Hi Albert,

 

Just wanted to mention that I now do notice an improvement with Fidelizer.  I don't understand why I didn't before. I might have made a mistake or something changed.

 

I sent an email to FTDI (USB chip we use) regarding it and the USB 3.0 port issue to see if they have any answers.  I figure it is worth the cost of an email :}

 

Still working on some optimization...

 

Regards

TK  

 

Group: DynoMotion	Message: 3721	From: albertplatek	Date: 2/5/2012
Subject: Re: Coord buffer overflow again

Hi Tom, My settings: Break angle 30 deg Look ahead 3 sec Collinear tolerance 0.001 in Corner tolerance 0.003 in Facet angle 1.5 deg Tau = 0.008 sec XYZ Accelarations 40 in/sec2 XYZ Velocity 4.5 in/sec In gcode file I use max Vel 270 in/min (4.5 in/sec) I have strange problem, everything was ok until I started machining bigger reliefs. At half of the job my stepper motors lost position probably becouse of high jerk of acceleration. Firstly I thought that maybe axis accelerations for my system are to high(40in/sec2). So I changed it to 20in/sec2 for test. But didnt help, still stepper motors stall at half of the gcode file. This file contain 548389 lines, my steppers stall between 226000 - 274000 line but not always in same place. I tried to find line where the problem occurs, but when I found line with problem and tried to execute 500 lines before, it was ok. Motors didnt stall. I found that motors stall when KMotionCNC start execute at least 100 000 lines before line number 274000. The problem disappear when I set break angle 5deg. "This diagnostic pointed out a side problem I hadn't noticed before. There is an annoying ~30 second delay when it runs the second time. The Interpreter keeps a record of all its state changes which allow Halts to go back to any previous state. In this test of 100x500 subroutine calls results in about 1 million state changes. Ironically creating the data, making use of the data all happens in the blink of an eye, but discarding the data (CList.RemoveAll) takes 30 seconds! We will find a better way in the next version. For now just don't be surprised if there is an annoying delay on a second run of a big job." Can this "delay" make that my stepper motors stall? Best Regards Albert P³atek --- In DynoMotion@yahoogroups.com, Albert <albertplatek86@...> wrote: > > Hi Tom, > > Thank you. > Btw. I am trying to find out what changes make > Fidelizer in Windows. If I will find something > important I will let you know. > Best regards > Albert Platek > > > Dnia 2012-01-18 o godz. 23:47 Tom Kerekes <tk@...> napisał > (a): > > > Hi Albert, > > > > Just wanted to mention that I now do notice an improvement with > > Fidelizer. I don't understand why I didn't before. I might have > > made a mistake or something changed. > > > > I sent an email to FTDI (USB chip we use) regarding it and the USB > > 3.0 port issue to see if they have any answers. I figure it is > > worth the cost of an email :} > > > > Still working on some optimization... > > > > Regards > > TK > > > > > > From: albertplatek <albertplatek86@...> > > To: DynoMotion@yahoogroups.com > > Sent: Monday, January 16, 2012 12:01 PM > > Subject: [DynoMotion] Re: Coord buffer overflow again > > > > > > Hi Tom, > > Thank you for making changes in new version to increase the segment > > rate. > > > > Btw: > > I installed on second pc Windows 7 (Core2Duo cpu), and aslo tested > > with fidelizer. Strange but didn't make difference, as in your > > computer. > > > > I cant get why in this particular case (pc with WIN7 i7 cpu) it helps. > > Maybe it depends on type of host controller... > > > > Pc with WIN7 64bit (Core2Duo) has > > SiS 7001 PCI to USB Open Host Controller > > > > Pc with WIN7 64bit (i7) has > > Intel(R) 6 Series/C200 Series Chipset Family USB Enhanced Host > > Controller > > > > This computer has aslo another usb ports: Renesas Electronics USB > > 3.0 Host Controller > > which works very bad with KFlop. And on this port fidelizer make no > > difference. > > > > Best Regards > > Albert Platek > > > > --- In DynoMotion@yahoogroups.com, Tom Kerekes <tk@> wrote: > > > > > > Hi Albert, > > >  > > > Thanks for that interesting information. > > >  > > > I tried the Fidelizer on my i7 W7 32 machine and it didn't seem to > > make any difference. > > >  > > > I'm trying some things to increase the segment rate by sending > > blocks of segments and optimizing how they are handled in KFLOP. I > > f this bears fruit I'll let you know. > > >  > > > This diagnostic pointed out a side problem I hadn't noticed before > > . There is an annoying ~30 second delay when it runs the second ti > > me. The Interpreter keeps a record of all its state changes which > > allow Halts to go back to any previous state. In this test of 100x > > 500 subroutine calls results in about 1 million state changes. Iro > > nically creating the data, making use of the data all happens in the > > blink of an eye, but discarding the data (CList.RemoveAll) takes 30 > > seconds! We will find a better way in the next version. For now > > just don't be surprised if there is an annoying delay on a second > > run of a big job. > > >  > > > Regards > > > TK > > >  > > > > > > > > > ________________________________ > > > From: albertplatek <albertplatek86@> > > > To: DynoMotion@yahoogroups.com > > > Sent: Sunday, January 15, 2012 8:37 AM > > > Subject: [DynoMotion] Re: Coord buffer overflow again > > > > > > > > >  > > > Hi Tom, > > > First thank you Tom for the test file. > > > I made few tests and I got interesting results :) > > > > > > PC No1: > > > > > > Intel Pentium 4 2.6Ghz WinXP 32bit port USB2.0 > > > Max segments: 110 > > > Time: 31s > > > 100 circles x 110 segments/circle / 31 seconds =354 segments per > > second. > > > > > > PC No2: > > > > > > Intel Core2Duo 2.0 Ghz WinXP 32bit port USB2.0 > > > Max segments: 140 > > > Time: 31s > > > 100 circles x 140 segments/circle / 31 seconds = 451 segments per > > second. > > > > > > PC No3: > > > > > > i7 2670QM Win7 64bit port USB3.0 > > > Max segments:30 > > > Time:31s > > > 100 circles x 60 segments/circle / 31 seconds = 193 segments per > > second. > > > (thats why I had problem here, usb 3.0 fail) > > > > > > (Luckily I had 1 USB 2.0 port) > > > > > > i7 2670QM Win7 64bit port USB2.0 > > > Max segments:170 > > > Time:31s > > > 100 circles x 170 segments/circle / 31 seconds = 548 segments per > > second. > > > > > > i7 2670QM Win7 64bit port USB2.0 (with 1 core affinity for > > KmotionCNC and KmotionServer) > > > Max segments:200 > > > Time:31s > > > 100 circles x 200 segments/circle / 31 seconds =645 segments per > > second. > > > > > > I found on net some interesting software called Fidelizer, its > > software to optimize win7/vista for audio workstation, but it aslo > > tweak somehow usb throughput. > > > More detail here: /www.windowsxlive.net/fidelizer > > > > > > i7 2670QM Win7 64bit port USB2.0 (Fidelizer) > > > Max segments:410 > > > Time:31s > > > 100 circles x 410 segments/circle / 31 seconds = 1320 segments per > > second. (wow! tested few times with same result) > > > > > > Be aware of new USB 3.0 ports they may not work with good > > performance with KFlop, maybe becouse of bug in drivers. Windows 7 > > dont see them without additional drivers. > > > Fidelizer software only improved performance on pc with  Win7. > > > On WinXP seem to not work. I will soon try install on "PC No2" > > Windows 7 and then try fidelizer also. > > > > > > Best regards > > > Albert P³atek > > > --- In DynoMotion@yahoogroups.com, Tom Kerekes <tk@> wrote: > > > > > > > > Hi Albert, > > > >  > > > > Your application/system is quite demanding :} > > > >  > > > > That portion of GCode you sent is 100,000 lines of all G1 > > movements with typical tiny lengths of only 50~150um. The > > collinear tolerance of 0.001 (25um) will cause many blocks to > > be combined together so the real required vector rate will be more d > > ependent on the path curvature rather than the original GCode. > > > >  > > > > On my system (i7 W7 32-bit) it runs ok with a facet angle 0f 0.4 > > degrees but can't keep up at 0.3 degrees. > > > >  > > > > I did some calculations for a system like yours with 40 in/sec2 > > acceleration. Here is what I find: > > > >  > > > > That acceleration allows going around a 0.1 inch radius circle > > at 120ipm (which is about 3 circles per second) > > > >  > > > > With a facet angle setting of 1.5 degrees that requires 240 > > segments per circle which would require 720 segments per second > > > >  > > > > I created a diagnostic test GCode file that programmatically > > generates an N sided Polygon 100 times with radius 0.1 inches > > and allows you to specify the number of polygon sides (it also gradu > > ally shifts the "circle" to make the plot more interesting). > > (see attachment).  When using the diagnostic make sure to set > > the Collinear tolerance to zero and the Corner Rounding to zero so t > > he original segments are used. By changing N (number of sides > > we can determine the segment download rate). > > > >  > > > > For example on my system it works with N = 200 and fails with N > > = 240. The total time to do all 100 circles takes about 30 > > seconds. So: > > > >  > > > > 100 circles x 200 segments/circle / 30 seconds = 666 segments > > per second. > > > >  > > > > In simulation mode the Interpreter+Graphics finishes the job in > > ~ 5 seconds so it isn't a significant factor > > > >  > > > > With the GCode Window closed it simulates in ~ 4 seconds so the > > Graphical Display is not a significant factor. > > > >  > > > > You might run the tests on your two systems to see if you get > > drastically different results. > > > >  > > > > But to try to answer your original question of why one CPU would > > be significantly different I don't know. Google finds some > > benchmarking tools but they are mainly for High Speed USB > > devices like flash drives. Some sites state if an older USB > > 1.1 device is attached it slows everything down. > > > >  > > > > Regards > > > > TK > > > >  > > > > > > > > From: albertplatek albertplatek86@ > > > > To: DynoMotion@yahoogroups.com > > > > Sent: Saturday, January 14, 2012 11:02 AM > > > > Subject: [DynoMotion] Re: Coord buffer overflow again > > > > > > > > > > > >  > > > > Hi Tom, > > > > How to determine USB throughput on pc? or what pc/system is best > > for KFlop? > > > > > > > > I have still sometimes cord buffer overflow on second pc even I > > > > set lower accelaration 25 and bigger facet angle 2.5 it sometimes > > > > overflow. Its very strange pc has windows xp clean installation, > > > > Integrated Intel Graphics. > > > > On first pc works perfect even with high acceleration 40, facet > > angle 2.0 (also windows xp, but ati graphics) > > > > > > > > Break angle 30 > > > > Look ahead 3 > > > > Collinear tolerance 0.001 > > > > Corner tolerance 0.005 > > > > Vel 5 > > > > > > > > Best Regards > > > > Albert Platek > > > > > > > > --- In DynoMotion@yahoogroups.com, Albert P³atek > > albertplatek86@ wrote: > > > > > > > > > > Hi Tom, > > > > > GCode file is attached with this message. > > > > > Yes feedrate is 300ipm. > > > > > What do you mean by "path generation slows down"? - i mean > > that when I am > > > > > changing view perspective in GViewer > > > > > the "simulated tool" catching up end of buffered path. So > > thats why I > > > > > thought that maybe gviewer impact on performance somehow. > > > > > > > > > > Best Regards > > > > > Albert PÃÆ'â€¦Ã¢â‚¬à ¡atek > > > > > > > > > > > > > > > 2012/1/13 Tom Kerekes tk@ > > > > > > > > > > > > > > > > > Hi Albert, > > > > > > > > > > > > Yes please send the file. Is the feedrate set high 300ipm? > > > > > > > > > > > > What do you mean by "path generation slows down"? > > > > > > > > > > > > Thanks > > > > > > TK > > > > > > > > > > > > > > > > > > > > > > > > *From:* albertplatek albertplatek86@ > > > > > > *To:* DynoMotion@yahoogroups.com > > > > > > *Sent:* Thursday, January 12, 2012 3:35 PM > > > > > > *Subject:* [DynoMotion] Re: Coord buffer overflow again > > > > > > > > > > > > > > > > > > Hi Tom, > > > > > > > > > > > > My settings: > > > > > > > > > > > > Break angle 30 > > > > > > Look ahead 3 > > > > > > Collinear tolerance 0.001 > > > > > > Corner tolerance 0.005 > > > > > > Facet angle 2.0 (if I go to 1.5, i get cord buffer overflow > > almost from > > > > > > program start) > > > > > > > > > > > > I use Tau = 0.01; (with facet angle 2.0 machine movements > > are fairly > > > > > > smooth) > > > > > > > > > > > > All 3 axis has same Vel/Accel parameters: > > > > > > Vel 5.0 > > > > > > Accel 40 > > > > > > > > > > > > Toolpath is parallel along X-axis (contain a lot of small > > arcs, I will > > > > > > send file tomorrow) > > > > > > > > > > > > With this settings my first router with old laptop works ok > > (Dell, core 2 > > > > > > duo, ati graphics card), > > > > > > but on second router (same settings) with new laptop dont > > work, i often > > > > > > get buffer overflows. (laptop Dell, i7 cpu, Win7 64bit, > > Nvidia Gt555) > > > > > > > > > > > > (Btw. when GViewer is open and I move view by mouse path > > generation slows > > > > > > down, is it normal?) > > > > > > > > > > > > Best Regards > > > > > > Albert PÃÆ'‚³atek > > > > > > > > > > > > --- In DynoMotion@yahoogroups.com, Tom Kerekes <tk@> wrote: > > > > > > > > > > > > > > Hi Albert, > > > > > > > ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > Could you send me the Gcode and all your settings?ÃÆ'Æ'à > > ¢â‚¬à ¡ I don't understand > > > > > > the collinear tolerance of 1?ÃÆ'Æ'Ã¢â‚¬à ¡ That > > seems huge.ÃÆ'Æ'Ã¢â‚¬à ¡ Did you mean 0.001? > > > > > > > ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > A corner rounding facet angle too small might be > > generating too many > > > > > > small segments.ÃÆ'Æ'Ã¢â‚¬à ¡ Are you using the > > Coordinated Motion Low Pass Filter > > > > > > Smoothing option?ÃÆ'Æ'Ã¢â‚¬à ¡ That might help make > > as smooth or smoother motion with > > > > > > larger facets. > > > > > > > ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > The USB throughput can vary from one machine to the other. > > > > > > > ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > No data for the GViewer comes from KFLOP it all comes from > > the > > > > > > trajectory planner on the PC. > > > > > > > ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > Regards > > > > > > > TK > > > > > > > ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > > > > > > > > From: albertplatek <albertplatek86@> > > > > > > > To: DynoMotion@yahoogroups.com > > > > > > > Sent: Thursday, January 12, 2012 11:11 AM > > > > > > > Subject: [DynoMotion] Coord buffer overflow again > > > > > > > > > > > > > > > > > > > > > ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > Hi Tom, > > > > > > > > > > > > > > 1. Why on 1 pc kmotioncnc works ok, but on second not, > > with same > > > > > > settings? > > > > > > > (collinear tollerance 1) > > > > > > > same file, same accelerations etc. > > > > > > > on second I have only usb 3.0 ports. > > > > > > > > > > > > > > 2. How much data come to Gviewer from kflop? if a lot is > > it possible > > > > > > > to bypass this if I dont need to use G viewer? > > > > > > > > > > > > > > I make a lot of reliefs in wood, with kinda high > > acceleleration 40, if > > > > > > > will be any way to get more performance from kflop with > > smoothing > > > > > > feature will be very good. > > > > > > > > > > > > > > Best Regards > > > > > > > Albert PÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚³atek > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > >

Group: DynoMotion	Message: 3724	From: Tom Kerekes	Date: 2/5/2012
Subject: Re: Coord buffer overflow again

Hi Albert,   Thanks for the very complete problem description, I can't tell you how much I appreciate that, but I can't think of anything that would cause this. Anything with large files I might expect some processing issue but that would result in a buffer underflow and an error message not incorrect motion that would result in a stall.  That delay I described in your email only happens when "Execute" is pushed so I don't think it is related.   There is an experimental version KMotion429x.exe on the download site that has this delay eliminated and also faster USB throughput, but I don't think either of those are related to this problem.   Can you post the file somewhere?  I'm thinking to write a C program that would run in KFLOP and watch for too high of acceleration or velocity by monitoring how ch0->Dest changes.  If it detects something bad that could be used as a trigger to help us track it down.   Which axes loose position?   Regards TK   Group: DynoMotion Message: 3725 From: Albert Płatek Date: 2/5/2012 Subject: Re: Coord buffer overflow again Attachments : Relief14-3725.zip Hi Tom,   Thank you for help. Mostly stall X axis, and sometimes Z. Y axis seem to be ok. I uploaded file as an attachment (zip).   Best Regards Albert Platek 2012/2/5 Tom Kerekes <tk@...>   Hi Albert,   Thanks for the very complete problem description, I can't tell you how much I appreciate that, but I can't think of anything that would cause this. Anything with large files I might expect some processing issue but that would result in a buffer underflow and an error message not incorrect motion that would result in a stall.  That delay I described in your email only happens when "Execute" is pushed so I don't think it is related.   There is an experimental version KMotion429x.exe on the download site that has this delay eliminated and also faster USB throughput, but I don't think either of those are related to this problem.   Can you post the file somewhere?  I'm thinking to write a C program that would run in KFLOP and watch for too high of acceleration or velocity by monitoring how ch0->Dest changes.  If it detects something bad that could be used as a trigger to help us track it down.   Which axes loose position?   Regards TK   Group: DynoMotion Message: 3736 From: albertplatek Date: 2/6/2012 Subject: Re: Coord buffer overflow again Hi Tom, Problem solved (probably), weird but after re-flash firmware 4.28 everything is working ok, no axis stall. btw. Experimental version KMotion429x.exe is not working, when I start execute gcode I get message "G Code Error GCode Aborted", this error occurs on any gcode program even on only few G01 lines. Tom if you have C program that watch for too high of acceleration please post it. If again in future something strange will happen I will try find out the problem. I am still not sure what was the problem in my system that axis were stalling. Best Regards Albert P³atek --- In DynoMotion@yahoogroups.com, Albert P³atek <albertplatek86@...> wrote: > > Hi Tom, > > Thank you for help. > Mostly stall X axis, and sometimes Z. > Y axis seem to be ok. > I uploaded file as an attachment (zip). > > Best Regards > Albert Platek > > 2012/2/5 Tom Kerekes <tk@...> > > > ** > > > > > > Hi Albert, > > > > Thanks for the very complete problem description, I can't tell you how > > much I appreciate that, but I can't think of anything that would cause > > this. Anything with large files I might expect some processing issue but > > that would result in a buffer underflow and an error message not incorrect > > motion that would result in a stall. That delay I described in your email > > only happens when "Execute" is pushed so I don't think it is related. > > > > There is an experimental version KMotion429x.exe on the download site > > that has this delay eliminated and also faster USB throughput, but I don't > > think either of those are related to this problem. > > > > Can you post the file somewhere? I'm thinking to write a C program that > > would run in KFLOP and watch for too high of acceleration or velocity by > > monitoring how ch0->Dest changes. If it detects something bad that could > > be used as a trigger to help us track it down. > > > > Which axes loose position? > > > > Regards > > TK > > > > > > *From:* albertplatek <albertplatek86@...> > > *To:* DynoMotion@yahoogroups.com > > *Sent:* Sunday, February 5, 2012 5:27 AM > > > > *Subject:* [DynoMotion] Re: Coord buffer overflow again > > > > > > Hi Tom, > > > > My settings: > > Break angle 30 deg > > Look ahead 3 sec > > Collinear tolerance 0.001 in > > Corner tolerance 0.003 in > > Facet angle 1.5 deg > > Tau = 0.008 sec > > XYZ Accelarations 40 in/sec2 > > XYZ Velocity 4.5 in/sec > > In gcode file I use max Vel 270 in/min (4.5 in/sec) > > > > I have strange problem, everything was ok until I started machining > > bigger reliefs. > > At half of the job my stepper motors lost position > > probably becouse of high jerk of acceleration. > > Firstly I thought that maybe axis accelerations for my system are to > > high(40in/sec2). So I changed it to 20in/sec2 for test. > > But didnt help, still stepper motors stall at half of the gcode file. > > > > This file contain 548389 lines, my steppers stall between 226000 - 274000 > > line but not always in same place. > > > > I tried to find line where the problem occurs, but when I found line with > > problem and tried to execute 500 lines before, it was ok. Motors didnt > > stall. > > I found that motors stall when KMotionCNC start execute at least > > 100 000 lines before line number 274000. > > > > The problem disappear when I set break angle 5deg. > > > > "This diagnostic pointed out a side problem I hadn't noticed before. There > > is an annoying ~30 second delay when it runs the second time. The > > Interpreter keeps a record of all its state changes which allow Halts to go > > back to any previous state. In this test of 100x500 subroutine calls > > results in about 1 million state changes. Ironically creating the data, > > making use of the data all happens in the blink of an eye, but discarding > > the data (CList.RemoveAll) takes 30 seconds! We will find a better way in > > the next version. For now just don't be surprised if there is an annoying > > delay on a second run of a big job." > > > > Can this "delay" make that my stepper motors stall? > > > > Best Regards > > Albert P³atek > > > > --- In DynoMotion@yahoogroups.com, Albert <albertplatek86@> wrote: > > > > > > Hi Tom, > > > > > > Thank you. > > > Btw. I am trying to find out what changes make > > > Fidelizer in Windows. If I will find something > > > important I will let you know. > > > Best regards > > > Albert Platek > > > > > > > > > Dnia 2012-01-18 o godz. 23:47 Tom Kerekes <tk@> napisaà ‚ > > > (a): > > > > > > > Hi Albert, > > > > > > > > Just wanted to mention that I now do notice an improvement with > > > > Fidelizer. I don't understand why I didn't before. I might have > > > > made a mistake or something changed. > > > > > > > > I sent an email to FTDI (USB chip we use) regarding it and the USB > > > > 3.0 port issue to see if they have any answers. I figure it is > > > > worth the cost of an email :} > > > > > > > > Still working on some optimization... > > > > > > > > Regards > > > > TK > > > > > > > > > > > > From: albertplatek <albertplatek86@> > > > > To: DynoMotion@yahoogroups.com > > > > Sent: Monday, January 16, 2012 12:01 PM > > > > Subject: [DynoMotion] Re: Coord buffer overflow again > > > > > > > > > > > > Hi Tom, > > > > Thank you for making changes in new version to increase the segment > > > > rate. > > > > > > > > Btw: > > > > I installed on second pc Windows 7 (Core2Duo cpu), and aslo tested > > > > with fidelizer. Strange but didn't make difference, as in your > > > > computer. > > > > > > > > I cant get why in this particular case (pc with WIN7 i7 cpu) it helps. > > > > Maybe it depends on type of host controller... > > > > > > > > Pc with WIN7 64bit (Core2Duo) has > > > > SiS 7001 PCI to USB Open Host Controller > > > > > > > > Pc with WIN7 64bit (i7) has > > > > Intel(R) 6 Series/C200 Series Chipset Family USB Enhanced Host > > > > Controller > > > > > > > > This computer has aslo another usb ports: Renesas Electronics USB > > > > 3.0 Host Controller > > > > which works very bad with KFlop. And on this port fidelizer make no > > > > difference. > > > > > > > > Best Regards > > > > Albert Platek > > > > > > > > --- In DynoMotion@yahoogroups.com, Tom Kerekes <tk@> wrote: > > > > > > > > > > Hi Albert, > > > > >  > > > > > Thanks for that interesting information. > > > > >  > > > > > I tried the Fidelizer on my i7 W7 32 machine and it didn't seem to > > > > make any difference. > > > > >  > > > > > I'm trying some things to increase the segment rate by sending > > > > blocks of segments and optimizing how they are handled in KFLOP. I > > > > f this bears fruit I'll let you know. > > > > >  > > > > > This diagnostic pointed out a side problem I hadn't noticed before > > > > . There is an annoying ~30 second delay when it runs the second ti > > > > me. The Interpreter keeps a record of all its state changes which > > > > allow Halts to go back to any previous state. In this test of 100x > > > > 500 subroutine calls results in about 1 million state changes. Iro > > > > nically creating the data, making use of the data all happens in the > > > > blink of an eye, but discarding the data (CList.RemoveAll) takes 30 > > > > seconds! We will find a better way in the next version. For now > > > > just don't be surprised if there is an annoying delay on a second > > > > run of a big job. > > > > >  > > > > > Regards > > > > > TK > > > > >  > > > > > > > > > > > > > > > ________________________________ > > > > > From: albertplatek <albertplatek86@> > > > > > To: DynoMotion@yahoogroups.com > > > > > Sent: Sunday, January 15, 2012 8:37 AM > > > > > Subject: [DynoMotion] Re: Coord buffer overflow again > > > > > > > > > > > > > > >  > > > > > Hi Tom, > > > > > First thank you Tom for the test file. > > > > > I made few tests and I got interesting results :) > > > > > > > > > > PC No1: > > > > > > > > > > Intel Pentium 4 2.6Ghz WinXP 32bit port USB2.0 > > > > > Max segments: 110 > > > > > Time: 31s > > > > > 100 circles x 110 segments/circle / 31 seconds =354 segments per > > > > second. > > > > > > > > > > PC No2: > > > > > > > > > > Intel Core2Duo 2.0 Ghz WinXP 32bit port USB2.0 > > > > > Max segments: 140 > > > > > Time: 31s > > > > > 100 circles x 140 segments/circle / 31 seconds = 451 segments per > > > > second. > > > > > > > > > > PC No3: > > > > > > > > > > i7 2670QM Win7 64bit port USB3.0 > > > > > Max segments:30 > > > > > Time:31s > > > > > 100 circles x 60 segments/circle / 31 seconds = 193 segments per > > > > second. > > > > > (thats why I had problem here, usb 3.0 fail) > > > > > > > > > > (Luckily I had 1 USB 2.0 port) > > > > > > > > > > i7 2670QM Win7 64bit port USB2.0 > > > > > Max segments:170 > > > > > Time:31s > > > > > 100 circles x 170 segments/circle / 31 seconds = 548 segments per > > > > second. > > > > > > > > > > i7 2670QM Win7 64bit port USB2.0 (with 1 core affinity for > > > > KmotionCNC and KmotionServer) > > > > > Max segments:200 > > > > > Time:31s > > > > > 100 circles x 200 segments/circle / 31 seconds =645 segments per > > > > second. > > > > > > > > > > I found on net some interesting software called Fidelizer, its > > > > software to optimize win7/vista for audio workstation, but it aslo > > > > tweak somehow usb throughput. > > > > > More detail here: /www.windowsxlive.net/fidelizer > > > > > > > > > > i7 2670QM Win7 64bit port USB2.0 (Fidelizer) > > > > > Max segments:410 > > > > > Time:31s > > > > > 100 circles x 410 segments/circle / 31 seconds = 1320 segments per > > > > second. (wow! tested few times with same result) > > > > > > > > > > Be aware of new USB 3.0 ports they may not work with good > > > > performance with KFlop, maybe becouse of bug in drivers. Windows 7 > > > > dont see them without additional drivers. > > > > > Fidelizer software only improved performance on pc with  Win7. > > > > > On WinXP seem to not work. I will soon try install on "PC No2" > > > > Windows 7 and then try fidelizer also. > > > > > > > > > > Best regards > > > > > Albert P³atek > > > > > --- In DynoMotion@yahoogroups.com, Tom Kerekes <tk@> wrote: > > > > > > > > > > > > Hi Albert, > > > > > > ÃÆ'‚ > > > > > > Your application/system is quite demanding :} > > > > > > ÃÆ'‚ > > > > > > That portion of GCode you sent is 100,000 lines of all G1 > > > > movements with typical tiny lengths of only 50~150um.ÃÆ'‚ The > > > > collinear tolerance of 0.001 (25um) will cause many blocksÃÆ'‚ to > > > > be combined together so the real required vector rate will be more d > > > > ependent on the path curvature rather than the original GCode.ÃÆ'‚ > > > > > > ÃÆ'‚ > > > > > > On my system (i7 W7 32-bit) it runs ok with a facet angle 0f 0.4 > > > > degrees butÃÆ'‚ can't keep up at 0.3 degrees. > > > > > > ÃÆ'‚ > > > > > > I did some calculations for a system like yours with 40 in/sec2 > > > > acceleration.ÃÆ'‚ Here is what I find: > > > > > > ÃÆ'‚ > > > > > > That acceleration allows going around a 0.1 inch radius circle > > > > at 120ipm (which is about 3 circles per second) > > > > > > ÃÆ'‚ > > > > > > With a facet angle setting of 1.5 degrees that requires 240 > > > > segments per circle which would require 720 segments per second > > > > > > ÃÆ'‚ > > > > > > I created a diagnostic test GCode file that programmatically > > > > generates an N sided Polygon 100 timesÃÆ'‚ with radius 0.1 inches > > > > and allows you to specify the number of polygon sides (it also gradu > > > > ally shifts the "circle" to make the plot more interesting).ÃÆ'‚ > > > > (see attachment). ÃÆ'‚ When using the diagnostic make sure to set > > > > the Collinear tolerance to zero and the Corner Rounding to zero so t > > > > he original segments are used.ÃÆ'‚ By changing N (number of sides > > > > we can determine the segment download rate). > > > > > > ÃÆ'‚ > > > > > > For example on my system it works with N = 200 and fails with N > > > > = 240.ÃÆ'‚ The total time to do all 100 circles takes about 30 > > > > seconds.ÃÆ'‚ So: > > > > > > ÃÆ'‚ > > > > > > 100 circles x 200 segments/circle / 30 seconds = 666 segments > > > > per second. > > > > > > ÃÆ'‚ > > > > > > In simulation mode the Interpreter+Graphics finishes the job in > > > > ~ 5 seconds so it isn't a significant factor > > > > > > ÃÆ'‚ > > > > > > With the GCode Window closed it simulates in ~ 4 seconds so the > > > > Graphical Display is not aÃÆ'‚ significant factor. > > > > > > ÃÆ'‚ > > > > > > You might run the tests on your two systems to see if you get > > > > drastically different results. > > > > > > ÃÆ'‚ > > > > > > But to try to answer your original question of why one CPU would > > > > be significantly different I don't know.ÃÆ'‚ Google finds some > > > > benchmarking tools but they areÃÆ'‚ mainly for High Speed USB > > > > devices like flash drives.ÃÆ'‚ Some sites state if an older USB > > > > 1.1 device is attached it slows everything down.ÃÆ'‚ > > > > > > ÃÆ'‚ > > > > > > Regards > > > > > > TKÃÆ'‚ > > > > > > ÃÆ'‚ > > > > > > > > > > > > From: albertplatek albertplatek86@ > > > > > > To: DynoMotion@yahoogroups.com > > > > > > Sent: Saturday, January 14, 2012 11:02 AM > > > > > > Subject: [DynoMotion] Re: Coord buffer overflow again > > > > > > > > > > > > > > > > > > ÃÆ'‚ > > > > > > Hi Tom, > > > > > > How to determine USB throughput on pc? or what pc/system is best > > > > for KFlop? > > > > > > > > > > > > I have still sometimes cord buffer overflow on second pc even I > > > > > > set lower accelaration 25 and bigger facet angle 2.5 it sometimes > > > > > > overflow. Its very strange pc has windows xp clean installation, > > > > > > Integrated Intel Graphics. > > > > > > On first pc works perfect even with high acceleration 40, facet > > > > angle 2.0 (also windows xp, but ati graphics) > > > > > > > > > > > > Break angle 30 > > > > > > Look ahead 3 > > > > > > Collinear tolerance 0.001 > > > > > > Corner tolerance 0.005 > > > > > > Vel 5 > > > > > > > > > > > > Best Regards > > > > > > Albert Platek > > > > > > > > > > > > --- In DynoMotion@yahoogroups.com, Albert PÃÆ'‚³atek > > > > albertplatek86@ wrote: > > > > > > > > > > > > > > Hi Tom, > > > > > > > GCode file is attached with this message. > > > > > > > Yes feedrate is 300ipm. > > > > > > > What do you mean by "path generation slows down"? - i mean > > > > that when I am > > > > > > > changing view perspective in GViewer > > > > > > > the "simulated tool" catching up end of buffered path. So > > > > thats why I > > > > > > > thought that maybe gviewer impact on performance somehow. > > > > > > > > > > > > > > Best Regards > > > > > > > Albert PÃÆ'Æ'…ÃÆ'¢â‚¬Å¡atek > > > > > > > > > > > > > > > > > > > > > 2012/1/13 Tom Kerekes tk@ > > > > > > > > > > > > > > > > > > > > > > > Hi Albert, > > > > > > > > > > > > > > > > Yes please send the file. Is the feedrate set high 300ipm? > > > > > > > > > > > > > > > > What do you mean by "path generation slows down"? > > > > > > > > > > > > > > > > Thanks > > > > > > > > TK > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > *From:* albertplatek albertplatek86@ > > > > > > > > *To:* DynoMotion@yahoogroups.com > > > > > > > > *Sent:* Thursday, January 12, 2012 3:35 PM > > > > > > > > *Subject:* [DynoMotion] Re: Coord buffer overflow again > > > > > > > > > > > > > > > > > > > > > > > > Hi Tom, > > > > > > > > > > > > > > > > My settings: > > > > > > > > > > > > > > > > Break angle 30 > > > > > > > > Look ahead 3 > > > > > > > > Collinear tolerance 0.001 > > > > > > > > Corner tolerance 0.005 > > > > > > > > Facet angle 2.0 (if I go to 1.5, i get cord buffer overflow > > > > almost from > > > > > > > > program start) > > > > > > > > > > > > > > > > I use Tau = 0.01; (with facet angle 2.0 machine movements > > > > are fairly > > > > > > > > smooth) > > > > > > > > > > > > > > > > All 3 axis has same Vel/Accel parameters: > > > > > > > > Vel 5.0 > > > > > > > > Accel 40 > > > > > > > > > > > > > > > > Toolpath is parallel along X-axis (contain a lot of small > > > > arcs, I will > > > > > > > > send file tomorrow) > > > > > > > > > > > > > > > > With this settings my first router with old laptop works ok > > > > (Dell, core 2 > > > > > > > > duo, ati graphics card), > > > > > > > > but on second router (same settings) with new laptop dont > > > > work, i often > > > > > > > > get buffer overflows. (laptop Dell, i7 cpu, Win7 64bit, > > > > Nvidia Gt555) > > > > > > > > > > > > > > > > (Btw. when GViewer is open and I move view by mouse path > > > > generation slows > > > > > > > > down, is it normal?) > > > > > > > > > > > > > > > > Best Regards > > > > > > > > Albert PÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚³atek > > > > > > > > > > > > > > > > --- In DynoMotion@yahoogroups.com, Tom Kerekes <tk@> wrote: > > > > > > > > > > > > > > > > > > Hi Albert, > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ > > > > > > > > > Could you send me the Gcode and all your settings?ÃÆ'Æ'Æ'ÃÆ' > > > > ¢â‚¬Å¡ I don't understand > > > > > > > > the collinear tolerance of 1?ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ That > > > > seems huge.ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ Did you mean 0.001? > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ > > > > > > > > > A corner rounding facet angle too small might be > > > > generating too many > > > > > > > > small segments.ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ Are you using the > > > > Coordinated Motion Low Pass Filter > > > > > > > > Smoothing option?ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ That might help make > > > > as smooth or smoother motion with > > > > > > > > larger facets. > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ > > > > > > > > > The USB throughput can vary from one machine to the other. > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ > > > > > > > > > No data for the GViewer comes from KFLOP it all comes from > > > > the > > > > > > > > trajectory planner on the PC. > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ > > > > > > > > > Regards > > > > > > > > > TK > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ > > > > > > > > > > > > > > > > > > From: albertplatek <albertplatek86@> > > > > > > > > > To: DynoMotion@yahoogroups.com > > > > > > > > > Sent: Thursday, January 12, 2012 11:11 AM > > > > > > > > > Subject: [DynoMotion] Coord buffer overflow again > > > > > > > > > > > > > > > > > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ > > > > > > > > > Hi Tom, > > > > > > > > > > > > > > > > > > 1. Why on 1 pc kmotioncnc works ok, but on second not, > > > > with same > > > > > > > > settings? > > > > > > > > > (collinear tollerance 1) > > > > > > > > > same file, same accelerations etc. > > > > > > > > > on second I have only usb 3.0 ports. > > > > > > > > > > > > > > > > > > 2. How much data come to Gviewer from kflop? if a lot is > > > > it possible > > > > > > > > > to bypass this if I dont need to use G viewer? > > > > > > > > > > > > > > > > > > I make a lot of reliefs in wood, with kinda high > > > > acceleleration 40, if > > > > > > > > > will be any way to get more performance from kflop with > > > > smoothing > > > > > > > > feature will be very good. > > > > > > > > > > > > > > > > > > Best Regards > > > > > > > > > Albert PÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚³atek > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > Group: DynoMotion Message: 3738 From: albertplatek Date: 2/6/2012 Subject: Re: Coord buffer overflow again Hi Tom, Sorry Tom but still this happens randomly :/ I was just lucky that after firmware flash 1 job was ok. I use ver 4.28. I dont know how to determine line numner of this jerk, but you can see huge acceleration! X motor stalled. It was runinng about 10min from program start. Acceleration was set 32in/sec2 Break Angle 30 Tau 0.008 Collinear Tolerance 0.0005 Corener Tolerance 0.01 Facet angle 0.5 Vel 4in/sec Here is my simple C program to check acceleration please check if it is correct: #include "KMotionDef.h" #define channel ch0 #define time 0.00018 main() { double X1,X2,speed1,speed2,accel; int k; while(1) { for (k=0; k<1; k++) WaitNextTimeSlice(); X1 = channel->Dest; for (k=0; k<1; k++) WaitNextTimeSlice(); X2 = channel->Dest; speed1=(X2-X1)/(time); for (k=0; k<1; k++) WaitNextTimeSlice(); X1 = channel->Dest; for (k=0; k<1; k++) WaitNextTimeSlice(); X2 = channel->Dest; speed2=(X2-X1)/(time); accel=(speed1-speed2)/(time); accel=(accel/254000)/2; if(accel>45) { printf("AccelError= %f\n",accel); } } } Here log: AccelError= 45.009219 AccelError= 45.229875 AccelError= 46.294106 AccelError= 45.055529 AccelError= 45.747547 AccelError= 45.070729 AccelError= 45.360541 AccelError= 45.765590 AccelError= 46.302076 AccelError= 46.775562 AccelError= 47.051896 AccelError= 45.225101 AccelError= 45.730674 AccelError= 46.758273 AccelError= 46.838768 AccelError= 45.698572 AccelError= 46.918682 AccelError= 45.204219 AccelError= 45.900150 AccelError= 47.023519 AccelError= 47.551550 AccelError= 48.022541 AccelError= 46.395295 AccelError= 81526.314840 AccelError= 72853.056587 AccelError= 65101.872298 AccelError= 58175.992747 AccelError= 51986.611437 AccelError= 46456.046360 AccelError= 41513.718492 AccelError= 37097.300673 AccelError= 33150.797142 AccelError= 29622.770379 AccelError= 26468.350534 AccelError= 23652.797849 AccelError= 21136.323622 AccelError= 18888.213633 AccelError= 16879.065077 AccelError= 15083.474245 AccelError= 13479.186003 AccelError= 12045.566722 AccelError= 10764.204608 AccelError= 9619.483512 AccelError= 8596.141212 AccelError= 7682.015002 AccelError= 6866.662006 AccelError= 6139.961139 AccelError= 5490.446488 AccelError= 4910.042286 AccelError= 4391.574405 AccelError= 3928.114510 AccelError= 3514.000485 AccelError= 3144.002655 AccelError= 2813.316136 AccelError= 2517.500643 AccelError= 2253.490792 AccelError= 2017.432428 AccelError= 1806.266152 AccelError= 1617.706032 AccelError= 1448.982225 AccelError= 1298.398178 AccelError= 1163.621865 AccelError= 1043.472200 AccelError= 935.821407 AccelError= 839.766485 AccelError= 753.863074 AccelError= 677.101783 AccelError= 608.587862 AccelError= 547.209083 AccelError= 492.589836 AccelError= 443.554631 AccelError= 399.735924 AccelError= 360.639632 AccelError= 325.643422 AccelError= 294.370902 AccelError= 266.425861 AccelError= 241.478288 AccelError= 219.191672 AccelError= 199.246085 AccelError= 181.422761 AccelError= 165.532090 AccelError= 151.295897 AccelError= 138.574472 AccelError= 127.206640 AccelError= 117.084135 AccelError= 108.003075 AccelError= 99.888264 AccelError= 92.598183 AccelError= 86.122270 AccelError= 76.770777 AccelError= 65.225363 AccelError= 54.908134 AccelError= 45.688617 AccelError= 45.070435 AccelError= 45.411744 AccelError= 45.770036 AccelError= 45.871485 AccelError= 45.239885 AccelError= 45.631836 AccelError= 45.001237 AccelError= 45.373033 AccelError= 45.577907 AccelError= 46.322608 AccelError= 46.958820 AccelError= 46.038556 AccelError= 45.018088 If you test it on your PC and will be ok, then maybe something is wrong with my laptop. Then I will test on other PC. Best Regards Albert P³atek --- In DynoMotion@yahoogroups.com, "albertplatek" <albertplatek86@...> wrote: > > Hi Tom, > > Problem solved (probably), weird but after re-flash firmware 4.28 > everything is working ok, no axis stall. > > btw. Experimental version KMotion429x.exe is not working, when I start > execute gcode I get message "G Code Error GCode Aborted", this error > occurs on any gcode program even on only few G01 lines. > > Tom if you have C program that watch for too high of acceleration > please post it. If again in future something strange will > happen I will try find out the problem. > I am still not sure what was the problem in my system that axis > were stalling. > > Best Regards > Albert P³atek > > > > --- In DynoMotion@yahoogroups.com, Albert P³atek <albertplatek86@> wrote: > > > > Hi Tom, > > > > Thank you for help. > > Mostly stall X axis, and sometimes Z. > > Y axis seem to be ok. > > I uploaded file as an attachment (zip). > > > > Best Regards > > Albert Platek > > > > 2012/2/5 Tom Kerekes <tk@> > > > > > ** > > > > > > > > > Hi Albert, > > > > > > Thanks for the very complete problem description, I can't tell you how > > > much I appreciate that, but I can't think of anything that would cause > > > this. Anything with large files I might expect some processing issue but > > > that would result in a buffer underflow and an error message not incorrect > > > motion that would result in a stall. That delay I described in your email > > > only happens when "Execute" is pushed so I don't think it is related. > > > > > > There is an experimental version KMotion429x.exe on the download site > > > that has this delay eliminated and also faster USB throughput, but I don't > > > think either of those are related to this problem. > > > > > > Can you post the file somewhere? I'm thinking to write a C program that > > > would run in KFLOP and watch for too high of acceleration or velocity by > > > monitoring how ch0->Dest changes. If it detects something bad that could > > > be used as a trigger to help us track it down. > > > > > > Which axes loose position? > > > > > > Regards > > > TK > > > > > > > > > *From:* albertplatek <albertplatek86@> > > > *To:* DynoMotion@yahoogroups.com > > > *Sent:* Sunday, February 5, 2012 5:27 AM > > > > > > *Subject:* [DynoMotion] Re: Coord buffer overflow again > > > > > > > > > Hi Tom, > > > > > > My settings: > > > Break angle 30 deg > > > Look ahead 3 sec > > > Collinear tolerance 0.001 in > > > Corner tolerance 0.003 in > > > Facet angle 1.5 deg > > > Tau = 0.008 sec > > > XYZ Accelarations 40 in/sec2 > > > XYZ Velocity 4.5 in/sec > > > In gcode file I use max Vel 270 in/min (4.5 in/sec) > > > > > > I have strange problem, everything was ok until I started machining > > > bigger reliefs. > > > At half of the job my stepper motors lost position > > > probably becouse of high jerk of acceleration. > > > Firstly I thought that maybe axis accelerations for my system are to > > > high(40in/sec2). So I changed it to 20in/sec2 for test. > > > But didnt help, still stepper motors stall at half of the gcode file. > > > > > > This file contain 548389 lines, my steppers stall between 226000 - 274000 > > > line but not always in same place. > > > > > > I tried to find line where the problem occurs, but when I found line with > > > problem and tried to execute 500 lines before, it was ok. Motors didnt > > > stall. > > > I found that motors stall when KMotionCNC start execute at least > > > 100 000 lines before line number 274000. > > > > > > The problem disappear when I set break angle 5deg. > > > > > > "This diagnostic pointed out a side problem I hadn't noticed before. There > > > is an annoying ~30 second delay when it runs the second time. The > > > Interpreter keeps a record of all its state changes which allow Halts to go > > > back to any previous state. In this test of 100x500 subroutine calls > > > results in about 1 million state changes. Ironically creating the data, > > > making use of the data all happens in the blink of an eye, but discarding > > > the data (CList.RemoveAll) takes 30 seconds! We will find a better way in > > > the next version. For now just don't be surprised if there is an annoying > > > delay on a second run of a big job." > > > > > > Can this "delay" make that my stepper motors stall? > > > > > > Best Regards > > > Albert P³atek > > > > > > --- In DynoMotion@yahoogroups.com, Albert <albertplatek86@> wrote: > > > > > > > > Hi Tom, > > > > > > > > Thank you. > > > > Btw. I am trying to find out what changes make > > > > Fidelizer in Windows. If I will find something > > > > important I will let you know. > > > > Best regards > > > > Albert Platek > > > > > > > > > > > > Dnia 2012-01-18 o godz. 23:47 Tom Kerekes <tk@> napisaà ‚ > > > > (a): > > > > > > > > > Hi Albert, > > > > > > > > > > Just wanted to mention that I now do notice an improvement with > > > > > Fidelizer. I don't understand why I didn't before. I might have > > > > > made a mistake or something changed. > > > > > > > > > > I sent an email to FTDI (USB chip we use) regarding it and the USB > > > > > 3.0 port issue to see if they have any answers. I figure it is > > > > > worth the cost of an email :} > > > > > > > > > > Still working on some optimization... > > > > > > > > > > Regards > > > > > TK > > > > > > > > > > > > > > > From: albertplatek <albertplatek86@> > > > > > To: DynoMotion@yahoogroups.com > > > > > Sent: Monday, January 16, 2012 12:01 PM > > > > > Subject: [DynoMotion] Re: Coord buffer overflow again > > > > > > > > > > > > > > > Hi Tom, > > > > > Thank you for making changes in new version to increase the segment > > > > > rate. > > > > > > > > > > Btw: > > > > > I installed on second pc Windows 7 (Core2Duo cpu), and aslo tested > > > > > with fidelizer. Strange but didn't make difference, as in your > > > > > computer. > > > > > > > > > > I cant get why in this particular case (pc with WIN7 i7 cpu) it helps. > > > > > Maybe it depends on type of host controller... > > > > > > > > > > Pc with WIN7 64bit (Core2Duo) has > > > > > SiS 7001 PCI to USB Open Host Controller > > > > > > > > > > Pc with WIN7 64bit (i7) has > > > > > Intel(R) 6 Series/C200 Series Chipset Family USB Enhanced Host > > > > > Controller > > > > > > > > > > This computer has aslo another usb ports: Renesas Electronics USB > > > > > 3.0 Host Controller > > > > > which works very bad with KFlop. And on this port fidelizer make no > > > > > difference. > > > > > > > > > > Best Regards > > > > > Albert Platek > > > > > > > > > > --- In DynoMotion@yahoogroups.com, Tom Kerekes <tk@> wrote: > > > > > > > > > > > > Hi Albert, > > > > > >  > > > > > > Thanks for that interesting information. > > > > > >  > > > > > > I tried the Fidelizer on my i7 W7 32 machine and it didn't seem to > > > > > make any difference. > > > > > >  > > > > > > I'm trying some things to increase the segment rate by sending > > > > > blocks of segments and optimizing how they are handled in KFLOP. I > > > > > f this bears fruit I'll let you know. > > > > > >  > > > > > > This diagnostic pointed out a side problem I hadn't noticed before > > > > > . There is an annoying ~30 second delay when it runs the second ti > > > > > me. The Interpreter keeps a record of all its state changes which > > > > > allow Halts to go back to any previous state. In this test of 100x > > > > > 500 subroutine calls results in about 1 million state changes. Iro > > > > > nically creating the data, making use of the data all happens in the > > > > > blink of an eye, but discarding the data (CList.RemoveAll) takes 30 > > > > > seconds! We will find a better way in the next version. For now > > > > > just don't be surprised if there is an annoying delay on a second > > > > > run of a big job. > > > > > >  > > > > > > Regards > > > > > > TK > > > > > >  > > > > > > > > > > > > > > > > > > ________________________________ > > > > > > From: albertplatek <albertplatek86@> > > > > > > To: DynoMotion@yahoogroups.com > > > > > > Sent: Sunday, January 15, 2012 8:37 AM > > > > > > Subject: [DynoMotion] Re: Coord buffer overflow again > > > > > > > > > > > > > > > > > >  > > > > > > Hi Tom, > > > > > > First thank you Tom for the test file. > > > > > > I made few tests and I got interesting results :) > > > > > > > > > > > > PC No1: > > > > > > > > > > > > Intel Pentium 4 2.6Ghz WinXP 32bit port USB2.0 > > > > > > Max segments: 110 > > > > > > Time: 31s > > > > > > 100 circles x 110 segments/circle / 31 seconds =354 segments per > > > > > second. > > > > > > > > > > > > PC No2: > > > > > > > > > > > > Intel Core2Duo 2.0 Ghz WinXP 32bit port USB2.0 > > > > > > Max segments: 140 > > > > > > Time: 31s > > > > > > 100 circles x 140 segments/circle / 31 seconds = 451 segments per > > > > > second. > > > > > > > > > > > > PC No3: > > > > > > > > > > > > i7 2670QM Win7 64bit port USB3.0 > > > > > > Max segments:30 > > > > > > Time:31s > > > > > > 100 circles x 60 segments/circle / 31 seconds = 193 segments per > > > > > second. > > > > > > (thats why I had problem here, usb 3.0 fail) > > > > > > > > > > > > (Luckily I had 1 USB 2.0 port) > > > > > > > > > > > > i7 2670QM Win7 64bit port USB2.0 > > > > > > Max segments:170 > > > > > > Time:31s > > > > > > 100 circles x 170 segments/circle / 31 seconds = 548 segments per > > > > > second. > > > > > > > > > > > > i7 2670QM Win7 64bit port USB2.0 (with 1 core affinity for > > > > > KmotionCNC and KmotionServer) > > > > > > Max segments:200 > > > > > > Time:31s > > > > > > 100 circles x 200 segments/circle / 31 seconds =645 segments per > > > > > second. > > > > > > > > > > > > I found on net some interesting software called Fidelizer, its > > > > > software to optimize win7/vista for audio workstation, but it aslo > > > > > tweak somehow usb throughput. > > > > > > More detail here: /www.windowsxlive.net/fidelizer > > > > > > > > > > > > i7 2670QM Win7 64bit port USB2.0 (Fidelizer) > > > > > > Max segments:410 > > > > > > Time:31s > > > > > > 100 circles x 410 segments/circle / 31 seconds = 1320 segments per > > > > > second. (wow! tested few times with same result) > > > > > > > > > > > > Be aware of new USB 3.0 ports they may not work with good > > > > > performance with KFlop, maybe becouse of bug in drivers. Windows 7 > > > > > dont see them without additional drivers. > > > > > > Fidelizer software only improved performance on pc with  Win7. > > > > > > On WinXP seem to not work. I will soon try install on "PC No2" > > > > > Windows 7 and then try fidelizer also. > > > > > > > > > > > > Best regards > > > > > > Albert P³atek > > > > > > --- In DynoMotion@yahoogroups.com, Tom Kerekes <tk@> wrote: > > > > > > > > > > > > > > Hi Albert, > > > > > > > ÃÆ'‚ > > > > > > > Your application/system is quite demanding :} > > > > > > > ÃÆ'‚ > > > > > > > That portion of GCode you sent is 100,000 lines of all G1 > > > > > movements with typical tiny lengths of only 50~150um.ÃÆ'‚ The > > > > > collinear tolerance of 0.001 (25um) will cause many blocksÃÆ'‚ to > > > > > be combined together so the real required vector rate will be more d > > > > > ependent on the path curvature rather than the original GCode.ÃÆ'‚ > > > > > > > ÃÆ'‚ > > > > > > > On my system (i7 W7 32-bit) it runs ok with a facet angle 0f 0.4 > > > > > degrees butÃÆ'‚ can't keep up at 0.3 degrees. > > > > > > > ÃÆ'‚ > > > > > > > I did some calculations for a system like yours with 40 in/sec2 > > > > > acceleration.ÃÆ'‚ Here is what I find: > > > > > > > ÃÆ'‚ > > > > > > > That acceleration allows going around a 0.1 inch radius circle > > > > > at 120ipm (which is about 3 circles per second) > > > > > > > ÃÆ'‚ > > > > > > > With a facet angle setting of 1.5 degrees that requires 240 > > > > > segments per circle which would require 720 segments per second > > > > > > > ÃÆ'‚ > > > > > > > I created a diagnostic test GCode file that programmatically > > > > > generates an N sided Polygon 100 timesÃÆ'‚ with radius 0.1 inches > > > > > and allows you to specify the number of polygon sides (it also gradu > > > > > ally shifts the "circle" to make the plot more interesting).ÃÆ'‚ > > > > > (see attachment). ÃÆ'‚ When using the diagnostic make sure to set > > > > > the Collinear tolerance to zero and the Corner Rounding to zero so t > > > > > he original segments are used.ÃÆ'‚ By changing N (number of sides > > > > > we can determine the segment download rate). > > > > > > > ÃÆ'‚ > > > > > > > For example on my system it works with N = 200 and fails with N > > > > > = 240.ÃÆ'‚ The total time to do all 100 circles takes about 30 > > > > > seconds.ÃÆ'‚ So: > > > > > > > ÃÆ'‚ > > > > > > > 100 circles x 200 segments/circle / 30 seconds = 666 segments > > > > > per second. > > > > > > > ÃÆ'‚ > > > > > > > In simulation mode the Interpreter+Graphics finishes the job in > > > > > ~ 5 seconds so it isn't a significant factor > > > > > > > ÃÆ'‚ > > > > > > > With the GCode Window closed it simulates in ~ 4 seconds so the > > > > > Graphical Display is not aÃÆ'‚ significant factor. > > > > > > > ÃÆ'‚ > > > > > > > You might run the tests on your two systems to see if you get > > > > > drastically different results. > > > > > > > ÃÆ'‚ > > > > > > > But to try to answer your original question of why one CPU would > > > > > be significantly different I don't know.ÃÆ'‚ Google finds some > > > > > benchmarking tools but they areÃÆ'‚ mainly for High Speed USB > > > > > devices like flash drives.ÃÆ'‚ Some sites state if an older USB > > > > > 1.1 device is attached it slows everything down.ÃÆ'‚ > > > > > > > ÃÆ'‚ > > > > > > > Regards > > > > > > > TKÃÆ'‚ > > > > > > > ÃÆ'‚ > > > > > > > > > > > > > > From: albertplatek albertplatek86@ > > > > > > > To: DynoMotion@yahoogroups.com > > > > > > > Sent: Saturday, January 14, 2012 11:02 AM > > > > > > > Subject: [DynoMotion] Re: Coord buffer overflow again > > > > > > > > > > > > > > > > > > > > > ÃÆ'‚ > > > > > > > Hi Tom, > > > > > > > How to determine USB throughput on pc? or what pc/system is best > > > > > for KFlop? > > > > > > > > > > > > > > I have still sometimes cord buffer overflow on second pc even I > > > > > > > set lower accelaration 25 and bigger facet angle 2.5 it sometimes > > > > > > > overflow. Its very strange pc has windows xp clean installation, > > > > > > > Integrated Intel Graphics. > > > > > > > On first pc works perfect even with high acceleration 40, facet > > > > > angle 2.0 (also windows xp, but ati graphics) > > > > > > > > > > > > > > Break angle 30 > > > > > > > Look ahead 3 > > > > > > > Collinear tolerance 0.001 > > > > > > > Corner tolerance 0.005 > > > > > > > Vel 5 > > > > > > > > > > > > > > Best Regards > > > > > > > Albert Platek > > > > > > > > > > > > > > --- In DynoMotion@yahoogroups.com, Albert PÃÆ'‚³atek > > > > > albertplatek86@ wrote: > > > > > > > > > > > > > > > > Hi Tom, > > > > > > > > GCode file is attached with this message. > > > > > > > > Yes feedrate is 300ipm. > > > > > > > > What do you mean by "path generation slows down"? - i mean > > > > > that when I am > > > > > > > > changing view perspective in GViewer > > > > > > > > the "simulated tool" catching up end of buffered path. So > > > > > thats why I > > > > > > > > thought that maybe gviewer impact on performance somehow. > > > > > > > > > > > > > > > > Best Regards > > > > > > > > Albert PÃÆ'Æ'…ÃÆ'¢â‚¬Å¡atek > > > > > > > > > > > > > > > > > > > > > > > > 2012/1/13 Tom Kerekes tk@ > > > > > > > > > > > > > > > > > > > > > > > > > > Hi Albert, > > > > > > > > > > > > > > > > > > Yes please send the file. Is the feedrate set high 300ipm? > > > > > > > > > > > > > > > > > > What do you mean by "path generation slows down"? > > > > > > > > > > > > > > > > > > Thanks > > > > > > > > > TK > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > *From:* albertplatek albertplatek86@ > > > > > > > > > *To:* DynoMotion@yahoogroups.com > > > > > > > > > *Sent:* Thursday, January 12, 2012 3:35 PM > > > > > > > > > *Subject:* [DynoMotion] Re: Coord buffer overflow again > > > > > > > > > > > > > > > > > > > > > > > > > > > Hi Tom, > > > > > > > > > > > > > > > > > > My settings: > > > > > > > > > > > > > > > > > > Break angle 30 > > > > > > > > > Look ahead 3 > > > > > > > > > Collinear tolerance 0.001 > > > > > > > > > Corner tolerance 0.005 > > > > > > > > > Facet angle 2.0 (if I go to 1.5, i get cord buffer overflow > > > > > almost from > > > > > > > > > program start) > > > > > > > > > > > > > > > > > > I use Tau = 0.01; (with facet angle 2.0 machine movements > > > > > are fairly > > > > > > > > > smooth) > > > > > > > > > > > > > > > > > > All 3 axis has same Vel/Accel parameters: > > > > > > > > > Vel 5.0 > > > > > > > > > Accel 40 > > > > > > > > > > > > > > > > > > Toolpath is parallel along X-axis (contain a lot of small > > > > > arcs, I will > > > > > > > > > send file tomorrow) > > > > > > > > > > > > > > > > > > With this settings my first router with old laptop works ok > > > > > (Dell, core 2 > > > > > > > > > duo, ati graphics card), > > > > > > > > > but on second router (same settings) with new laptop dont > > > > > work, i often > > > > > > > > > get buffer overflows. (laptop Dell, i7 cpu, Win7 64bit, > > > > > Nvidia Gt555) > > > > > > > > > > > > > > > > > > (Btw. when GViewer is open and I move view by mouse path > > > > > generation slows > > > > > > > > > down, is it normal?) > > > > > > > > > > > > > > > > > > Best Regards > > > > > > > > > Albert PÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚³atek > > > > > > > > > > > > > > > > > > --- In DynoMotion@yahoogroups.com, Tom Kerekes <tk@> wrote: > > > > > > > > > > > > > > > > > > > > Hi Albert, > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ > > > > > > > > > > Could you send me the Gcode and all your settings?ÃÆ'Æ'Æ'ÃÆ' > > > > > ¢â‚¬Å¡ I don't understand > > > > > > > > > the collinear tolerance of 1?ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ That > > > > > seems huge.ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ Did you mean 0.001? > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ > > > > > > > > > > A corner rounding facet angle too small might be > > > > > generating too many > > > > > > > > > small segments.ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ Are you using the > > > > > Coordinated Motion Low Pass Filter > > > > > > > > > Smoothing option?ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ That might help make > > > > > as smooth or smoother motion with > > > > > > > > > larger facets. > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ > > > > > > > > > > The USB throughput can vary from one machine to the other. > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ > > > > > > > > > > No data for the GViewer comes from KFLOP it all comes from > > > > > the > > > > > > > > > trajectory planner on the PC. > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ > > > > > > > > > > Regards > > > > > > > > > > TK > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ > > > > > > > > > > > > > > > > > > > > From: albertplatek <albertplatek86@> > > > > > > > > > > To: DynoMotion@yahoogroups.com > > > > > > > > > > Sent: Thursday, January 12, 2012 11:11 AM > > > > > > > > > > Subject: [DynoMotion] Coord buffer overflow again > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ > > > > > > > > > > Hi Tom, > > > > > > > > > > > > > > > > > > > > 1. Why on 1 pc kmotioncnc works ok, but on second not, > > > > > with same > > > > > > > > > settings? > > > > > > > > > > (collinear tollerance 1) > > > > > > > > > > same file, same accelerations etc. > > > > > > > > > > on second I have only usb 3.0 ports. > > > > > > > > > > > > > > > > > > > > 2. How much data come to Gviewer from kflop? if a lot is > > > > > it possible > > > > > > > > > > to bypass this if I dont need to use G viewer? > > > > > > > > > > > > > > > > > > > > I make a lot of reliefs in wood, with kinda high > > > > > acceleleration 40, if > > > > > > > > > > will be any way to get more performance from kflop with > > > > > smoothing > > > > > > > > > feature will be very good. > > > > > > > > > > > > > > > > > > > > Best Regards > > > > > > > > > > Albert PÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚³atek > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > Group: DynoMotion Message: 3753 From: Tom Kerekes Date: 2/6/2012 Subject: Re: Coord buffer overflow again Attachments : WatchXAccel-3753.c Hi Albert,   Your program seems reasonable but the numbers are crazy.   So are you saying that it did not report any errors until after 10 minutes then printed those?   I made a similar program and ran it over night and didn't see such a problem.  See attached.  It displays the velocity and acceleration into unused axes 6 and 7 Destinations on the Axis Screen. It also captures the Max Velocity and Acceleration and displays those as 6 and 7 Positions.   To be honest I see some tiny glitches and anomolies probably due to numerical rounding that I would like to better understand, but with the smoothing filter of 8ms it attenuates them way down.  The biggest acceleration spike I see in the entire multi hour job is 90 in/sec2.   Did you have the smoothing turned on?   Is your Resolution 254,000 steps per inch?   You don't have an MPG connected do you?   I will try your program to see if I get similar results.   Thanks TK Group: DynoMotion Message: 3777 From: Albert Płatek Date: 2/7/2012 Subject: Re: Coord buffer overflow again [1 Attachment] Hi Tom, "So are you saying that it did not report any errors until after 10 minutes then printed those?" - It was printing accel peak around 60in/sec2 sometimes from start and after 10 min this big error.  Second job execute was ok by about 20min then again error.  Smoothing was Tau= 0.008.   I just made new test on other pc with file "test_tolerance_0.005.ngc" . And I got 1166in/sec2 acceleration peak... and it stalled X motor. It is same program as before but with toolpath tolerance of 0.005. I thought that maybe this will help. But it didnt. (old file Relief14.ngc had toolpath tolerance of 0.01) Small programs works even with extream acceleration of 60, but this program and other bigger programs fail even with acceleration of 20. Here all my settings: Trajectory planner Break Angle 30 degrees Look Ahead 3 sec Collinear Tolerance 0.0005 in Corner Tolerance 0.01 in Facet Angle 1.5 Axis Parameters X Resolution 254000 Vel 5 Accel 40 Y Resolution 127000 Vel 4 Accel 10 Z Resolution 254000 Vel 5 Accel 40 A Resolution 7055.5555 Vel 100 Accel 15 (I dont use this axis in programs at moment) Smoothing Tau = 0.01 Gamepad disabled Lathe disabled I dont use MPG and it is not connected. Gcode program (test_tolerance_0.005.ngc) started from first line, I was only capturing Acceleration > 58. Wt, lut 07, 2012, 12:42:31  KMotion Program Started TAU = 0.010000 ACCEL= 58.677936 ACCEL= 58.927802 ACCEL= 58.960033 ACCEL= 60.769580 ACCEL= 58.512884 ACCEL= 58.924281 ACCEL= 59.024541 ACCEL= 58.133892 ACCEL= 61.956729 ACCEL= 59.006616 ACCEL= 60.105317 ACCEL= 58.828274 ACCEL= 60.622193 ACCEL= 58.060298 ACCEL= 60.837834 ACCEL= 58.185082 ACCEL= 58.925759 ACCEL= 59.444038 ACCEL= 61.017190 ACCEL= 59.082489 ACCEL= 60.176545 ACCEL= 58.677989 ACCEL= 60.129486 ACCEL= 59.890981 ACCEL= 63.703736 ACCEL= 58.300426 ACCEL= 58.390017 ACCEL= 64.994889 ACCEL= 61.344673 ACCEL= 61.264344 ACCEL= 60.206659 ACCEL= 61.110118 ACCEL= 58.861575 ACCEL= 62.844393 ACCEL= 58.712601 ACCEL= 62.743716 ACCEL= 59.412413 ACCEL= 59.090724 ACCEL= 59.607079 ACCEL= 58.240991 ACCEL= 58.200542 ACCEL= 62.400201 ACCEL= 64.401300 ACCEL= 60.829834 ACCEL= 60.765056 ACCEL= 58.584326 ACCEL= 59.368296 ACCEL= 58.845166 ACCEL= 63.692201 ACCEL= 59.398388 ACCEL= 58.492754 ACCEL= 58.201424 ACCEL= 59.713704 ACCEL= 59.488310 ACCEL= 58.663588 ACCEL= 62.887233 ACCEL= 58.256548 ACCEL= 58.483484 ACCEL= 58.042246 ACCEL= 58.518015 ACCEL= 61.312055 ACCEL= 60.224222 ACCEL= 60.041371 ACCEL= 61.019005 ACCEL= 59.857207 ACCEL= 60.667182 ACCEL= 62.517183 ACCEL= 62.978943 ACCEL= 60.672158 ACCEL= 62.764637 ACCEL= 60.755186 ACCEL= 58.177374 ACCEL= 58.144752 ACCEL= 62.349561 ACCEL= 59.094084 ACCEL= 59.707611 ACCEL= 60.042771 ACCEL= 59.493590 ACCEL= 58.381740 ACCEL= 58.552432 ACCEL= 60.385809 ACCEL= 61.290063 ACCEL= 63.347741 ACCEL= 59.724964 ACCEL= 66.035414 ACCEL= 58.827807 ACCEL= 60.704297 ACCEL= 60.619439 ACCEL= 60.255062 ACCEL= 59.246636 ACCEL= 58.966244 ACCEL= 62.119289 ACCEL= 61.110753 ACCEL= 64.018676 ACCEL= 63.911028 ACCEL= 60.811810 ACCEL= 59.177635 ACCEL= 61.196853 ACCEL= 58.680083 ACCEL= 59.527456 ACCEL= 58.087839 ACCEL= 59.119465 ACCEL= 62.853732 ACCEL= 59.093458 ACCEL= 59.161929 ACCEL= 61.429357 ACCEL= 58.803188 ACCEL= 59.175222 ACCEL= 60.219244 ACCEL= 58.596575 ACCEL= 58.764846 ACCEL= 58.426363 ACCEL= 60.086656 ACCEL= 59.610107 ACCEL= 59.229815 ACCEL= 58.044244 ACCEL= 63.695621 ACCEL= 58.131310 ACCEL= 59.107058 ACCEL= 58.927665 ACCEL= 61.165485 ACCEL= 58.315186 ACCEL= 60.871648 ACCEL= 60.176458 ACCEL= 59.214217 ACCEL= 59.208457 ACCEL= 62.621824 ACCEL= 58.398335 ACCEL= 61.152436 ACCEL= 58.604786 ACCEL= 58.729217 ACCEL= 59.923425 ACCEL= 59.677849 ACCEL= 58.123086 ACCEL= 60.433701 ACCEL= 58.291893 ACCEL= 60.252449 ACCEL= 60.417103 ACCEL= 60.433561 ACCEL= 61.127250 ACCEL= 61.402805 ACCEL= 58.202147 ACCEL= 62.299667 ACCEL= 62.973329 ACCEL= 61.678133 ACCEL= 58.407879 ACCEL= 59.317560 ACCEL= 58.968064 ACCEL= 58.048711 ACCEL= 58.288685 ACCEL= 58.241909 ACCEL= 59.930745 ACCEL= 59.450933 ACCEL= 59.439070 ACCEL= 58.143918 ACCEL= 58.259981 ACCEL= 58.482531 ACCEL= 58.101452 ACCEL= 60.411654 ACCEL= 60.631439 ACCEL= 62.469302 ACCEL= 60.847604 ACCEL= 58.350368 ACCEL= 1166.694794 ACCEL= 450.118253 ACCEL= 186.931892 ACCEL= 91.714558 ACCEL= 60.190257 ACCEL= 60.530275 ACCEL= 59.028122 Here I stopped program line was N463261. Smoothing TAU 0.01   In attechement  INIT.c config program. GCode program is in Yahoo Files folder "Albert CNC files "  (test_tolerance_0.005.part01.rar and test_tolerance_0.005.part02.rar)   Best Regards Albert Platek 2012/2/7 Tom Kerekes <tk@...>   [Attachment(s) from Tom Kerekes included below] Hi Albert,   Your program seems reasonable but the numbers are crazy.   So are you saying that it did not report any errors until after 10 minutes then printed those?   I made a similar program and ran it over night and didn't see such a problem.  See attached.  It displays the velocity and acceleration into unused axes 6 and 7 Destinations on the Axis Screen. It also captures the Max Velocity and Acceleration and displays those as 6 and 7 Positions.   To be honest I see some tiny glitches and anomolies probably due to numerical rounding that I would like to better understand, but with the smoothing filter of 8ms it attenuates them way down.  The biggest acceleration spike I see in the entire multi hour job is 90 in/sec2.   Did you have the smoothing turned on?   Is your Resolution 254,000 steps per inch?   You don't have an MPG connected do you?   I will try your program to see if I get similar results.   Thanks TK Group: DynoMotion Message: 3808 From: albertplatek Date: 2/8/2012 Subject: Re: Coord buffer overflow again [1 Attachment] Hi Tom, I think that in 99% I found bug, it was gamepad connected into usb. Even if it was disabled in KMmotionCNC, it was causing randomly this crazy jerk of acceleration. Today I tested this big gcode program 3 times without gamepad and everything is fine! :) Best Regards Albert P³atek --- In DynoMotion@yahoogroups.com, Albert P³atek <albertplatek86@...> wrote: > > Hi Tom, > "So are you saying that it did not report any errors until after 10 > minutes then printed those?" > - It was printing accel peak around 60in/sec2 sometimes from start and > after 10 min this big error. Second job execute was ok by about 20min then > again error. > Smoothing was Tau= 0.008. > > > I just made new test on other pc with file "test_tolerance_0.005.ngc" . And > I got *1166in/sec2* acceleration peak... and it stalled X motor. > > It is same program as before but with toolpath tolerance of 0.005. I > thought that maybe this will help. But it didnt. (old file Relief14.ngc had > toolpath tolerance of 0.01) > Small programs works even with extream acceleration of 60, but this program > and other bigger programs fail even with acceleration of 20. > > Here all my settings: > > *Trajectory planner* > > Break Angle 30 degrees > Look Ahead 3 sec > Collinear Tolerance 0.0005 in > Corner Tolerance 0.01 in > Facet Angle 1.5 > > *Axis Parameters* > > X Resolution 254000 Vel 5 Accel 40 > Y Resolution 127000 Vel 4 Accel 10 > Z Resolution 254000 Vel 5 Accel 40 > A Resolution 7055.5555 Vel 100 Accel 15 (I dont use this axis in programs > at moment) > > Smoothing Tau = 0.01 > > Gamepad disabled > Lathe disabled > I dont use MPG and it is not connected. > > Gcode program (test_tolerance_0.005.ngc) started from first line, I was > only capturing Acceleration > 58. > > Wt, lut 07, 2012, 12:42:31 KMotion Program Started > TAU = 0.010000 > ACCEL= 58.677936 > ACCEL= 58.927802 > ACCEL= 58.960033 > ACCEL= 60.769580 > ACCEL= 58.512884 > ACCEL= 58.924281 > ACCEL= 59.024541 > ACCEL= 58.133892 > ACCEL= 61.956729 > ACCEL= 59.006616 > ACCEL= 60.105317 > ACCEL= 58.828274 > ACCEL= 60.622193 > ACCEL= 58.060298 > ACCEL= 60.837834 > ACCEL= 58.185082 > ACCEL= 58.925759 > ACCEL= 59.444038 > ACCEL= 61.017190 > ACCEL= 59.082489 > ACCEL= 60.176545 > ACCEL= 58.677989 > ACCEL= 60.129486 > ACCEL= 59.890981 > ACCEL= 63.703736 > ACCEL= 58.300426 > ACCEL= 58.390017 > ACCEL= 64.994889 > ACCEL= 61.344673 > ACCEL= 61.264344 > ACCEL= 60.206659 > ACCEL= 61.110118 > ACCEL= 58.861575 > ACCEL= 62.844393 > ACCEL= 58.712601 > ACCEL= 62.743716 > ACCEL= 59.412413 > ACCEL= 59.090724 > ACCEL= 59.607079 > ACCEL= 58.240991 > ACCEL= 58.200542 > ACCEL= 62.400201 > ACCEL= 64.401300 > ACCEL= 60.829834 > ACCEL= 60.765056 > ACCEL= 58.584326 > ACCEL= 59.368296 > ACCEL= 58.845166 > ACCEL= 63.692201 > ACCEL= 59.398388 > ACCEL= 58.492754 > ACCEL= 58.201424 > ACCEL= 59.713704 > ACCEL= 59.488310 > ACCEL= 58.663588 > ACCEL= 62.887233 > ACCEL= 58.256548 > ACCEL= 58.483484 > ACCEL= 58.042246 > ACCEL= 58.518015 > ACCEL= 61.312055 > ACCEL= 60.224222 > ACCEL= 60.041371 > ACCEL= 61.019005 > ACCEL= 59.857207 > ACCEL= 60.667182 > ACCEL= 62.517183 > ACCEL= 62.978943 > ACCEL= 60.672158 > ACCEL= 62.764637 > ACCEL= 60.755186 > ACCEL= 58.177374 > ACCEL= 58.144752 > ACCEL= 62.349561 > ACCEL= 59.094084 > ACCEL= 59.707611 > ACCEL= 60.042771 > ACCEL= 59.493590 > ACCEL= 58.381740 > ACCEL= 58.552432 > ACCEL= 60.385809 > ACCEL= 61.290063 > ACCEL= 63.347741 > ACCEL= 59.724964 > ACCEL= 66.035414 > ACCEL= 58.827807 > ACCEL= 60.704297 > ACCEL= 60.619439 > ACCEL= 60.255062 > ACCEL= 59.246636 > ACCEL= 58.966244 > ACCEL= 62.119289 > ACCEL= 61.110753 > ACCEL= 64.018676 > ACCEL= 63.911028 > ACCEL= 60.811810 > ACCEL= 59.177635 > ACCEL= 61.196853 > ACCEL= 58.680083 > ACCEL= 59.527456 > ACCEL= 58.087839 > ACCEL= 59.119465 > ACCEL= 62.853732 > ACCEL= 59.093458 > ACCEL= 59.161929 > ACCEL= 61.429357 > ACCEL= 58.803188 > ACCEL= 59.175222 > ACCEL= 60.219244 > ACCEL= 58.596575 > ACCEL= 58.764846 > ACCEL= 58.426363 > ACCEL= 60.086656 > ACCEL= 59.610107 > ACCEL= 59.229815 > ACCEL= 58.044244 > ACCEL= 63.695621 > ACCEL= 58.131310 > ACCEL= 59.107058 > ACCEL= 58.927665 > ACCEL= 61.165485 > ACCEL= 58.315186 > ACCEL= 60.871648 > ACCEL= 60.176458 > ACCEL= 59.214217 > ACCEL= 59.208457 > ACCEL= 62.621824 > ACCEL= 58.398335 > ACCEL= 61.152436 > ACCEL= 58.604786 > ACCEL= 58.729217 > ACCEL= 59.923425 > ACCEL= 59.677849 > ACCEL= 58.123086 > ACCEL= 60.433701 > ACCEL= 58.291893 > ACCEL= 60.252449 > ACCEL= 60.417103 > ACCEL= 60.433561 > ACCEL= 61.127250 > ACCEL= 61.402805 > ACCEL= 58.202147 > ACCEL= 62.299667 > ACCEL= 62.973329 > ACCEL= 61.678133 > ACCEL= 58.407879 > ACCEL= 59.317560 > ACCEL= 58.968064 > ACCEL= 58.048711 > ACCEL= 58.288685 > ACCEL= 58.241909 > ACCEL= 59.930745 > ACCEL= 59.450933 > ACCEL= 59.439070 > ACCEL= 58.143918 > ACCEL= 58.259981 > ACCEL= 58.482531 > ACCEL= 58.101452 > ACCEL= 60.411654 > ACCEL= 60.631439 > ACCEL= 62.469302 > ACCEL= 60.847604 > ACCEL= 58.350368 > *ACCEL= 1166.694794 > ACCEL= 450.118253* > ACCEL= 186.931892 > ACCEL= 91.714558 > ACCEL= 60.190257 > ACCEL= 60.530275 > ACCEL= 59.028122 > > Here I stopped program line was N463261. > Smoothing TAU 0.01 > > In attechement INIT.c config program. > GCode program is in Yahoo Files folder "Albert CNC > files " (test_tolerance_0.005.part01.rar > and test_tolerance_0.005.part02.rar) > > Best Regards > Albert Platek > > > 2012/2/7 Tom Kerekes <tk@...> > > > ** > > > > [Attachment(s)<https://mail.google.com/mail/html/compose/static_files/blank_quirks.html#1355587bcb0eb6ec_TopText>from Tom Kerekes included below] > > > > Hi Albert, > > > > Your program seems reasonable but the numbers are crazy. > > > > So are you saying that it did not report any errors until after 10 minutes > > then printed those? > > > > I made a similar program and ran it over night and didn't see such a > > problem. See attached. It displays the velocity and acceleration into > > unused axes 6 and 7 Destinations on the Axis Screen. It also captures the > > Max Velocity and Acceleration and displays those as 6 and 7 Positions. > > > > To be honest I see some tiny glitches and anomolies probably due to > > numerical rounding that I would like to better understand, but with the > > smoothing filter of 8ms it attenuates them way down. The biggest > > acceleration spike I see in the entire multi hour job is 90 in/sec2. > > > > Did you have the smoothing turned on? > > > > Is your Resolution 254,000 steps per inch? > > > > You don't have an MPG connected do you? > > > > I will try your program to see if I get similar results. > > > > Thanks > > TK > > > > *From:* albertplatek <albertplatek86@...> > > *To:* DynoMotion@yahoogroups.com > > *Sent:* Monday, February 6, 2012 8:24 AM > > *Subject:* [DynoMotion] Re: Coord buffer overflow again > > > > > > Hi Tom, > > > > Sorry Tom but still this happens randomly :/ I was just lucky that after > > firmware flash 1 job was ok. > > I use ver 4.28. > > > > I dont know how to determine line numner of this jerk, but you can see > > huge acceleration! X motor stalled. > > It was runinng about 10min from program start. > > Acceleration was set 32in/sec2 > > Break Angle 30 > > Tau 0.008 > > Collinear Tolerance 0.0005 > > Corener Tolerance 0.01 > > Facet angle 0.5 > > Vel 4in/sec > > > > Here is my simple C program to check acceleration please check if it is > > correct: > > > > #include "KMotionDef.h" > > #define channel ch0 > > #define time 0.00018 > > main() > > { > > double X1,X2,speed1,speed2,accel; > > int k; > > > > while(1) > > { > > for (k=0; k<1; k++) WaitNextTimeSlice(); > > X1 = channel->Dest; > > for (k=0; k<1; k++) WaitNextTimeSlice(); > > X2 = channel->Dest; > > speed1=(X2-X1)/(time); > > > > for (k=0; k<1; k++) WaitNextTimeSlice(); > > X1 = channel->Dest; > > for (k=0; k<1; k++) WaitNextTimeSlice(); > > X2 = channel->Dest; > > speed2=(X2-X1)/(time); > > > > accel=(speed1-speed2)/(time); > > accel=(accel/254000)/2; > > if(accel>45) > > { > > printf("AccelError= %f\n",accel); > > } > > } > > } > > > > Here log: > > > > AccelError= 45.009219 > > AccelError= 45.229875 > > AccelError= 46.294106 > > AccelError= 45.055529 > > AccelError= 45.747547 > > AccelError= 45.070729 > > AccelError= 45.360541 > > AccelError= 45.765590 > > AccelError= 46.302076 > > AccelError= 46.775562 > > AccelError= 47.051896 > > AccelError= 45.225101 > > AccelError= 45.730674 > > AccelError= 46.758273 > > AccelError= 46.838768 > > AccelError= 45.698572 > > AccelError= 46.918682 > > AccelError= 45.204219 > > AccelError= 45.900150 > > AccelError= 47.023519 > > AccelError= 47.551550 > > AccelError= 48.022541 > > AccelError= 46.395295 > > AccelError= 81526.314840 > > AccelError= 72853.056587 > > AccelError= 65101.872298 > > AccelError= 58175.992747 > > AccelError= 51986.611437 > > AccelError= 46456.046360 > > AccelError= 41513.718492 > > AccelError= 37097.300673 > > AccelError= 33150.797142 > > AccelError= 29622.770379 > > AccelError= 26468.350534 > > AccelError= 23652.797849 > > AccelError= 21136.323622 > > AccelError= 18888.213633 > > AccelError= 16879.065077 > > AccelError= 15083.474245 > > AccelError= 13479.186003 > > AccelError= 12045.566722 > > AccelError= 10764.204608 > > AccelError= 9619.483512 > > AccelError= 8596.141212 > > AccelError= 7682.015002 > > AccelError= 6866.662006 > > AccelError= 6139.961139 > > AccelError= 5490.446488 > > AccelError= 4910.042286 > > AccelError= 4391.574405 > > AccelError= 3928.114510 > > AccelError= 3514.000485 > > AccelError= 3144.002655 > > AccelError= 2813.316136 > > AccelError= 2517.500643 > > AccelError= 2253.490792 > > AccelError= 2017.432428 > > AccelError= 1806.266152 > > AccelError= 1617.706032 > > AccelError= 1448.982225 > > AccelError= 1298.398178 > > AccelError= 1163.621865 > > AccelError= 1043.472200 > > AccelError= 935.821407 > > AccelError= 839.766485 > > AccelError= 753.863074 > > AccelError= 677.101783 > > AccelError= 608.587862 > > AccelError= 547.209083 > > AccelError= 492.589836 > > AccelError= 443.554631 > > AccelError= 399.735924 > > AccelError= 360.639632 > > AccelError= 325.643422 > > AccelError= 294.370902 > > AccelError= 266.425861 > > AccelError= 241.478288 > > AccelError= 219.191672 > > AccelError= 199.246085 > > AccelError= 181.422761 > > AccelError= 165.532090 > > AccelError= 151.295897 > > AccelError= 138.574472 > > AccelError= 127.206640 > > AccelError= 117.084135 > > AccelError= 108.003075 > > AccelError= 99.888264 > > AccelError= 92.598183 > > AccelError= 86.122270 > > AccelError= 76.770777 > > AccelError= 65.225363 > > AccelError= 54.908134 > > AccelError= 45.688617 > > AccelError= 45.070435 > > AccelError= 45.411744 > > AccelError= 45.770036 > > AccelError= 45.871485 > > AccelError= 45.239885 > > AccelError= 45.631836 > > AccelError= 45.001237 > > AccelError= 45.373033 > > AccelError= 45.577907 > > AccelError= 46.322608 > > AccelError= 46.958820 > > AccelError= 46.038556 > > AccelError= 45.018088 > > > > If you test it on your PC and will be ok, then maybe something is wrong > > with my laptop. Then I will test on other PC. > > > > Best Regards > > Albert P³atek > > > > --- In DynoMotion@yahoogroups.com, "albertplatek" <albertplatek86@> > > wrote: > > > > > > Hi Tom, > > > > > > Problem solved (probably), weird but after re-flash firmware 4.28 > > > everything is working ok, no axis stall. > > > > > > btw. Experimental version KMotion429x.exe is not working, when I start > > > execute gcode I get message "G Code Error GCode Aborted", this error > > > occurs on any gcode program even on only few G01 lines. > > > > > > Tom if you have C program that watch for too high of acceleration > > > please post it. If again in future something strange will > > > happen I will try find out the problem. > > > I am still not sure what was the problem in my system that axis > > > were stalling. > > > > > > Best Regards > > > Albert P³atek > > > > > > > > > > > > --- In DynoMotion@yahoogroups.com, Albert P³atek <albertplatek86@> > > wrote: > > > > > > > > Hi Tom, > > > > > > > > Thank you for help. > > > > Mostly stall X axis, and sometimes Z. > > > > Y axis seem to be ok. > > > > I uploaded file as an attachment (zip). > > > > > > > > Best Regards > > > > Albert Platek > > > > > > > > 2012/2/5 Tom Kerekes <tk@> > > > > > > > > > ** > > > > > > > > > > > > > > > Hi Albert, > > > > > > > > > > Thanks for the very complete problem description, I can't tell you > > how > > > > > much I appreciate that, but I can't think of anything that would > > cause > > > > > this. Anything with large files I might expect some processing issue > > but > > > > > that would result in a buffer underflow and an error message not > > incorrect > > > > > motion that would result in a stall. That delay I described in your > > email > > > > > only happens when "Execute" is pushed so I don't think it is related. > > > > > > > > > > There is an experimental version KMotion429x.exe on the download site > > > > > that has this delay eliminated and also faster USB throughput, but I > > don't > > > > > think either of those are related to this problem. > > > > > > > > > > Can you post the file somewhere? I'm thinking to write a C program > > that > > > > > would run in KFLOP and watch for too high of acceleration or > > velocity by > > > > > monitoring how ch0->Dest changes. If it detects something bad that > > could > > > > > be used as a trigger to help us track it down. > > > > > > > > > > Which axes loose position? > > > > > > > > > > Regards > > > > > TK > > > > > > > > > > > > > > > *From:* albertplatek <albertplatek86@> > > > > > *To:* DynoMotion@yahoogroups.com > > > > > *Sent:* Sunday, February 5, 2012 5:27 AM > > > > > > > > > > *Subject:* [DynoMotion] Re: Coord buffer overflow again > > > > > > > > > > > > > > > Hi Tom, > > > > > > > > > > My settings: > > > > > Break angle 30 deg > > > > > Look ahead 3 sec > > > > > Collinear tolerance 0.001 in > > > > > Corner tolerance 0.003 in > > > > > Facet angle 1.5 deg > > > > > Tau = 0.008 sec > > > > > XYZ Accelarations 40 in/sec2 > > > > > XYZ Velocity 4.5 in/sec > > > > > In gcode file I use max Vel 270 in/min (4.5 in/sec) > > > > > > > > > > I have strange problem, everything was ok until I started machining > > > > > bigger reliefs. > > > > > At half of the job my stepper motors lost position > > > > > probably becouse of high jerk of acceleration. > > > > > Firstly I thought that maybe axis accelerations for my system are to > > > > > high(40in/sec2). So I changed it to 20in/sec2 for test. > > > > > But didnt help, still stepper motors stall at half of the gcode file. > > > > > > > > > > This file contain 548389 lines, my steppers stall between 226000 - > > 274000 > > > > > line but not always in same place. > > > > > > > > > > I tried to find line where the problem occurs, but when I found line > > with > > > > > problem and tried to execute 500 lines before, it was ok. Motors > > didnt > > > > > stall. > > > > > I found that motors stall when KMotionCNC start execute at least > > > > > 100 000 lines before line number 274000. > > > > > > > > > > The problem disappear when I set break angle 5deg. > > > > > > > > > > "This diagnostic pointed out a side problem I hadn't noticed before. > > There > > > > > is an annoying ~30 second delay when it runs the second time. The > > > > > Interpreter keeps a record of all its state changes which allow > > Halts to go > > > > > back to any previous state. In this test of 100x500 subroutine calls > > > > > results in about 1 million state changes. Ironically creating the > > data, > > > > > making use of the data all happens in the blink of an eye, but > > discarding > > > > > the data (CList.RemoveAll) takes 30 seconds! We will find a better > > way in > > > > > the next version. For now just don't be surprised if there is an > > annoying > > > > > delay on a second run of a big job." > > > > > > > > > > Can this "delay" make that my stepper motors stall? > > > > > > > > > > Best Regards > > > > > Albert P³atek > > > > > > > > > > --- In DynoMotion@yahoogroups.com, Albert <albertplatek86@> wrote: > > > > > > > > > > > > Hi Tom, > > > > > > > > > > > > Thank you. > > > > > > Btw. I am trying to find out what changes make > > > > > > Fidelizer in Windows. If I will find something > > > > > > important I will let you know. > > > > > > Best regards > > > > > > Albert Platek > > > > > > > > > > > > > > > > > > Dnia 2012-01-18 o godz. 23:47 Tom Kerekes <tk@> napisaÅ‚ > > > > > > (a): > > > > > > > > > > > > > Hi Albert, > > > > > > > > > > > > > > Just wanted to mention that I now do notice an improvement with > > > > > > > Fidelizer. I don't understand why I didn't before. I might have > > > > > > > made a mistake or something changed. > > > > > > > > > > > > > > I sent an email to FTDI (USB chip we use) regarding it and the > > USB > > > > > > > 3.0 port issue to see if they have any answers. I figure it is > > > > > > > worth the cost of an email :} > > > > > > > > > > > > > > Still working on some optimization... > > > > > > > > > > > > > > Regards > > > > > > > TK > > > > > > > > > > > > > > > > > > > > > From: albertplatek <albertplatek86@> > > > > > > > To: DynoMotion@yahoogroups.com > > > > > > > Sent: Monday, January 16, 2012 12:01 PM > > > > > > > Subject: [DynoMotion] Re: Coord buffer overflow again > > > > > > > > > > > > > > > > > > > > > Hi Tom, > > > > > > > Thank you for making changes in new version to increase the > > segment > > > > > > > rate. > > > > > > > > > > > > > > Btw: > > > > > > > I installed on second pc Windows 7 (Core2Duo cpu), and aslo > > tested > > > > > > > with fidelizer. Strange but didn't make difference, as in your > > > > > > > computer. > > > > > > > > > > > > > > I cant get why in this particular case (pc with WIN7 i7 cpu) it > > helps. > > > > > > > Maybe it depends on type of host controller... > > > > > > > > > > > > > > Pc with WIN7 64bit (Core2Duo) has > > > > > > > SiS 7001 PCI to USB Open Host Controller > > > > > > > > > > > > > > Pc with WIN7 64bit (i7) has > > > > > > > Intel(R) 6 Series/C200 Series Chipset Family USB Enhanced Host > > > > > > > Controller > > > > > > > > > > > > > > This computer has aslo another usb ports: Renesas Electronics USB > > > > > > > 3.0 Host Controller > > > > > > > which works very bad with KFlop. And on this port fidelizer make > > no > > > > > > > difference. > > > > > > > > > > > > > > Best Regards > > > > > > > Albert Platek > > > > > > > > > > > > > > --- In DynoMotion@yahoogroups.com, Tom Kerekes <tk@> wrote: > > > > > > > > > > > > > > > > Hi Albert, > > > > > > > > ÃÆ'‚ > > > > > > > > Thanks for that interesting information. > > > > > > > > ÃÆ'‚ > > > > > > > > I tried the Fidelizer on my i7 W7 32 machine and it didn't > > seem to > > > > > > > make any difference. > > > > > > > > ÃÆ'‚ > > > > > > > > I'm trying some things to increase the segment rate by sending > > > > > > > blocks of segments and optimizing how they are handled in > > KFLOP.ÃÆ'‚ I > > > > > > > f this bears fruit I'll let you know. > > > > > > > > ÃÆ'‚ > > > > > > > > This diagnostic pointed out a side problem I hadn't noticed > > before > > > > > > > .ÃÆ'‚ There is an annoying ~30 second delay when it runs the > > second ti > > > > > > > me.ÃÆ'‚ The Interpreter keeps a record of all its state changes > > which > > > > > > > allow Halts to go back to any previous state.ÃÆ'‚ In this test > > of 100x > > > > > > > 500 subroutine calls results in about 1 million state > > changes.ÃÆ'‚ Iro > > > > > > > nically creating the data, making use of the data all happens in > > the > > > > > > > blink of an eye, but discarding the data (CList.RemoveAll) takes > > 30 > > > > > > > seconds!ÃÆ'‚ We will find a better way in the next > > version.ÃÆ'‚ For now > > > > > > > just don't be surprisedÃÆ'‚ if there is an annoying delay on a > > second > > > > > > > run of a big job. > > > > > > > > ÃÆ'‚ > > > > > > > > Regards > > > > > > > > TK > > > > > > > > ÃÆ'‚ > > > > > > > > > > > > > > > > > > > > > > > > ________________________________ > > > > > > > > From: albertplatek <albertplatek86@> > > > > > > > > To: DynoMotion@yahoogroups.com > > > > > > > > Sent: Sunday, January 15, 2012 8:37 AM > > > > > > > > Subject: [DynoMotion] Re: Coord buffer overflow again > > > > > > > > > > > > > > > > > > > > > > > > ÃÆ'‚ > > > > > > > > Hi Tom, > > > > > > > > First thank you Tom for the test file. > > > > > > > > I made few tests and I got interesting results :) > > > > > > > > > > > > > > > > PC No1: > > > > > > > > > > > > > > > > Intel Pentium 4 2.6Ghz WinXP 32bit port USB2.0 > > > > > > > > Max segments: 110ÃÆ'‚ > > > > > > > > Time: 31s > > > > > > > > 100 circles x 110 segments/circle / 31 seconds =354 segments > > per > > > > > > > second. > > > > > > > > > > > > > > > > PC No2: > > > > > > > > > > > > > > > > Intel Core2Duo 2.0 Ghz WinXP 32bit port USB2.0 > > > > > > > > Max segments: 140ÃÆ'‚ > > > > > > > > Time: 31s > > > > > > > > 100 circles x 140 segments/circle / 31 seconds = 451 segments > > per > > > > > > > second. > > > > > > > > > > > > > > > > PC No3: > > > > > > > > > > > > > > > > i7 2670QM Win7 64bit port USB3.0 > > > > > > > > Max segments:30ÃÆ'‚ > > > > > > > > Time:31s > > > > > > > > 100 circles x 60 segments/circle / 31 seconds = 193 segments > > per > > > > > > > second. > > > > > > > > (thats why I had problem here, usb 3.0 fail) > > > > > > > > > > > > > > > > (Luckily I had 1 USB 2.0 port) > > > > > > > > > > > > > > > > i7 2670QM Win7 64bit port USB2.0 > > > > > > > > Max segments:170 > > > > > > > > Time:31s > > > > > > > > 100 circles x 170 segments/circle / 31 seconds = 548 segments > > per > > > > > > > second. > > > > > > > > > > > > > > > > i7 2670QM Win7 64bit port USB2.0 (with 1 core affinity for > > > > > > > KmotionCNC and KmotionServer) > > > > > > > > Max segments:200 > > > > > > > > Time:31s > > > > > > > > 100 circles x 200 segments/circle / 31 seconds =645 segments > > per > > > > > > > second. > > > > > > > > > > > > > > > > I found on net some interesting software called Fidelizer, its > > > > > > > software to optimize win7/vista for audio workstation, but it > > aslo > > > > > > > tweak somehow usb throughput. > > > > > > > > More detail here: /www.windowsxlive.net/fidelizer > > > > > > > > > > > > > > > > i7 2670QM Win7 64bit port USB2.0 (Fidelizer) > > > > > > > > Max segments:410 > > > > > > > > Time:31s > > > > > > > > 100 circles x 410 segments/circle / 31 seconds = 1320 segments > > per > > > > > > > second. (wow! tested few times with same result) > > > > > > > > > > > > > > > > Be aware of new USB 3.0 ports they may not work with good > > > > > > > performance with KFlop, maybe becouse of bug in drivers. Windows > > 7 > > > > > > > dont see them without additional drivers. > > > > > > > > Fidelizer software only improved performance on pc with ÃÆ'‚ > > Win7. > > > > > > > > On WinXP seem to not work. I will soon try install on "PC No2" > > > > > > > Windows 7 and then try fidelizer also. > > > > > > > > > > > > > > > > Best regards > > > > > > > > Albert PÃÆ'‚³atek > > > > > > > > --- In DynoMotion@yahoogroups.com, Tom Kerekes <tk@> wrote: > > > > > > > > > > > > > > > > > > Hi Albert, > > > > > > > > > ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚ > > > > > > > > > Your application/system is quite demanding :} > > > > > > > > > ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚ > > > > > > > > > That portion of GCode you sent is 100,000 lines of all G1 > > > > > > > movements with typical tiny lengths of only > > 50~150um.ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚ The > > > > > > > collinear tolerance of 0.001 (25um) will cause many > > blocksÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚ to > > > > > > > be combined together so the real required vector rate will be > > more d > > > > > > > ependent on the path curvature rather than the original > > GCode.ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚ > > > > > > > > > ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚ > > > > > > > > > On my system (i7 W7 32-bit) it runs ok with a facet angle 0f > > 0.4 > > > > > > > degrees butÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚ can't keep up at 0.3 degrees. > > > > > > > > > ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚ > > > > > > > > > I did some calculations for a system like yours with 40 > > in/sec2 > > > > > > > acceleration.ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚ Here is what I find: > > > > > > > > > ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚ > > > > > > > > > That acceleration allows going around a 0.1 inch radius > > circle > > > > > > > at 120ipm (which is about 3 circles per second) > > > > > > > > > ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚ > > > > > > > > > With a facet angle setting of 1.5 degrees that requires 240 > > > > > > > segments per circle which would require 720 segments per second > > > > > > > > > ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚ > > > > > > > > > I created a diagnostic test GCode file that programmatically > > > > > > > generates an N sided Polygon 100 timesÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚ with > > radius 0.1 inches > > > > > > > and allows you to specify the number of polygon sides (it also > > gradu > > > > > > > ally shifts the "circle" to make the plot more > > interesting).ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚ > > > > > > > (see attachment). ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚ When using the diagnostic > > make sure to set > > > > > > > the Collinear tolerance to zero and the Corner Rounding to zero > > so t > > > > > > > he original segments are used.ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚ By changing N > > (number of sides > > > > > > > we can determine the segment download rate). > > > > > > > > > ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚ > > > > > > > > > For example on my system it works with N = 200 and fails > > with N > > > > > > > = 240.ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚ The total time to do all 100 circles > > takes about 30 > > > > > > > seconds.ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚ So: > > > > > > > > > ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚ > > > > > > > > > 100 circles x 200 segments/circle / 30 seconds = 666 segments > > > > > > > per second. > > > > > > > > > ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚ > > > > > > > > > In simulation mode the Interpreter+Graphics finishes the job > > in > > > > > > > ~ 5 seconds so it isn't a significant factor > > > > > > > > > ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚ > > > > > > > > > With the GCode Window closed it simulates in ~ 4 seconds so > > the > > > > > > > Graphical Display is not aÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚ significant factor. > > > > > > > > > ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚ > > > > > > > > > You might run the tests on your two systems to see if you get > > > > > > > drastically different results. > > > > > > > > > ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚ > > > > > > > > > But to try to answer your original question of why one CPU > > would > > > > > > > be significantly different I don't know.ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚ Google > > finds some > > > > > > > benchmarking tools but they areÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚ mainly for High > > Speed USB > > > > > > > devices like flash drives.ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚ Some sites state if > > an older USB > > > > > > > 1.1 device is attached it slows everything down.ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚ > > > > > > > > > ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚ > > > > > > > > > Regards > > > > > > > > > TKÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚ > > > > > > > > > ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚ > > > > > > > > > > > > > > > > > > From: albertplatek albertplatek86@ > > > > > > > > > To: DynoMotion@yahoogroups.com > > > > > > > > > Sent: Saturday, January 14, 2012 11:02 AM > > > > > > > > > Subject: [DynoMotion] Re: Coord buffer overflow again > > > > > > > > > > > > > > > > > > > > > > > > > > > ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚ > > > > > > > > > Hi Tom, > > > > > > > > > How to determine USB throughput on pc? or what pc/system is > > best > > > > > > > for KFlop? > > > > > > > > > > > > > > > > > > I have still sometimes cord buffer overflow on second pc > > even I > > > > > > > > > set lower accelaration 25 and bigger facet angle 2.5 it > > sometimes > > > > > > > > > overflow. Its very strange pc has windows xp clean > > installation, > > > > > > > > > Integrated Intel Graphics. > > > > > > > > > On first pc works perfect even with high acceleration 40, > > facet > > > > > > > angle 2.0 (also windows xp, but ati graphics) > > > > > > > > > > > > > > > > > > Break angle 30 > > > > > > > > > Look ahead 3 > > > > > > > > > Collinear tolerance 0.001 > > > > > > > > > Corner tolerance 0.005 > > > > > > > > > Vel 5 > > > > > > > > > > > > > > > > > > Best Regards > > > > > > > > > Albert Platek > > > > > > > > > > > > > > > > > > --- In DynoMotion@yahoogroups.com, Albert > > PÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚³atek > > > > > > > albertplatek86@ wrote: > > > > > > > > > > > > > > > > > > > > Hi Tom, > > > > > > > > > > GCode file is attached with this message. > > > > > > > > > > Yes feedrate is 300ipm. > > > > > > > > > > What do you mean by "path generation slows down"? - i mean > > > > > > > that when I am > > > > > > > > > > changing view perspective in GViewer > > > > > > > > > > the "simulated tool" catching up end of buffered path. So > > > > > > > thats why I > > > > > > > > > > thought that maybe gviewer impact on performance somehow. > > > > > > > > > > > > > > > > > > > > Best Regards > > > > > > > > > > Albert > > PÃÆ'Æ'Æ'ÃÆ'¢â‚¬Â¦ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡atek > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > 2012/1/13 Tom Kerekes tk@ > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > Hi Albert, > > > > > > > > > > > > > > > > > > > > > > Yes please send the file. Is the feedrate set high > > 300ipm? > > > > > > > > > > > > > > > > > > > > > > What do you mean by "path generation slows down"? > > > > > > > > > > > > > > > > > > > > > > Thanks > > > > > > > > > > > TK > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > *From:* albertplatek albertplatek86@ > > > > > > > > > > > *To:* DynoMotion@yahoogroups.com > > > > > > > > > > > *Sent:* Thursday, January 12, 2012 3:35 PM > > > > > > > > > > > *Subject:* [DynoMotion] Re: Coord buffer overflow again > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > Hi Tom, > > > > > > > > > > > > > > > > > > > > > > My settings: > > > > > > > > > > > > > > > > > > > > > > Break angle 30 > > > > > > > > > > > Look ahead 3 > > > > > > > > > > > Collinear tolerance 0.001 > > > > > > > > > > > Corner tolerance 0.005 > > > > > > > > > > > Facet angle 2.0 (if I go to 1.5, i get cord buffer > > overflow > > > > > > > almost from > > > > > > > > > > > program start) > > > > > > > > > > > > > > > > > > > > > > I use Tau = 0.01; (with facet angle 2.0 machine movements > > > > > > > are fairly > > > > > > > > > > > smooth) > > > > > > > > > > > > > > > > > > > > > > All 3 axis has same Vel/Accel parameters: > > > > > > > > > > > Vel 5.0 > > > > > > > > > > > Accel 40 > > > > > > > > > > > > > > > > > > > > > > Toolpath is parallel along X-axis (contain a lot of small > > > > > > > arcs, I will > > > > > > > > > > > send file tomorrow) > > > > > > > > > > > > > > > > > > > > > > With this settings my first router with old laptop works > > ok > > > > > > > (Dell, core 2 > > > > > > > > > > > duo, ati graphics card), > > > > > > > > > > > but on second router (same settings) with new laptop dont > > > > > > > work, i often > > > > > > > > > > > get buffer overflows. (laptop Dell, i7 cpu, Win7 64bit, > > > > > > > Nvidia Gt555) > > > > > > > > > > > > > > > > > > > > > > (Btw. when GViewer is open and I move view by mouse path > > > > > > > generation slows > > > > > > > > > > > down, is it normal?) > > > > > > > > > > > > > > > > > > > > > > Best Regards > > > > > > > > > > > Albert PÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚³atek > > > > > > > > > > > > > > > > > > > > > > --- In DynoMotion@yahoogroups.com, Tom Kerekes <tk@> > > wrote: > > > > > > > > > > > > > > > > > > > > > > > > Hi Albert, > > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ > > > > > > > > > > > > Could you send me the Gcode and all your > > settings?ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ' > > > > > > > ¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ I don't understand > > > > > > > > > > > the collinear tolerance of > > 1?ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ That > > > > > > > seems huge.ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ Did you > > mean 0.001? > > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ > > > > > > > > > > > > A corner rounding facet angle too small might be > > > > > > > generating too many > > > > > > > > > > > small > > segments.ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ Are you using the > > > > > > > Coordinated Motion Low Pass Filter > > > > > > > > > > > Smoothing > > option?ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ That might help make > > > > > > > as smooth or smoother motion with > > > > > > > > > > > larger facets. > > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ > > > > > > > > > > > > The USB throughput can vary from one machine to the > > other. > > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ > > > > > > > > > > > > No data for the GViewer comes from KFLOP it all comes > > from > > > > > > > the > > > > > > > > > > > trajectory planner on the PC. > > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ > > > > > > > > > > > > Regards > > > > > > > > > > > > TK > > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ > > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ > > > > > > > > > > > > > > > > > > > > > > > > From: albertplatek <albertplatek86@> > > > > > > > > > > > > To: DynoMotion@yahoogroups.com > > > > > > > > > > > > Sent: Thursday, January 12, 2012 11:11 AM > > > > > > > > > > > > Subject: [DynoMotion] Coord buffer overflow again > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ > > > > > > > > > > > > Hi Tom, > > > > > > > > > > > > > > > > > > > > > > > > 1. Why on 1 pc kmotioncnc works ok, but on second not, > > > > > > > with same > > > > > > > > > > > settings? > > > > > > > > > > > > (collinear tollerance 1) > > > > > > > > > > > > same file, same accelerations etc. > > > > > > > > > > > > on second I have only usb 3.0 ports. > > > > > > > > > > > > > > > > > > > > > > > > 2. How much data come to Gviewer from kflop? if a lot > > is > > > > > > > it possible > > > > > > > > > > > > to bypass this if I dont need to use G viewer? > > > > > > > > > > > > > > > > > > > > > > > > I make a lot of reliefs in wood, with kinda high > > > > > > > acceleleration 40, if > > > > > > > > > > > > will be any way to get more performance from kflop with > > > > > > > smoothing > > > > > > > > > > > feature will be very good. > > > > > > > > > > > > > > > > > > > > > > > > Best Regards > > > > > > > > > > > > Albert > > PÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚³atek > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > Group: DynoMotion Message: 3812 From: Tom Kerekes Date: 2/8/2012 Subject: Re: Coord buffer overflow again Hi Albert,   Wow that is strange.  I just double checked and V4.28 is in fact testing the Gampad enable option.  So I have no explaination how the Gampad could make a difference.  Possibly it somehow corrupts the USB communication - but that should generate an error.  Maybe it causes Coordinated Motion Segments to just get completely lost (skipped).   That is something I couldn't understand about the problem, there should be no randomness.  Everything we are doing should be completely deterministic.  The same GCode and the same Settings should generate the exact same motion every time.   Have you tried re-connecting the GamPad and pushing the buttons to see if you can force errors? Regards TK     Group: DynoMotion Message: 3814 From: albertplatek Date: 2/8/2012 Subject: Re: Coord buffer overflow again Hi Tom, "Have you tried re-connecting the GamPad and pushing the buttons to see if you can force errors?" - no nothing happen, accelerations are normal. I am happy that it is finally working and I finished this big reliefs. I can test tomorrow same gcode with gamepad connected to pc and check if it will happen again. Best Regards Albert P³atek --- In DynoMotion@yahoogroups.com, Tom Kerekes <tk@...> wrote: > > Hi Albert, >   > Wow that is strange.  I just double checked and V4.28 is in fact testing the Gampad enable option.  So I have no explaination how the Gampad could make a difference.  Possibly it somehow corrupts the USB communication - but that should generate an error.  Maybe it causes Coordinated Motion Segments to just get completely lost (skipped). >   > That is something I couldn't understand about the problem, there should be no randomness.  Everything we are doing should be completely deterministic.  The same GCode and the same Settings should generate the exact same motion every time. >   > Have you tried re-connecting the GamPad and pushing the buttons to see if you can force errors? > > Regards > TK >   >   > >   > Hi Tom, > > I think that in 99% I found bug, it was gamepad connected into usb. > Even if it was disabled in KMmotionCNC, it was causing randomly this > crazy jerk of acceleration. Today I tested this big gcode program 3 times without gamepad and everything is fine! :) > > Best Regards > Albert P³atek > > --- In DynoMotion@yahoogroups.com, Albert P³atek <albertplatek86@> wrote: > > > > Hi Tom, > > "So are you saying that it did not report any errors until after 10 > > minutes then printed those?" > > - It was printing accel peak around 60in/sec2 sometimes from start and > > after 10 min this big error. Second job execute was ok by about 20min then > > again error. > > Smoothing was Tau= 0.008. > > > > > > I just made new test on other pc with file "test_tolerance_0.005.ngc" . And > > I got *1166in/sec2* acceleration peak... and it stalled X motor. > > > > It is same program as before but with toolpath tolerance of 0.005. I > > thought that maybe this will help. But it didnt. (old file Relief14.ngc had > > toolpath tolerance of 0.01) > > Small programs works even with extream acceleration of 60, but this program > > and other bigger programs fail even with acceleration of 20. > > > > Here all my settings: > > > > *Trajectory planner* > > > > Break Angle 30 degrees > > Look Ahead 3 sec > > Collinear Tolerance 0.0005 in > > Corner Tolerance 0.01 in > > Facet Angle 1.5 > > > > *Axis Parameters* > > > > X Resolution 254000 Vel 5 Accel 40 > > Y Resolution 127000 Vel 4 Accel 10 > > Z Resolution 254000 Vel 5 Accel 40 > > A Resolution 7055.5555 Vel 100 Accel 15 (I dont use this axis in programs > > at moment) > > > > Smoothing Tau = 0.01 > > > > Gamepad disabled > > Lathe disabled > > I dont use MPG and it is not connected. > > > > Gcode program (test_tolerance_0.005.ngc) started from first line, I was > > only capturing Acceleration > 58. > > > > Wt, lut 07, 2012, 12:42:31 KMotion Program Started > > TAU = 0.010000 > > ACCEL= 58.677936 > > ACCEL= 58.927802 > > ACCEL= 58.960033 > > ACCEL= 60.769580 > > ACCEL= 58.512884 > > ACCEL= 58.924281 > > ACCEL= 59.024541 > > ACCEL= 58.133892 > > ACCEL= 61.956729 > > ACCEL= 59.006616 > > ACCEL= 60.105317 > > ACCEL= 58.828274 > > ACCEL= 60.622193 > > ACCEL= 58.060298 > > ACCEL= 60.837834 > > ACCEL= 58.185082 > > ACCEL= 58.925759 > > ACCEL= 59.444038 > > ACCEL= 61.017190 > > ACCEL= 59.082489 > > ACCEL= 60.176545 > > ACCEL= 58.677989 > > ACCEL= 60.129486 > > ACCEL= 59.890981 > > ACCEL= 63.703736 > > ACCEL= 58.300426 > > ACCEL= 58.390017 > > ACCEL= 64.994889 > > ACCEL= 61.344673 > > ACCEL= 61.264344 > > ACCEL= 60.206659 > > ACCEL= 61.110118 > > ACCEL= 58.861575 > > ACCEL= 62.844393 > > ACCEL= 58.712601 > > ACCEL= 62.743716 > > ACCEL= 59.412413 > > ACCEL= 59.090724 > > ACCEL= 59.607079 > > ACCEL= 58.240991 > > ACCEL= 58.200542 > > ACCEL= 62.400201 > > ACCEL= 64.401300 > > ACCEL= 60.829834 > > ACCEL= 60.765056 > > ACCEL= 58.584326 > > ACCEL= 59.368296 > > ACCEL= 58.845166 > > ACCEL= 63.692201 > > ACCEL= 59.398388 > > ACCEL= 58.492754 > > ACCEL= 58.201424 > > ACCEL= 59.713704 > > ACCEL= 59.488310 > > ACCEL= 58.663588 > > ACCEL= 62.887233 > > ACCEL= 58.256548 > > ACCEL= 58.483484 > > ACCEL= 58.042246 > > ACCEL= 58.518015 > > ACCEL= 61.312055 > > ACCEL= 60.224222 > > ACCEL= 60.041371 > > ACCEL= 61.019005 > > ACCEL= 59.857207 > > ACCEL= 60.667182 > > ACCEL= 62.517183 > > ACCEL= 62.978943 > > ACCEL= 60.672158 > > ACCEL= 62.764637 > > ACCEL= 60.755186 > > ACCEL= 58.177374 > > ACCEL= 58.144752 > > ACCEL= 62.349561 > > ACCEL= 59.094084 > > ACCEL= 59.707611 > > ACCEL= 60.042771 > > ACCEL= 59.493590 > > ACCEL= 58.381740 > > ACCEL= 58.552432 > > ACCEL= 60.385809 > > ACCEL= 61.290063 > > ACCEL= 63.347741 > > ACCEL= 59.724964 > > ACCEL= 66.035414 > > ACCEL= 58.827807 > > ACCEL= 60.704297 > > ACCEL= 60.619439 > > ACCEL= 60.255062 > > ACCEL= 59.246636 > > ACCEL= 58.966244 > > ACCEL= 62.119289 > > ACCEL= 61.110753 > > ACCEL= 64.018676 > > ACCEL= 63.911028 > > ACCEL= 60.811810 > > ACCEL= 59.177635 > > ACCEL= 61.196853 > > ACCEL= 58.680083 > > ACCEL= 59.527456 > > ACCEL= 58.087839 > > ACCEL= 59.119465 > > ACCEL= 62.853732 > > ACCEL= 59.093458 > > ACCEL= 59.161929 > > ACCEL= 61.429357 > > ACCEL= 58.803188 > > ACCEL= 59.175222 > > ACCEL= 60.219244 > > ACCEL= 58.596575 > > ACCEL= 58.764846 > > ACCEL= 58.426363 > > ACCEL= 60.086656 > > ACCEL= 59.610107 > > ACCEL= 59.229815 > > ACCEL= 58.044244 > > ACCEL= 63.695621 > > ACCEL= 58.131310 > > ACCEL= 59.107058 > > ACCEL= 58.927665 > > ACCEL= 61.165485 > > ACCEL= 58.315186 > > ACCEL= 60.871648 > > ACCEL= 60.176458 > > ACCEL= 59.214217 > > ACCEL= 59.208457 > > ACCEL= 62.621824 > > ACCEL= 58.398335 > > ACCEL= 61.152436 > > ACCEL= 58.604786 > > ACCEL= 58.729217 > > ACCEL= 59.923425 > > ACCEL= 59.677849 > > ACCEL= 58.123086 > > ACCEL= 60.433701 > > ACCEL= 58.291893 > > ACCEL= 60.252449 > > ACCEL= 60.417103 > > ACCEL= 60.433561 > > ACCEL= 61.127250 > > ACCEL= 61.402805 > > ACCEL= 58.202147 > > ACCEL= 62.299667 > > ACCEL= 62.973329 > > ACCEL= 61.678133 > > ACCEL= 58.407879 > > ACCEL= 59.317560 > > ACCEL= 58.968064 > > ACCEL= 58.048711 > > ACCEL= 58.288685 > > ACCEL= 58.241909 > > ACCEL= 59.930745 > > ACCEL= 59.450933 > > ACCEL= 59.439070 > > ACCEL= 58.143918 > > ACCEL= 58.259981 > > ACCEL= 58.482531 > > ACCEL= 58.101452 > > ACCEL= 60.411654 > > ACCEL= 60.631439 > > ACCEL= 62.469302 > > ACCEL= 60.847604 > > ACCEL= 58.350368 > > *ACCEL= 1166.694794 > > ACCEL= 450.118253* > > ACCEL= 186.931892 > > ACCEL= 91.714558 > > ACCEL= 60.190257 > > ACCEL= 60.530275 > > ACCEL= 59.028122 > > > > Here I stopped program line was N463261. > > Smoothing TAU 0.01 > > > > In attechement INIT.c config program. > > GCode program is in Yahoo Files folder "Albert CNC > > files " (test_tolerance_0.005.part01.rar > > and test_tolerance_0.005.part02.rar) > > > > Best Regards > > Albert Platek > > > > > > 2012/2/7 Tom Kerekes <tk@> > > > > > ** > > > > > > [Attachment(s)<https://mail.google.com/mail/html/compose/static_files/blank_quirks.html#1355587bcb0eb6ec_TopText>from Tom Kerekes included below] > > > > > > Hi Albert, > > > > > > Your program seems reasonable but the numbers are crazy. > > > > > > So are you saying that it did not report any errors until after 10 minutes > > > then printed those? > > > > > > I made a similar program and ran it over night and didn't see such a > > > problem. See attached. It displays the velocity and acceleration into > > > unused axes 6 and 7 Destinations on the Axis Screen. It also captures the > > > Max Velocity and Acceleration and displays those as 6 and 7 Positions. > > > > > > To be honest I see some tiny glitches and anomolies probably due to > > > numerical rounding that I would like to better understand, but with the > > > smoothing filter of 8ms it attenuates them way down. The biggest > > > acceleration spike I see in the entire multi hour job is 90 in/sec2. > > > > > > Did you have the smoothing turned on? > > > > > > Is your Resolution 254,000 steps per inch? > > > > > > You don't have an MPG connected do you? > > > > > > I will try your program to see if I get similar results. > > > > > > Thanks > > > TK > > > > > > *From:* albertplatek <albertplatek86@> > > > *To:* DynoMotion@yahoogroups.com > > > *Sent:* Monday, February 6, 2012 8:24 AM > > > *Subject:* [DynoMotion] Re: Coord buffer overflow again > > > > > > > > > Hi Tom, > > > > > > Sorry Tom but still this happens randomly :/ I was just lucky that after > > > firmware flash 1 job was ok. > > > I use ver 4.28. > > > > > > I dont know how to determine line numner of this jerk, but you can see > > > huge acceleration! X motor stalled. > > > It was runinng about 10min from program start. > > > Acceleration was set 32in/sec2 > > > Break Angle 30 > > > Tau 0.008 > > > Collinear Tolerance 0.0005 > > > Corener Tolerance 0.01 > > > Facet angle 0.5 > > > Vel 4in/sec > > > > > > Here is my simple C program to check acceleration please check if it is > > > correct: > > > > > > #include "KMotionDef.h" > > > #define channel ch0 > > > #define time 0.00018 > > > main() > > > { > > > double X1,X2,speed1,speed2,accel; > > > int k; > > > > > > while(1) > > > { > > > for (k=0; k<1; k++) WaitNextTimeSlice(); > > > X1 = channel->Dest; > > > for (k=0; k<1; k++) WaitNextTimeSlice(); > > > X2 = channel->Dest; > > > speed1=(X2-X1)/(time); > > > > > > for (k=0; k<1; k++) WaitNextTimeSlice(); > > > X1 = channel->Dest; > > > for (k=0; k<1; k++) WaitNextTimeSlice(); > > > X2 = channel->Dest; > > > speed2=(X2-X1)/(time); > > > > > > accel=(speed1-speed2)/(time); > > > accel=(accel/254000)/2; > > > if(accel>45) > > > { > > > printf("AccelError= %f\n",accel); > > > } > > > } > > > } > > > > > > Here log: > > > > > > AccelError= 45.009219 > > > AccelError= 45.229875 > > > AccelError= 46.294106 > > > AccelError= 45.055529 > > > AccelError= 45.747547 > > > AccelError= 45.070729 > > > AccelError= 45.360541 > > > AccelError= 45.765590 > > > AccelError= 46.302076 > > > AccelError= 46.775562 > > > AccelError= 47.051896 > > > AccelError= 45.225101 > > > AccelError= 45.730674 > > > AccelError= 46.758273 > > > AccelError= 46.838768 > > > AccelError= 45.698572 > > > AccelError= 46.918682 > > > AccelError= 45.204219 > > > AccelError= 45.900150 > > > AccelError= 47.023519 > > > AccelError= 47.551550 > > > AccelError= 48.022541 > > > AccelError= 46.395295 > > > AccelError= 81526.314840 > > > AccelError= 72853.056587 > > > AccelError= 65101.872298 > > > AccelError= 58175.992747 > > > AccelError= 51986.611437 > > > AccelError= 46456.046360 > > > AccelError= 41513.718492 > > > AccelError= 37097.300673 > > > AccelError= 33150.797142 > > > AccelError= 29622.770379 > > > AccelError= 26468.350534 > > > AccelError= 23652.797849 > > > AccelError= 21136.323622 > > > AccelError= 18888.213633 > > > AccelError= 16879.065077 > > > AccelError= 15083.474245 > > > AccelError= 13479.186003 > > > AccelError= 12045.566722 > > > AccelError= 10764.204608 > > > AccelError= 9619.483512 > > > AccelError= 8596.141212 > > > AccelError= 7682.015002 > > > AccelError= 6866.662006 > > > AccelError= 6139.961139 > > > AccelError= 5490.446488 > > > AccelError= 4910.042286 > > > AccelError= 4391.574405 > > > AccelError= 3928.114510 > > > AccelError= 3514.000485 > > > AccelError= 3144.002655 > > > AccelError= 2813.316136 > > > AccelError= 2517.500643 > > > AccelError= 2253.490792 > > > AccelError= 2017.432428 > > > AccelError= 1806.266152 > > > AccelError= 1617.706032 > > > AccelError= 1448.982225 > > > AccelError= 1298.398178 > > > AccelError= 1163.621865 > > > AccelError= 1043.472200 > > > AccelError= 935.821407 > > > AccelError= 839.766485 > > > AccelError= 753.863074 > > > AccelError= 677.101783 > > > AccelError= 608.587862 > > > AccelError= 547.209083 > > > AccelError= 492.589836 > > > AccelError= 443.554631 > > > AccelError= 399.735924 > > > AccelError= 360.639632 > > > AccelError= 325.643422 > > > AccelError= 294.370902 > > > AccelError= 266.425861 > > > AccelError= 241.478288 > > > AccelError= 219.191672 > > > AccelError= 199.246085 > > > AccelError= 181.422761 > > > AccelError= 165.532090 > > > AccelError= 151.295897 > > > AccelError= 138.574472 > > > AccelError= 127.206640 > > > AccelError= 117.084135 > > > AccelError= 108.003075 > > > AccelError= 99.888264 > > > AccelError= 92.598183 > > > AccelError= 86.122270 > > > AccelError= 76.770777 > > > AccelError= 65.225363 > > > AccelError= 54.908134 > > > AccelError= 45.688617 > > > AccelError= 45.070435 > > > AccelError= 45.411744 > > > AccelError= 45.770036 > > > AccelError= 45.871485 > > > AccelError= 45.239885 > > > AccelError= 45.631836 > > > AccelError= 45.001237 > > > AccelError= 45.373033 > > > AccelError= 45.577907 > > > AccelError= 46.322608 > > > AccelError= 46.958820 > > > AccelError= 46.038556 > > > AccelError= 45.018088 > > > > > > If you test it on your PC and will be ok, then maybe something is wrong > > > with my laptop. Then I will test on other PC. > > > > > > Best Regards > > > Albert P³atek > > > > > > --- In DynoMotion@yahoogroups.com, "albertplatek" <albertplatek86@> > > > wrote: > > > > > > > > Hi Tom, > > > > > > > > Problem solved (probably), weird but after re-flash firmware 4.28 > > > > everything is working ok, no axis stall. > > > > > > > > btw. Experimental version KMotion429x.exe is not working, when I start > > > > execute gcode I get message "G Code Error GCode Aborted", this error > > > > occurs on any gcode program even on only few G01 lines. > > > > > > > > Tom if you have C program that watch for too high of acceleration > > > > please post it. If again in future something strange will > > > > happen I will try find out the problem. > > > > I am still not sure what was the problem in my system that axis > > > > were stalling. > > > > > > > > Best Regards > > > > Albert P³atek > > > > > > > > > > > > > > > > --- In DynoMotion@yahoogroups.com, Albert P³atek <albertplatek86@> > > > wrote: > > > > > > > > > > Hi Tom, > > > > > > > > > > Thank you for help. > > > > > Mostly stall X axis, and sometimes Z. > > > > > Y axis seem to be ok. > > > > > I uploaded file as an attachment (zip). > > > > > > > > > > Best Regards > > > > > Albert Platek > > > > > > > > > > 2012/2/5 Tom Kerekes <tk@> > > > > > > > > > > > ** > > > > > > > > > > > > > > > > > > Hi Albert, > > > > > > > > > > > > Thanks for the very complete problem description, I can't tell you > > > how > > > > > > much I appreciate that, but I can't think of anything that would > > > cause > > > > > > this. Anything with large files I might expect some processing issue > > > but > > > > > > that would result in a buffer underflow and an error message not > > > incorrect > > > > > > motion that would result in a stall. That delay I described in your > > > email > > > > > > only happens when "Execute" is pushed so I don't think it is related. > > > > > > > > > > > > There is an experimental version KMotion429x.exe on the download site > > > > > > that has this delay eliminated and also faster USB throughput, but I > > > don't > > > > > > think either of those are related to this problem. > > > > > > > > > > > > Can you post the file somewhere? I'm thinking to write a C program > > > that > > > > > > would run in KFLOP and watch for too high of acceleration or > > > velocity by > > > > > > monitoring how ch0->Dest changes. If it detects something bad that > > > could > > > > > > be used as a trigger to help us track it down. > > > > > > > > > > > > Which axes loose position? > > > > > > > > > > > > Regards > > > > > > TK > > > > > > > > > > > > > > > > > > *From:* albertplatek <albertplatek86@> > > > > > > *To:* DynoMotion@yahoogroups.com > > > > > > *Sent:* Sunday, February 5, 2012 5:27 AM > > > > > > > > > > > > *Subject:* [DynoMotion] Re: Coord buffer overflow again > > > > > > > > > > > > > > > > > > Hi Tom, > > > > > > > > > > > > My settings: > > > > > > Break angle 30 deg > > > > > > Look ahead 3 sec > > > > > > Collinear tolerance 0.001 in > > > > > > Corner tolerance 0.003 in > > > > > > Facet angle 1.5 deg > > > > > > Tau = 0.008 sec > > > > > > XYZ Accelarations 40 in/sec2 > > > > > > XYZ Velocity 4.5 in/sec > > > > > > In gcode file I use max Vel 270 in/min (4.5 in/sec) > > > > > > > > > > > > I have strange problem, everything was ok until I started machining > > > > > > bigger reliefs. > > > > > > At half of the job my stepper motors lost position > > > > > > probably becouse of high jerk of acceleration. > > > > > > Firstly I thought that maybe axis accelerations for my system are to > > > > > > high(40in/sec2). So I changed it to 20in/sec2 for test. > > > > > > But didnt help, still stepper motors stall at half of the gcode file. > > > > > > > > > > > > This file contain 548389 lines, my steppers stall between 226000 - > > > 274000 > > > > > > line but not always in same place. > > > > > > > > > > > > I tried to find line where the problem occurs, but when I found line > > > with > > > > > > problem and tried to execute 500 lines before, it was ok. Motors > > > didnt > > > > > > stall. > > > > > > I found that motors stall when KMotionCNC start execute at least > > > > > > 100 000 lines before line number 274000. > > > > > > > > > > > > The problem disappear when I set break angle 5deg. > > > > > > > > > > > > "This diagnostic pointed out a side problem I hadn't noticed before. > > > There > > > > > > is an annoying ~30 second delay when it runs the second time. The > > > > > > Interpreter keeps a record of all its state changes which allow > > > Halts to go > > > > > > back to any previous state. In this test of 100x500 subroutine calls > > > > > > results in about 1 million state changes. Ironically creating the > > > data, > > > > > > making use of the data all happens in the blink of an eye, but > > > discarding > > > > > > the data (CList.RemoveAll) takes 30 seconds! We will find a better > > > way in > > > > > > the next version. For now just don't be surprised if there is an > > > annoying > > > > > > delay on a second run of a big job." > > > > > > > > > > > > Can this "delay" make that my stepper motors stall? > > > > > > > > > > > > Best Regards > > > > > > Albert P³atek > > > > > > > > > > > > --- In DynoMotion@yahoogroups.com, Albert <albertplatek86@> wrote: > > > > > > > > > > > > > > Hi Tom, > > > > > > > > > > > > > > Thank you. > > > > > > > Btw. I am trying to find out what changes make > > > > > > > Fidelizer in Windows. If I will find something > > > > > > > important I will let you know. > > > > > > > Best regards > > > > > > > Albert Platek > > > > > > > > > > > > > > > > > > > > > Dnia 2012-01-18 o godz. 23:47 Tom Kerekes <tk@> napisaÃÆ'â€¦Ã¢â‚¬à ¡ > > > > > > > (a): > > > > > > > > > > > > > > > Hi Albert, > > > > > > > > > > > > > > > > Just wanted to mention that I now do notice an improvement with > > > > > > > > Fidelizer. I don't understand why I didn't before. I might have > > > > > > > > made a mistake or something changed. > > > > > > > > > > > > > > > > I sent an email to FTDI (USB chip we use) regarding it and the > > > USB > > > > > > > > 3.0 port issue to see if they have any answers. I figure it is > > > > > > > > worth the cost of an email :} > > > > > > > > > > > > > > > > Still working on some optimization... > > > > > > > > > > > > > > > > Regards > > > > > > > > TK > > > > > > > > > > > > > > > > > > > > > > > > From: albertplatek <albertplatek86@> > > > > > > > > To: DynoMotion@yahoogroups.com > > > > > > > > Sent: Monday, January 16, 2012 12:01 PM > > > > > > > > Subject: [DynoMotion] Re: Coord buffer overflow again > > > > > > > > > > > > > > > > > > > > > > > > Hi Tom, > > > > > > > > Thank you for making changes in new version to increase the > > > segment > > > > > > > > rate. > > > > > > > > > > > > > > > > Btw: > > > > > > > > I installed on second pc Windows 7 (Core2Duo cpu), and aslo > > > tested > > > > > > > > with fidelizer. Strange but didn't make difference, as in your > > > > > > > > computer. > > > > > > > > > > > > > > > > I cant get why in this particular case (pc with WIN7 i7 cpu) it > > > helps. > > > > > > > > Maybe it depends on type of host controller... > > > > > > > > > > > > > > > > Pc with WIN7 64bit (Core2Duo) has > > > > > > > > SiS 7001 PCI to USB Open Host Controller > > > > > > > > > > > > > > > > Pc with WIN7 64bit (i7) has > > > > > > > > Intel(R) 6 Series/C200 Series Chipset Family USB Enhanced Host > > > > > > > > Controller > > > > > > > > > > > > > > > > This computer has aslo another usb ports: Renesas Electronics USB > > > > > > > > 3.0 Host Controller > > > > > > > > which works very bad with KFlop. And on this port fidelizer make > > > no > > > > > > > > difference. > > > > > > > > > > > > > > > > Best Regards > > > > > > > > Albert Platek > > > > > > > > > > > > > > > > --- In DynoMotion@yahoogroups.com, Tom Kerekes <tk@> wrote: > > > > > > > > > > > > > > > > > > Hi Albert, > > > > > > > > > ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > > > Thanks for that interesting information. > > > > > > > > > ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > > > I tried the Fidelizer on my i7 W7 32 machine and it didn't > > > seem to > > > > > > > > make any difference. > > > > > > > > > ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > > > I'm trying some things to increase the segment rate by sending > > > > > > > > blocks of segments and optimizing how they are handled in > > > KFLOP.ÃÆ'Æ'Ã¢â‚¬à ¡ I > > > > > > > > f this bears fruit I'll let you know. > > > > > > > > > ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > > > This diagnostic pointed out a side problem I hadn't noticed > > > before > > > > > > > > .ÃÆ'Æ'Ã¢â‚¬à ¡ There is an annoying ~30 second delay when it runs the > > > second ti > > > > > > > > me.ÃÆ'Æ'Ã¢â‚¬à ¡ The Interpreter keeps a record of all its state changes > > > which > > > > > > > > allow Halts to go back to any previous state.ÃÆ'Æ'Ã¢â‚¬à ¡ In this test > > > of 100x > > > > > > > > 500 subroutine calls results in about 1 million state > > > changes.ÃÆ'Æ'Ã¢â‚¬à ¡ Iro > > > > > > > > nically creating the data, making use of the data all happens in > > > the > > > > > > > > blink of an eye, but discarding the data (CList.RemoveAll) takes > > > 30 > > > > > > > > seconds!ÃÆ'Æ'Ã¢â‚¬à ¡ We will find a better way in the next > > > version.ÃÆ'Æ'Ã¢â‚¬à ¡ For now > > > > > > > > just don't be surprisedÃÆ'Æ'Ã¢â‚¬à ¡ if there is an annoying delay on a > > > second > > > > > > > > run of a big job. > > > > > > > > > ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > > > Regards > > > > > > > > > TK > > > > > > > > > ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > > > > > > > > > > > > > > > > > > > > > ________________________________ > > > > > > > > > From: albertplatek <albertplatek86@> > > > > > > > > > To: DynoMotion@yahoogroups.com > > > > > > > > > Sent: Sunday, January 15, 2012 8:37 AM > > > > > > > > > Subject: [DynoMotion] Re: Coord buffer overflow again > > > > > > > > > > > > > > > > > > > > > > > > > > > ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > > > Hi Tom, > > > > > > > > > First thank you Tom for the test file. > > > > > > > > > I made few tests and I got interesting results :) > > > > > > > > > > > > > > > > > > PC No1: > > > > > > > > > > > > > > > > > > Intel Pentium 4 2.6Ghz WinXP 32bit port USB2.0 > > > > > > > > > Max segments: 110ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > > > Time: 31s > > > > > > > > > 100 circles x 110 segments/circle / 31 seconds =354 segments > > > per > > > > > > > > second. > > > > > > > > > > > > > > > > > > PC No2: > > > > > > > > > > > > > > > > > > Intel Core2Duo 2.0 Ghz WinXP 32bit port USB2.0 > > > > > > > > > Max segments: 140ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > > > Time: 31s > > > > > > > > > 100 circles x 140 segments/circle / 31 seconds = 451 segments > > > per > > > > > > > > second. > > > > > > > > > > > > > > > > > > PC No3: > > > > > > > > > > > > > > > > > > i7 2670QM Win7 64bit port USB3.0 > > > > > > > > > Max segments:30ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > > > Time:31s > > > > > > > > > 100 circles x 60 segments/circle / 31 seconds = 193 segments > > > per > > > > > > > > second. > > > > > > > > > (thats why I had problem here, usb 3.0 fail) > > > > > > > > > > > > > > > > > > (Luckily I had 1 USB 2.0 port) > > > > > > > > > > > > > > > > > > i7 2670QM Win7 64bit port USB2.0 > > > > > > > > > Max segments:170 > > > > > > > > > Time:31s > > > > > > > > > 100 circles x 170 segments/circle / 31 seconds = 548 segments > > > per > > > > > > > > second. > > > > > > > > > > > > > > > > > > i7 2670QM Win7 64bit port USB2.0 (with 1 core affinity for > > > > > > > > KmotionCNC and KmotionServer) > > > > > > > > > Max segments:200 > > > > > > > > > Time:31s > > > > > > > > > 100 circles x 200 segments/circle / 31 seconds =645 segments > > > per > > > > > > > > second. > > > > > > > > > > > > > > > > > > I found on net some interesting software called Fidelizer, its > > > > > > > > software to optimize win7/vista for audio workstation, but it > > > aslo > > > > > > > > tweak somehow usb throughput. > > > > > > > > > More detail here: /www.windowsxlive.net/fidelizer > > > > > > > > > > > > > > > > > > i7 2670QM Win7 64bit port USB2.0 (Fidelizer) > > > > > > > > > Max segments:410 > > > > > > > > > Time:31s > > > > > > > > > 100 circles x 410 segments/circle / 31 seconds = 1320 segments > > > per > > > > > > > > second. (wow! tested few times with same result) > > > > > > > > > > > > > > > > > > Be aware of new USB 3.0 ports they may not work with good > > > > > > > > performance with KFlop, maybe becouse of bug in drivers. Windows > > > 7 > > > > > > > > dont see them without additional drivers. > > > > > > > > > Fidelizer software only improved performance on pc with ÃÆ'Æ'Ã¢â‚¬à ¡ > > > Win7. > > > > > > > > > On WinXP seem to not work. I will soon try install on "PC No2" > > > > > > > > Windows 7 and then try fidelizer also. > > > > > > > > > > > > > > > > > > Best regards > > > > > > > > > Albert PÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚³atek > > > > > > > > > --- In DynoMotion@yahoogroups.com, Tom Kerekes <tk@> wrote: > > > > > > > > > > > > > > > > > > > > Hi Albert, > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > > > > Your application/system is quite demanding :} > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > > > > That portion of GCode you sent is 100,000 lines of all G1 > > > > > > > > movements with typical tiny lengths of only > > > 50~150um.ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ The > > > > > > > > collinear tolerance of 0.001 (25um) will cause many > > > blocksÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ to > > > > > > > > be combined together so the real required vector rate will be > > > more d > > > > > > > > ependent on the path curvature rather than the original > > > GCode.ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > > > > On my system (i7 W7 32-bit) it runs ok with a facet angle 0f > > > 0.4 > > > > > > > > degrees butÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ can't keep up at 0.3 degrees. > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > > > > I did some calculations for a system like yours with 40 > > > in/sec2 > > > > > > > > acceleration.ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ Here is what I find: > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > > > > That acceleration allows going around a 0.1 inch radius > > > circle > > > > > > > > at 120ipm (which is about 3 circles per second) > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > > > > With a facet angle setting of 1.5 degrees that requires 240 > > > > > > > > segments per circle which would require 720 segments per second > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > > > > I created a diagnostic test GCode file that programmatically > > > > > > > > generates an N sided Polygon 100 timesÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ with > > > radius 0.1 inches > > > > > > > > and allows you to specify the number of polygon sides (it also > > > gradu > > > > > > > > ally shifts the "circle" to make the plot more > > > interesting).ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > > (see attachment). ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ When using the diagnostic > > > make sure to set > > > > > > > > the Collinear tolerance to zero and the Corner Rounding to zero > > > so t > > > > > > > > he original segments are used.ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ By changing N > > > (number of sides > > > > > > > > we can determine the segment download rate). > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > > > > For example on my system it works with N = 200 and fails > > > with N > > > > > > > > = 240.ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ The total time to do all 100 circles > > > takes about 30 > > > > > > > > seconds.ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ So: > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > > > > 100 circles x 200 segments/circle / 30 seconds = 666 segments > > > > > > > > per second. > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > > > > In simulation mode the Interpreter+Graphics finishes the job > > > in > > > > > > > > ~ 5 seconds so it isn't a significant factor > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > > > > With the GCode Window closed it simulates in ~ 4 seconds so > > > the > > > > > > > > Graphical Display is not aÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ significant factor. > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > > > > You might run the tests on your two systems to see if you get > > > > > > > > drastically different results. > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > > > > But to try to answer your original question of why one CPU > > > would > > > > > > > > be significantly different I don't know.ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ Google > > > finds some > > > > > > > > benchmarking tools but they areÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ mainly for High > > > Speed USB > > > > > > > > devices like flash drives.ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ Some sites state if > > > an older USB > > > > > > > > 1.1 device is attached it slows everything down.ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > > > > Regards > > > > > > > > > > TKÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > > > > > > > > > > > > > > From: albertplatek albertplatek86@ > > > > > > > > > > To: DynoMotion@yahoogroups.com > > > > > > > > > > Sent: Saturday, January 14, 2012 11:02 AM > > > > > > > > > > Subject: [DynoMotion] Re: Coord buffer overflow again > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > > > > Hi Tom, > > > > > > > > > > How to determine USB throughput on pc? or what pc/system is > > > best > > > > > > > > for KFlop? > > > > > > > > > > > > > > > > > > > > I have still sometimes cord buffer overflow on second pc > > > even I > > > > > > > > > > set lower accelaration 25 and bigger facet angle 2.5 it > > > sometimes > > > > > > > > > > overflow. Its very strange pc has windows xp clean > > > installation, > > > > > > > > > > Integrated Intel Graphics. > > > > > > > > > > On first pc works perfect even with high acceleration 40, > > > facet > > > > > > > > angle 2.0 (also windows xp, but ati graphics) > > > > > > > > > > > > > > > > > > > > Break angle 30 > > > > > > > > > > Look ahead 3 > > > > > > > > > > Collinear tolerance 0.001 > > > > > > > > > > Corner tolerance 0.005 > > > > > > > > > > Vel 5 > > > > > > > > > > > > > > > > > > > > Best Regards > > > > > > > > > > Albert Platek > > > > > > > > > > > > > > > > > > > > --- In DynoMotion@yahoogroups.com, Albert > > > PÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚³atek > > > > > > > > albertplatek86@ wrote: > > > > > > > > > > > > > > > > > > > > > > Hi Tom, > > > > > > > > > > > GCode file is attached with this message. > > > > > > > > > > > Yes feedrate is 300ipm. > > > > > > > > > > > What do you mean by "path generation slows down"? - i mean > > > > > > > > that when I am > > > > > > > > > > > changing view perspective in GViewer > > > > > > > > > > > the "simulated tool" catching up end of buffered path. So > > > > > > > > thats why I > > > > > > > > > > > thought that maybe gviewer impact on performance somehow. > > > > > > > > > > > > > > > > > > > > > > Best Regards > > > > > > > > > > > Albert > > > PÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'‚¦ÃÆ'Æ'Æ'ÃÆ'‚¢ÃÆ'Æ'¢ÃÆ'¢â‚¬Å¡ÃÆ'‚¬ÃÆ'Æ'…ÃÆ'‚¡atek > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > 2012/1/13 Tom Kerekes tk@ > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > Hi Albert, > > > > > > > > > > > > > > > > > > > > > > > > Yes please send the file. Is the feedrate set high > > > 300ipm? > > > > > > > > > > > > > > > > > > > > > > > > What do you mean by "path generation slows down"? > > > > > > > > > > > > > > > > > > > > > > > > Thanks > > > > > > > > > > > > TK > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > *From:* albertplatek albertplatek86@ > > > > > > > > > > > > *To:* DynoMotion@yahoogroups.com > > > > > > > > > > > > *Sent:* Thursday, January 12, 2012 3:35 PM > > > > > > > > > > > > *Subject:* [DynoMotion] Re: Coord buffer overflow again > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > Hi Tom, > > > > > > > > > > > > > > > > > > > > > > > > My settings: > > > > > > > > > > > > > > > > > > > > > > > > Break angle 30 > > > > > > > > > > > > Look ahead 3 > > > > > > > > > > > > Collinear tolerance 0.001 > > > > > > > > > > > > Corner tolerance 0.005 > > > > > > > > > > > > Facet angle 2.0 (if I go to 1.5, i get cord buffer > > > overflow > > > > > > > > almost from > > > > > > > > > > > > program start) > > > > > > > > > > > > > > > > > > > > > > > > I use Tau = 0.01; (with facet angle 2.0 machine movements > > > > > > > > are fairly > > > > > > > > > > > > smooth) > > > > > > > > > > > > > > > > > > > > > > > > All 3 axis has same Vel/Accel parameters: > > > > > > > > > > > > Vel 5.0 > > > > > > > > > > > > Accel 40 > > > > > > > > > > > > > > > > > > > > > > > > Toolpath is parallel along X-axis (contain a lot of small > > > > > > > > arcs, I will > > > > > > > > > > > > send file tomorrow) > > > > > > > > > > > > > > > > > > > > > > > > With this settings my first router with old laptop works > > > ok > > > > > > > > (Dell, core 2 > > > > > > > > > > > > duo, ati graphics card), > > > > > > > > > > > > but on second router (same settings) with new laptop dont > > > > > > > > work, i often > > > > > > > > > > > > get buffer overflows. (laptop Dell, i7 cpu, Win7 64bit, > > > > > > > > Nvidia Gt555) > > > > > > > > > > > > > > > > > > > > > > > > (Btw. when GViewer is open and I move view by mouse path > > > > > > > > generation slows > > > > > > > > > > > > down, is it normal?) > > > > > > > > > > > > > > > > > > > > > > > > Best Regards > > > > > > > > > > > > Albert PÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚³atek > > > > > > > > > > > > > > > > > > > > > > > > --- In DynoMotion@yahoogroups.com, Tom Kerekes <tk@> > > > wrote: > > > > > > > > > > > > > > > > > > > > > > > > > > Hi Albert, > > > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'†'ÃÆ'Æ'Æ'ÃÆ'‚¢ÃÆ'Æ'¢ÃÆ'¢â‚¬Å¡ÃÆ'‚¬ÃÆ'Æ'…ÃÆ'‚¡ > > > > > > > > > > > > > Could you send me the Gcode and all your > > > settings?ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'†'ÃÆ'Æ'Æ' > > > > > > > > ÃÆ'‚¢ÃÆ'Æ'¢ÃÆ'¢â‚¬Å¡ÃÆ'‚¬ÃÆ'Æ'…ÃÆ'‚¡ I don't understand > > > > > > > > > > > > the collinear tolerance of > > > 1?ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'†'ÃÆ'Æ'Æ'ÃÆ'‚¢ÃÆ'Æ'¢ÃÆ'¢â‚¬Å¡ÃÆ'‚¬ÃÆ'Æ'…ÃÆ'‚¡ That > > > > > > > > seems huge.ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'†'ÃÆ'Æ'Æ'ÃÆ'‚¢ÃÆ'Æ'¢ÃÆ'¢â‚¬Å¡ÃÆ'‚¬ÃÆ'Æ'…ÃÆ'‚¡ Did you > > > mean 0.001? > > > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'†'ÃÆ'Æ'Æ'ÃÆ'‚¢ÃÆ'Æ'¢ÃÆ'¢â‚¬Å¡ÃÆ'‚¬ÃÆ'Æ'…ÃÆ'‚¡ > > > > > > > > > > > > > A corner rounding facet angle too small might be > > > > > > > > generating too many > > > > > > > > > > > > small > > > segments.ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'†'ÃÆ'Æ'Æ'ÃÆ'‚¢ÃÆ'Æ'¢ÃÆ'¢â‚¬Å¡ÃÆ'‚¬ÃÆ'Æ'…ÃÆ'‚¡ Are you using the > > > > > > > > Coordinated Motion Low Pass Filter > > > > > > > > > > > > Smoothing > > > option?ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'†'ÃÆ'Æ'Æ'ÃÆ'‚¢ÃÆ'Æ'¢ÃÆ'¢â‚¬Å¡ÃÆ'‚¬ÃÆ'Æ'…ÃÆ'‚¡ That might help make > > > > > > > > as smooth or smoother motion with > > > > > > > > > > > > larger facets. > > > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'†'ÃÆ'Æ'Æ'ÃÆ'‚¢ÃÆ'Æ'¢ÃÆ'¢â‚¬Å¡ÃÆ'‚¬ÃÆ'Æ'…ÃÆ'‚¡ > > > > > > > > > > > > > The USB throughput can vary from one machine to the > > > other. > > > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'†'ÃÆ'Æ'Æ'ÃÆ'‚¢ÃÆ'Æ'¢ÃÆ'¢â‚¬Å¡ÃÆ'‚¬ÃÆ'Æ'…ÃÆ'‚¡ > > > > > > > > > > > > > No data for the GViewer comes from KFLOP it all comes > > > from > > > > > > > > the > > > > > > > > > > > > trajectory planner on the PC. > > > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'†'ÃÆ'Æ'Æ'ÃÆ'‚¢ÃÆ'Æ'¢ÃÆ'¢â‚¬Å¡ÃÆ'‚¬ÃÆ'Æ'…ÃÆ'‚¡ > > > > > > > > > > > > > Regards > > > > > > > > > > > > > TK > > > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'†'ÃÆ'Æ'Æ'ÃÆ'‚¢ÃÆ'Æ'¢ÃÆ'¢â‚¬Å¡ÃÆ'‚¬ÃÆ'Æ'…ÃÆ'‚¡ > > > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'†'ÃÆ'Æ'Æ'ÃÆ'‚¢ÃÆ'Æ'¢ÃÆ'¢â‚¬Å¡ÃÆ'‚¬ÃÆ'Æ'…ÃÆ'‚¡ > > > > > > > > > > > > > > > > > > > > > > > > > > From: albertplatek <albertplatek86@> > > > > > > > > > > > > > To: DynoMotion@yahoogroups.com > > > > > > > > > > > > > Sent: Thursday, January 12, 2012 11:11 AM > > > > > > > > > > > > > Subject: [DynoMotion] Coord buffer overflow again > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'†'ÃÆ'Æ'Æ'ÃÆ'‚¢ÃÆ'Æ'¢ÃÆ'¢â‚¬Å¡ÃÆ'‚¬ÃÆ'Æ'…ÃÆ'‚¡ > > > > > > > > > > > > > Hi Tom, > > > > > > > > > > > > > > > > > > > > > > > > > > 1. Why on 1 pc kmotioncnc works ok, but on second not, > > > > > > > > with same > > > > > > > > > > > > settings? > > > > > > > > > > > > > (collinear tollerance 1) > > > > > > > > > > > > > same file, same accelerations etc. > > > > > > > > > > > > > on second I have only usb 3.0 ports. > > > > > > > > > > > > > > > > > > > > > > > > > > 2. How much data come to Gviewer from kflop? if a lot > > > is > > > > > > > > it possible > > > > > > > > > > > > > to bypass this if I dont need to use G viewer? > > > > > > > > > > > > > > > > > > > > > > > > > > I make a lot of reliefs in wood, with kinda high > > > > > > > > acceleleration 40, if > > > > > > > > > > > > > will be any way to get more performance from kflop with > > > > > > > > smoothing > > > > > > > > > > > > feature will be very good. > > > > > > > > > > > > > > > > > > > > > > > > > > Best Regards > > > > > > > > > > > > > Albert > > > PÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'†'ÃÆ'Æ'Æ'ÃÆ'‚¢ÃÆ'Æ'¢ÃÆ'¢â‚¬Å¡ÃÆ'‚¬ÃÆ'Æ'…ÃÆ'‚¡ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚³atek > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > ________________________________ > > From: albertplatek <albertplatek86@...> > To: DynoMotion@yahoogroups.com > Sent: Wednesday, February 8, 2012 8:03 AM > Subject: [DynoMotion] Re: Coord buffer overflow again > Group: DynoMotion Message: 3816 From: albertplatek Date: 2/9/2012 Subject: Re: Coord buffer overflow again Hi Tom, I made test with gamepad conencted to pc, and after 20min I got this jerk of acceleration. After this I again made test without gamepad and it was working fine. Gamepad is not important for me, so I will not use it in future. Best Regards Albert P³atek --- In DynoMotion@yahoogroups.com, "albertplatek" <albertplatek86@...> wrote: > > Hi Tom, > "Have you tried re-connecting the GamPad and pushing the buttons to see if you can force errors?" > - no nothing happen, accelerations are normal. > I am happy that it is finally working and I finished this big reliefs. I can test tomorrow same gcode with gamepad connected to pc and check if it will happen again. > > Best Regards > Albert P³atek > > > --- In DynoMotion@yahoogroups.com, Tom Kerekes <tk@> wrote: > > > > Hi Albert, > >   > > Wow that is strange.  I just double checked and V4.28 is in fact testing the Gampad enable option.  So I have no explaination how the Gampad could make a difference.  Possibly it somehow corrupts the USB communication - but that should generate an error.  Maybe it causes Coordinated Motion Segments to just get completely lost (skipped). > >   > > That is something I couldn't understand about the problem, there should be no randomness.  Everything we are doing should be completely deterministic.  The same GCode and the same Settings should generate the exact same motion every time. > >   > > Have you tried re-connecting the GamPad and pushing the buttons to see if you can force errors? > > > > Regards > > TK > >   > >   > > > >   > > Hi Tom, > > > > I think that in 99% I found bug, it was gamepad connected into usb. > > Even if it was disabled in KMmotionCNC, it was causing randomly this > > crazy jerk of acceleration. Today I tested this big gcode program 3 times without gamepad and everything is fine! :) > > > > Best Regards > > Albert P³atek > > > > --- In DynoMotion@yahoogroups.com, Albert P³atek <albertplatek86@> wrote: > > > > > > Hi Tom, > > > "So are you saying that it did not report any errors until after 10 > > > minutes then printed those?" > > > - It was printing accel peak around 60in/sec2 sometimes from start and > > > after 10 min this big error. Second job execute was ok by about 20min then > > > again error. > > > Smoothing was Tau= 0.008. > > > > > > > > > I just made new test on other pc with file "test_tolerance_0.005.ngc" . And > > > I got *1166in/sec2* acceleration peak... and it stalled X motor. > > > > > > It is same program as before but with toolpath tolerance of 0.005. I > > > thought that maybe this will help. But it didnt. (old file Relief14.ngc had > > > toolpath tolerance of 0.01) > > > Small programs works even with extream acceleration of 60, but this program > > > and other bigger programs fail even with acceleration of 20. > > > > > > Here all my settings: > > > > > > *Trajectory planner* > > > > > > Break Angle 30 degrees > > > Look Ahead 3 sec > > > Collinear Tolerance 0.0005 in > > > Corner Tolerance 0.01 in > > > Facet Angle 1.5 > > > > > > *Axis Parameters* > > > > > > X Resolution 254000 Vel 5 Accel 40 > > > Y Resolution 127000 Vel 4 Accel 10 > > > Z Resolution 254000 Vel 5 Accel 40 > > > A Resolution 7055.5555 Vel 100 Accel 15 (I dont use this axis in programs > > > at moment) > > > > > > Smoothing Tau = 0.01 > > > > > > Gamepad disabled > > > Lathe disabled > > > I dont use MPG and it is not connected. > > > > > > Gcode program (test_tolerance_0.005.ngc) started from first line, I was > > > only capturing Acceleration > 58. > > > > > > Wt, lut 07, 2012, 12:42:31 KMotion Program Started > > > TAU = 0.010000 > > > ACCEL= 58.677936 > > > ACCEL= 58.927802 > > > ACCEL= 58.960033 > > > ACCEL= 60.769580 > > > ACCEL= 58.512884 > > > ACCEL= 58.924281 > > > ACCEL= 59.024541 > > > ACCEL= 58.133892 > > > ACCEL= 61.956729 > > > ACCEL= 59.006616 > > > ACCEL= 60.105317 > > > ACCEL= 58.828274 > > > ACCEL= 60.622193 > > > ACCEL= 58.060298 > > > ACCEL= 60.837834 > > > ACCEL= 58.185082 > > > ACCEL= 58.925759 > > > ACCEL= 59.444038 > > > ACCEL= 61.017190 > > > ACCEL= 59.082489 > > > ACCEL= 60.176545 > > > ACCEL= 58.677989 > > > ACCEL= 60.129486 > > > ACCEL= 59.890981 > > > ACCEL= 63.703736 > > > ACCEL= 58.300426 > > > ACCEL= 58.390017 > > > ACCEL= 64.994889 > > > ACCEL= 61.344673 > > > ACCEL= 61.264344 > > > ACCEL= 60.206659 > > > ACCEL= 61.110118 > > > ACCEL= 58.861575 > > > ACCEL= 62.844393 > > > ACCEL= 58.712601 > > > ACCEL= 62.743716 > > > ACCEL= 59.412413 > > > ACCEL= 59.090724 > > > ACCEL= 59.607079 > > > ACCEL= 58.240991 > > > ACCEL= 58.200542 > > > ACCEL= 62.400201 > > > ACCEL= 64.401300 > > > ACCEL= 60.829834 > > > ACCEL= 60.765056 > > > ACCEL= 58.584326 > > > ACCEL= 59.368296 > > > ACCEL= 58.845166 > > > ACCEL= 63.692201 > > > ACCEL= 59.398388 > > > ACCEL= 58.492754 > > > ACCEL= 58.201424 > > > ACCEL= 59.713704 > > > ACCEL= 59.488310 > > > ACCEL= 58.663588 > > > ACCEL= 62.887233 > > > ACCEL= 58.256548 > > > ACCEL= 58.483484 > > > ACCEL= 58.042246 > > > ACCEL= 58.518015 > > > ACCEL= 61.312055 > > > ACCEL= 60.224222 > > > ACCEL= 60.041371 > > > ACCEL= 61.019005 > > > ACCEL= 59.857207 > > > ACCEL= 60.667182 > > > ACCEL= 62.517183 > > > ACCEL= 62.978943 > > > ACCEL= 60.672158 > > > ACCEL= 62.764637 > > > ACCEL= 60.755186 > > > ACCEL= 58.177374 > > > ACCEL= 58.144752 > > > ACCEL= 62.349561 > > > ACCEL= 59.094084 > > > ACCEL= 59.707611 > > > ACCEL= 60.042771 > > > ACCEL= 59.493590 > > > ACCEL= 58.381740 > > > ACCEL= 58.552432 > > > ACCEL= 60.385809 > > > ACCEL= 61.290063 > > > ACCEL= 63.347741 > > > ACCEL= 59.724964 > > > ACCEL= 66.035414 > > > ACCEL= 58.827807 > > > ACCEL= 60.704297 > > > ACCEL= 60.619439 > > > ACCEL= 60.255062 > > > ACCEL= 59.246636 > > > ACCEL= 58.966244 > > > ACCEL= 62.119289 > > > ACCEL= 61.110753 > > > ACCEL= 64.018676 > > > ACCEL= 63.911028 > > > ACCEL= 60.811810 > > > ACCEL= 59.177635 > > > ACCEL= 61.196853 > > > ACCEL= 58.680083 > > > ACCEL= 59.527456 > > > ACCEL= 58.087839 > > > ACCEL= 59.119465 > > > ACCEL= 62.853732 > > > ACCEL= 59.093458 > > > ACCEL= 59.161929 > > > ACCEL= 61.429357 > > > ACCEL= 58.803188 > > > ACCEL= 59.175222 > > > ACCEL= 60.219244 > > > ACCEL= 58.596575 > > > ACCEL= 58.764846 > > > ACCEL= 58.426363 > > > ACCEL= 60.086656 > > > ACCEL= 59.610107 > > > ACCEL= 59.229815 > > > ACCEL= 58.044244 > > > ACCEL= 63.695621 > > > ACCEL= 58.131310 > > > ACCEL= 59.107058 > > > ACCEL= 58.927665 > > > ACCEL= 61.165485 > > > ACCEL= 58.315186 > > > ACCEL= 60.871648 > > > ACCEL= 60.176458 > > > ACCEL= 59.214217 > > > ACCEL= 59.208457 > > > ACCEL= 62.621824 > > > ACCEL= 58.398335 > > > ACCEL= 61.152436 > > > ACCEL= 58.604786 > > > ACCEL= 58.729217 > > > ACCEL= 59.923425 > > > ACCEL= 59.677849 > > > ACCEL= 58.123086 > > > ACCEL= 60.433701 > > > ACCEL= 58.291893 > > > ACCEL= 60.252449 > > > ACCEL= 60.417103 > > > ACCEL= 60.433561 > > > ACCEL= 61.127250 > > > ACCEL= 61.402805 > > > ACCEL= 58.202147 > > > ACCEL= 62.299667 > > > ACCEL= 62.973329 > > > ACCEL= 61.678133 > > > ACCEL= 58.407879 > > > ACCEL= 59.317560 > > > ACCEL= 58.968064 > > > ACCEL= 58.048711 > > > ACCEL= 58.288685 > > > ACCEL= 58.241909 > > > ACCEL= 59.930745 > > > ACCEL= 59.450933 > > > ACCEL= 59.439070 > > > ACCEL= 58.143918 > > > ACCEL= 58.259981 > > > ACCEL= 58.482531 > > > ACCEL= 58.101452 > > > ACCEL= 60.411654 > > > ACCEL= 60.631439 > > > ACCEL= 62.469302 > > > ACCEL= 60.847604 > > > ACCEL= 58.350368 > > > *ACCEL= 1166.694794 > > > ACCEL= 450.118253* > > > ACCEL= 186.931892 > > > ACCEL= 91.714558 > > > ACCEL= 60.190257 > > > ACCEL= 60.530275 > > > ACCEL= 59.028122 > > > > > > Here I stopped program line was N463261. > > > Smoothing TAU 0.01 > > > > > > In attechement INIT.c config program. > > > GCode program is in Yahoo Files folder "Albert CNC > > > files " (test_tolerance_0.005.part01.rar > > > and test_tolerance_0.005.part02.rar) > > > > > > Best Regards > > > Albert Platek > > > > > > > > > 2012/2/7 Tom Kerekes <tk@> > > > > > > > ** > > > > > > > > [Attachment(s)<https://mail.google.com/mail/html/compose/static_files/blank_quirks.html#1355587bcb0eb6ec_TopText>from Tom Kerekes included below] > > > > > > > > Hi Albert, > > > > > > > > Your program seems reasonable but the numbers are crazy. > > > > > > > > So are you saying that it did not report any errors until after 10 minutes > > > > then printed those? > > > > > > > > I made a similar program and ran it over night and didn't see such a > > > > problem. See attached. It displays the velocity and acceleration into > > > > unused axes 6 and 7 Destinations on the Axis Screen. It also captures the > > > > Max Velocity and Acceleration and displays those as 6 and 7 Positions. > > > > > > > > To be honest I see some tiny glitches and anomolies probably due to > > > > numerical rounding that I would like to better understand, but with the > > > > smoothing filter of 8ms it attenuates them way down. The biggest > > > > acceleration spike I see in the entire multi hour job is 90 in/sec2. > > > > > > > > Did you have the smoothing turned on? > > > > > > > > Is your Resolution 254,000 steps per inch? > > > > > > > > You don't have an MPG connected do you? > > > > > > > > I will try your program to see if I get similar results. > > > > > > > > Thanks > > > > TK > > > > > > > > *From:* albertplatek <albertplatek86@> > > > > *To:* DynoMotion@yahoogroups.com > > > > *Sent:* Monday, February 6, 2012 8:24 AM > > > > *Subject:* [DynoMotion] Re: Coord buffer overflow again > > > > > > > > > > > > Hi Tom, > > > > > > > > Sorry Tom but still this happens randomly :/ I was just lucky that after > > > > firmware flash 1 job was ok. > > > > I use ver 4.28. > > > > > > > > I dont know how to determine line numner of this jerk, but you can see > > > > huge acceleration! X motor stalled. > > > > It was runinng about 10min from program start. > > > > Acceleration was set 32in/sec2 > > > > Break Angle 30 > > > > Tau 0.008 > > > > Collinear Tolerance 0.0005 > > > > Corener Tolerance 0.01 > > > > Facet angle 0.5 > > > > Vel 4in/sec > > > > > > > > Here is my simple C program to check acceleration please check if it is > > > > correct: > > > > > > > > #include "KMotionDef.h" > > > > #define channel ch0 > > > > #define time 0.00018 > > > > main() > > > > { > > > > double X1,X2,speed1,speed2,accel; > > > > int k; > > > > > > > > while(1) > > > > { > > > > for (k=0; k<1; k++) WaitNextTimeSlice(); > > > > X1 = channel->Dest; > > > > for (k=0; k<1; k++) WaitNextTimeSlice(); > > > > X2 = channel->Dest; > > > > speed1=(X2-X1)/(time); > > > > > > > > for (k=0; k<1; k++) WaitNextTimeSlice(); > > > > X1 = channel->Dest; > > > > for (k=0; k<1; k++) WaitNextTimeSlice(); > > > > X2 = channel->Dest; > > > > speed2=(X2-X1)/(time); > > > > > > > > accel=(speed1-speed2)/(time); > > > > accel=(accel/254000)/2; > > > > if(accel>45) > > > > { > > > > printf("AccelError= %f\n",accel); > > > > } > > > > } > > > > } > > > > > > > > Here log: > > > > > > > > AccelError= 45.009219 > > > > AccelError= 45.229875 > > > > AccelError= 46.294106 > > > > AccelError= 45.055529 > > > > AccelError= 45.747547 > > > > AccelError= 45.070729 > > > > AccelError= 45.360541 > > > > AccelError= 45.765590 > > > > AccelError= 46.302076 > > > > AccelError= 46.775562 > > > > AccelError= 47.051896 > > > > AccelError= 45.225101 > > > > AccelError= 45.730674 > > > > AccelError= 46.758273 > > > > AccelError= 46.838768 > > > > AccelError= 45.698572 > > > > AccelError= 46.918682 > > > > AccelError= 45.204219 > > > > AccelError= 45.900150 > > > > AccelError= 47.023519 > > > > AccelError= 47.551550 > > > > AccelError= 48.022541 > > > > AccelError= 46.395295 > > > > AccelError= 81526.314840 > > > > AccelError= 72853.056587 > > > > AccelError= 65101.872298 > > > > AccelError= 58175.992747 > > > > AccelError= 51986.611437 > > > > AccelError= 46456.046360 > > > > AccelError= 41513.718492 > > > > AccelError= 37097.300673 > > > > AccelError= 33150.797142 > > > > AccelError= 29622.770379 > > > > AccelError= 26468.350534 > > > > AccelError= 23652.797849 > > > > AccelError= 21136.323622 > > > > AccelError= 18888.213633 > > > > AccelError= 16879.065077 > > > > AccelError= 15083.474245 > > > > AccelError= 13479.186003 > > > > AccelError= 12045.566722 > > > > AccelError= 10764.204608 > > > > AccelError= 9619.483512 > > > > AccelError= 8596.141212 > > > > AccelError= 7682.015002 > > > > AccelError= 6866.662006 > > > > AccelError= 6139.961139 > > > > AccelError= 5490.446488 > > > > AccelError= 4910.042286 > > > > AccelError= 4391.574405 > > > > AccelError= 3928.114510 > > > > AccelError= 3514.000485 > > > > AccelError= 3144.002655 > > > > AccelError= 2813.316136 > > > > AccelError= 2517.500643 > > > > AccelError= 2253.490792 > > > > AccelError= 2017.432428 > > > > AccelError= 1806.266152 > > > > AccelError= 1617.706032 > > > > AccelError= 1448.982225 > > > > AccelError= 1298.398178 > > > > AccelError= 1163.621865 > > > > AccelError= 1043.472200 > > > > AccelError= 935.821407 > > > > AccelError= 839.766485 > > > > AccelError= 753.863074 > > > > AccelError= 677.101783 > > > > AccelError= 608.587862 > > > > AccelError= 547.209083 > > > > AccelError= 492.589836 > > > > AccelError= 443.554631 > > > > AccelError= 399.735924 > > > > AccelError= 360.639632 > > > > AccelError= 325.643422 > > > > AccelError= 294.370902 > > > > AccelError= 266.425861 > > > > AccelError= 241.478288 > > > > AccelError= 219.191672 > > > > AccelError= 199.246085 > > > > AccelError= 181.422761 > > > > AccelError= 165.532090 > > > > AccelError= 151.295897 > > > > AccelError= 138.574472 > > > > AccelError= 127.206640 > > > > AccelError= 117.084135 > > > > AccelError= 108.003075 > > > > AccelError= 99.888264 > > > > AccelError= 92.598183 > > > > AccelError= 86.122270 > > > > AccelError= 76.770777 > > > > AccelError= 65.225363 > > > > AccelError= 54.908134 > > > > AccelError= 45.688617 > > > > AccelError= 45.070435 > > > > AccelError= 45.411744 > > > > AccelError= 45.770036 > > > > AccelError= 45.871485 > > > > AccelError= 45.239885 > > > > AccelError= 45.631836 > > > > AccelError= 45.001237 > > > > AccelError= 45.373033 > > > > AccelError= 45.577907 > > > > AccelError= 46.322608 > > > > AccelError= 46.958820 > > > > AccelError= 46.038556 > > > > AccelError= 45.018088 > > > > > > > > If you test it on your PC and will be ok, then maybe something is wrong > > > > with my laptop. Then I will test on other PC. > > > > > > > > Best Regards > > > > Albert P³atek > > > > > > > > --- In DynoMotion@yahoogroups.com, "albertplatek" <albertplatek86@> > > > > wrote: > > > > > > > > > > Hi Tom, > > > > > > > > > > Problem solved (probably), weird but after re-flash firmware 4.28 > > > > > everything is working ok, no axis stall. > > > > > > > > > > btw. Experimental version KMotion429x.exe is not working, when I start > > > > > execute gcode I get message "G Code Error GCode Aborted", this error > > > > > occurs on any gcode program even on only few G01 lines. > > > > > > > > > > Tom if you have C program that watch for too high of acceleration > > > > > please post it. If again in future something strange will > > > > > happen I will try find out the problem. > > > > > I am still not sure what was the problem in my system that axis > > > > > were stalling. > > > > > > > > > > Best Regards > > > > > Albert P³atek > > > > > > > > > > > > > > > > > > > > --- In DynoMotion@yahoogroups.com, Albert P³atek <albertplatek86@> > > > > wrote: > > > > > > > > > > > > Hi Tom, > > > > > > > > > > > > Thank you for help. > > > > > > Mostly stall X axis, and sometimes Z. > > > > > > Y axis seem to be ok. > > > > > > I uploaded file as an attachment (zip). > > > > > > > > > > > > Best Regards > > > > > > Albert Platek > > > > > > > > > > > > 2012/2/5 Tom Kerekes <tk@> > > > > > > > > > > > > > ** > > > > > > > > > > > > > > > > > > > > > Hi Albert, > > > > > > > > > > > > > > Thanks for the very complete problem description, I can't tell you > > > > how > > > > > > > much I appreciate that, but I can't think of anything that would > > > > cause > > > > > > > this. Anything with large files I might expect some processing issue > > > > but > > > > > > > that would result in a buffer underflow and an error message not > > > > incorrect > > > > > > > motion that would result in a stall. That delay I described in your > > > > email > > > > > > > only happens when "Execute" is pushed so I don't think it is related. > > > > > > > > > > > > > > There is an experimental version KMotion429x.exe on the download site > > > > > > > that has this delay eliminated and also faster USB throughput, but I > > > > don't > > > > > > > think either of those are related to this problem. > > > > > > > > > > > > > > Can you post the file somewhere? I'm thinking to write a C program > > > > that > > > > > > > would run in KFLOP and watch for too high of acceleration or > > > > velocity by > > > > > > > monitoring how ch0->Dest changes. If it detects something bad that > > > > could > > > > > > > be used as a trigger to help us track it down. > > > > > > > > > > > > > > Which axes loose position? > > > > > > > > > > > > > > Regards > > > > > > > TK > > > > > > > > > > > > > > > > > > > > > *From:* albertplatek <albertplatek86@> > > > > > > > *To:* DynoMotion@yahoogroups.com > > > > > > > *Sent:* Sunday, February 5, 2012 5:27 AM > > > > > > > > > > > > > > *Subject:* [DynoMotion] Re: Coord buffer overflow again > > > > > > > > > > > > > > > > > > > > > Hi Tom, > > > > > > > > > > > > > > My settings: > > > > > > > Break angle 30 deg > > > > > > > Look ahead 3 sec > > > > > > > Collinear tolerance 0.001 in > > > > > > > Corner tolerance 0.003 in > > > > > > > Facet angle 1.5 deg > > > > > > > Tau = 0.008 sec > > > > > > > XYZ Accelarations 40 in/sec2 > > > > > > > XYZ Velocity 4.5 in/sec > > > > > > > In gcode file I use max Vel 270 in/min (4.5 in/sec) > > > > > > > > > > > > > > I have strange problem, everything was ok until I started machining > > > > > > > bigger reliefs. > > > > > > > At half of the job my stepper motors lost position > > > > > > > probably becouse of high jerk of acceleration. > > > > > > > Firstly I thought that maybe axis accelerations for my system are to > > > > > > > high(40in/sec2). So I changed it to 20in/sec2 for test. > > > > > > > But didnt help, still stepper motors stall at half of the gcode file. > > > > > > > > > > > > > > This file contain 548389 lines, my steppers stall between 226000 - > > > > 274000 > > > > > > > line but not always in same place. > > > > > > > > > > > > > > I tried to find line where the problem occurs, but when I found line > > > > with > > > > > > > problem and tried to execute 500 lines before, it was ok. Motors > > > > didnt > > > > > > > stall. > > > > > > > I found that motors stall when KMotionCNC start execute at least > > > > > > > 100 000 lines before line number 274000. > > > > > > > > > > > > > > The problem disappear when I set break angle 5deg. > > > > > > > > > > > > > > "This diagnostic pointed out a side problem I hadn't noticed before. > > > > There > > > > > > > is an annoying ~30 second delay when it runs the second time. The > > > > > > > Interpreter keeps a record of all its state changes which allow > > > > Halts to go > > > > > > > back to any previous state. In this test of 100x500 subroutine calls > > > > > > > results in about 1 million state changes. Ironically creating the > > > > data, > > > > > > > making use of the data all happens in the blink of an eye, but > > > > discarding > > > > > > > the data (CList.RemoveAll) takes 30 seconds! We will find a better > > > > way in > > > > > > > the next version. For now just don't be surprised if there is an > > > > annoying > > > > > > > delay on a second run of a big job." > > > > > > > > > > > > > > Can this "delay" make that my stepper motors stall? > > > > > > > > > > > > > > Best Regards > > > > > > > Albert P³atek > > > > > > > > > > > > > > --- In DynoMotion@yahoogroups.com, Albert <albertplatek86@> wrote: > > > > > > > > > > > > > > > > Hi Tom, > > > > > > > > > > > > > > > > Thank you. > > > > > > > > Btw. I am trying to find out what changes make > > > > > > > > Fidelizer in Windows. If I will find something > > > > > > > > important I will let you know. > > > > > > > > Best regards > > > > > > > > Albert Platek > > > > > > > > > > > > > > > > > > > > > > > > Dnia 2012-01-18 o godz. 23:47 Tom Kerekes <tk@> napisaÃÆ'â€¦Ã¢â‚¬à ¡ > > > > > > > > (a): > > > > > > > > > > > > > > > > > Hi Albert, > > > > > > > > > > > > > > > > > > Just wanted to mention that I now do notice an improvement with > > > > > > > > > Fidelizer. I don't understand why I didn't before. I might have > > > > > > > > > made a mistake or something changed. > > > > > > > > > > > > > > > > > > I sent an email to FTDI (USB chip we use) regarding it and the > > > > USB > > > > > > > > > 3.0 port issue to see if they have any answers. I figure it is > > > > > > > > > worth the cost of an email :} > > > > > > > > > > > > > > > > > > Still working on some optimization... > > > > > > > > > > > > > > > > > > Regards > > > > > > > > > TK > > > > > > > > > > > > > > > > > > > > > > > > > > > From: albertplatek <albertplatek86@> > > > > > > > > > To: DynoMotion@yahoogroups.com > > > > > > > > > Sent: Monday, January 16, 2012 12:01 PM > > > > > > > > > Subject: [DynoMotion] Re: Coord buffer overflow again > > > > > > > > > > > > > > > > > > > > > > > > > > > Hi Tom, > > > > > > > > > Thank you for making changes in new version to increase the > > > > segment > > > > > > > > > rate. > > > > > > > > > > > > > > > > > > Btw: > > > > > > > > > I installed on second pc Windows 7 (Core2Duo cpu), and aslo > > > > tested > > > > > > > > > with fidelizer. Strange but didn't make difference, as in your > > > > > > > > > computer. > > > > > > > > > > > > > > > > > > I cant get why in this particular case (pc with WIN7 i7 cpu) it > > > > helps. > > > > > > > > > Maybe it depends on type of host controller... > > > > > > > > > > > > > > > > > > Pc with WIN7 64bit (Core2Duo) has > > > > > > > > > SiS 7001 PCI to USB Open Host Controller > > > > > > > > > > > > > > > > > > Pc with WIN7 64bit (i7) has > > > > > > > > > Intel(R) 6 Series/C200 Series Chipset Family USB Enhanced Host > > > > > > > > > Controller > > > > > > > > > > > > > > > > > > This computer has aslo another usb ports: Renesas Electronics USB > > > > > > > > > 3.0 Host Controller > > > > > > > > > which works very bad with KFlop. And on this port fidelizer make > > > > no > > > > > > > > > difference. > > > > > > > > > > > > > > > > > > Best Regards > > > > > > > > > Albert Platek > > > > > > > > > > > > > > > > > > --- In DynoMotion@yahoogroups.com, Tom Kerekes <tk@> wrote: > > > > > > > > > > > > > > > > > > > > Hi Albert, > > > > > > > > > > ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > > > > Thanks for that interesting information. > > > > > > > > > > ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > > > > I tried the Fidelizer on my i7 W7 32 machine and it didn't > > > > seem to > > > > > > > > > make any difference. > > > > > > > > > > ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > > > > I'm trying some things to increase the segment rate by sending > > > > > > > > > blocks of segments and optimizing how they are handled in > > > > KFLOP.ÃÆ'Æ'Ã¢â‚¬à ¡ I > > > > > > > > > f this bears fruit I'll let you know. > > > > > > > > > > ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > > > > This diagnostic pointed out a side problem I hadn't noticed > > > > before > > > > > > > > > .ÃÆ'Æ'Ã¢â‚¬à ¡ There is an annoying ~30 second delay when it runs the > > > > second ti > > > > > > > > > me.ÃÆ'Æ'Ã¢â‚¬à ¡ The Interpreter keeps a record of all its state changes > > > > which > > > > > > > > > allow Halts to go back to any previous state.ÃÆ'Æ'Ã¢â‚¬à ¡ In this test > > > > of 100x > > > > > > > > > 500 subroutine calls results in about 1 million state > > > > changes.ÃÆ'Æ'Ã¢â‚¬à ¡ Iro > > > > > > > > > nically creating the data, making use of the data all happens in > > > > the > > > > > > > > > blink of an eye, but discarding the data (CList.RemoveAll) takes > > > > 30 > > > > > > > > > seconds!ÃÆ'Æ'Ã¢â‚¬à ¡ We will find a better way in the next > > > > version.ÃÆ'Æ'Ã¢â‚¬à ¡ For now > > > > > > > > > just don't be surprisedÃÆ'Æ'Ã¢â‚¬à ¡ if there is an annoying delay on a > > > > second > > > > > > > > > run of a big job. > > > > > > > > > > ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > > > > Regards > > > > > > > > > > TK > > > > > > > > > > ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > ________________________________ > > > > > > > > > > From: albertplatek <albertplatek86@> > > > > > > > > > > To: DynoMotion@yahoogroups.com > > > > > > > > > > Sent: Sunday, January 15, 2012 8:37 AM > > > > > > > > > > Subject: [DynoMotion] Re: Coord buffer overflow again > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > > > > Hi Tom, > > > > > > > > > > First thank you Tom for the test file. > > > > > > > > > > I made few tests and I got interesting results :) > > > > > > > > > > > > > > > > > > > > PC No1: > > > > > > > > > > > > > > > > > > > > Intel Pentium 4 2.6Ghz WinXP 32bit port USB2.0 > > > > > > > > > > Max segments: 110ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > > > > Time: 31s > > > > > > > > > > 100 circles x 110 segments/circle / 31 seconds =354 segments > > > > per > > > > > > > > > second. > > > > > > > > > > > > > > > > > > > > PC No2: > > > > > > > > > > > > > > > > > > > > Intel Core2Duo 2.0 Ghz WinXP 32bit port USB2.0 > > > > > > > > > > Max segments: 140ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > > > > Time: 31s > > > > > > > > > > 100 circles x 140 segments/circle / 31 seconds = 451 segments > > > > per > > > > > > > > > second. > > > > > > > > > > > > > > > > > > > > PC No3: > > > > > > > > > > > > > > > > > > > > i7 2670QM Win7 64bit port USB3.0 > > > > > > > > > > Max segments:30ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > > > > Time:31s > > > > > > > > > > 100 circles x 60 segments/circle / 31 seconds = 193 segments > > > > per > > > > > > > > > second. > > > > > > > > > > (thats why I had problem here, usb 3.0 fail) > > > > > > > > > > > > > > > > > > > > (Luckily I had 1 USB 2.0 port) > > > > > > > > > > > > > > > > > > > > i7 2670QM Win7 64bit port USB2.0 > > > > > > > > > > Max segments:170 > > > > > > > > > > Time:31s > > > > > > > > > > 100 circles x 170 segments/circle / 31 seconds = 548 segments > > > > per > > > > > > > > > second. > > > > > > > > > > > > > > > > > > > > i7 2670QM Win7 64bit port USB2.0 (with 1 core affinity for > > > > > > > > > KmotionCNC and KmotionServer) > > > > > > > > > > Max segments:200 > > > > > > > > > > Time:31s > > > > > > > > > > 100 circles x 200 segments/circle / 31 seconds =645 segments > > > > per > > > > > > > > > second. > > > > > > > > > > > > > > > > > > > > I found on net some interesting software called Fidelizer, its > > > > > > > > > software to optimize win7/vista for audio workstation, but it > > > > aslo > > > > > > > > > tweak somehow usb throughput. > > > > > > > > > > More detail here: /www.windowsxlive.net/fidelizer > > > > > > > > > > > > > > > > > > > > i7 2670QM Win7 64bit port USB2.0 (Fidelizer) > > > > > > > > > > Max segments:410 > > > > > > > > > > Time:31s > > > > > > > > > > 100 circles x 410 segments/circle / 31 seconds = 1320 segments > > > > per > > > > > > > > > second. (wow! tested few times with same result) > > > > > > > > > > > > > > > > > > > > Be aware of new USB 3.0 ports they may not work with good > > > > > > > > > performance with KFlop, maybe becouse of bug in drivers. Windows > > > > 7 > > > > > > > > > dont see them without additional drivers. > > > > > > > > > > Fidelizer software only improved performance on pc with ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > Win7. > > > > > > > > > > On WinXP seem to not work. I will soon try install on "PC No2" > > > > > > > > > Windows 7 and then try fidelizer also. > > > > > > > > > > > > > > > > > > > > Best regards > > > > > > > > > > Albert PÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚³atek > > > > > > > > > > --- In DynoMotion@yahoogroups.com, Tom Kerekes <tk@> wrote: > > > > > > > > > > > > > > > > > > > > > > Hi Albert, > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > > > > > Your application/system is quite demanding :} > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > > > > > That portion of GCode you sent is 100,000 lines of all G1 > > > > > > > > > movements with typical tiny lengths of only > > > > 50~150um.ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ The > > > > > > > > > collinear tolerance of 0.001 (25um) will cause many > > > > blocksÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ to > > > > > > > > > be combined together so the real required vector rate will be > > > > more d > > > > > > > > > ependent on the path curvature rather than the original > > > > GCode.ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > > > > > On my system (i7 W7 32-bit) it runs ok with a facet angle 0f > > > > 0.4 > > > > > > > > > degrees butÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ can't keep up at 0.3 degrees. > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > > > > > I did some calculations for a system like yours with 40 > > > > in/sec2 > > > > > > > > > acceleration.ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ Here is what I find: > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > > > > > That acceleration allows going around a 0.1 inch radius > > > > circle > > > > > > > > > at 120ipm (which is about 3 circles per second) > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > > > > > With a facet angle setting of 1.5 degrees that requires 240 > > > > > > > > > segments per circle which would require 720 segments per second > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > > > > > I created a diagnostic test GCode file that programmatically > > > > > > > > > generates an N sided Polygon 100 timesÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ with > > > > radius 0.1 inches > > > > > > > > > and allows you to specify the number of polygon sides (it also > > > > gradu > > > > > > > > > ally shifts the "circle" to make the plot more > > > > interesting).ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > > > (see attachment). ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ When using the diagnostic > > > > make sure to set > > > > > > > > > the Collinear tolerance to zero and the Corner Rounding to zero > > > > so t > > > > > > > > > he original segments are used.ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ By changing N > > > > (number of sides > > > > > > > > > we can determine the segment download rate). > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > > > > > For example on my system it works with N = 200 and fails > > > > with N > > > > > > > > > = 240.ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ The total time to do all 100 circles > > > > takes about 30 > > > > > > > > > seconds.ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ So: > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > > > > > 100 circles x 200 segments/circle / 30 seconds = 666 segments > > > > > > > > > per second. > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > > > > > In simulation mode the Interpreter+Graphics finishes the job > > > > in > > > > > > > > > ~ 5 seconds so it isn't a significant factor > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > > > > > With the GCode Window closed it simulates in ~ 4 seconds so > > > > the > > > > > > > > > Graphical Display is not aÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ significant factor. > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > > > > > You might run the tests on your two systems to see if you get > > > > > > > > > drastically different results. > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > > > > > But to try to answer your original question of why one CPU > > > > would > > > > > > > > > be significantly different I don't know.ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ Google > > > > finds some > > > > > > > > > benchmarking tools but they areÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ mainly for High > > > > Speed USB > > > > > > > > > devices like flash drives.ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ Some sites state if > > > > an older USB > > > > > > > > > 1.1 device is attached it slows everything down.ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > > > > > Regards > > > > > > > > > > > TKÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > > > > > > > > > > > > > > > > From: albertplatek albertplatek86@ > > > > > > > > > > > To: DynoMotion@yahoogroups.com > > > > > > > > > > > Sent: Saturday, January 14, 2012 11:02 AM > > > > > > > > > > > Subject: [DynoMotion] Re: Coord buffer overflow again > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ > > > > > > > > > > > Hi Tom, > > > > > > > > > > > How to determine USB throughput on pc? or what pc/system is > > > > best > > > > > > > > > for KFlop? > > > > > > > > > > > > > > > > > > > > > > I have still sometimes cord buffer overflow on second pc > > > > even I > > > > > > > > > > > set lower accelaration 25 and bigger facet angle 2.5 it > > > > sometimes > > > > > > > > > > > overflow. Its very strange pc has windows xp clean > > > > installation, > > > > > > > > > > > Integrated Intel Graphics. > > > > > > > > > > > On first pc works perfect even with high acceleration 40, > > > > facet > > > > > > > > > angle 2.0 (also windows xp, but ati graphics) > > > > > > > > > > > > > > > > > > > > > > Break angle 30 > > > > > > > > > > > Look ahead 3 > > > > > > > > > > > Collinear tolerance 0.001 > > > > > > > > > > > Corner tolerance 0.005 > > > > > > > > > > > Vel 5 > > > > > > > > > > > > > > > > > > > > > > Best Regards > > > > > > > > > > > Albert Platek > > > > > > > > > > > > > > > > > > > > > > --- In DynoMotion@yahoogroups.com, Albert > > > > PÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚³atek > > > > > > > > > albertplatek86@ wrote: > > > > > > > > > > > > > > > > > > > > > > > > Hi Tom, > > > > > > > > > > > > GCode file is attached with this message. > > > > > > > > > > > > Yes feedrate is 300ipm. > > > > > > > > > > > > What do you mean by "path generation slows down"? - i mean > > > > > > > > > that when I am > > > > > > > > > > > > changing view perspective in GViewer > > > > > > > > > > > > the "simulated tool" catching up end of buffered path. So > > > > > > > > > thats why I > > > > > > > > > > > > thought that maybe gviewer impact on performance somehow. > > > > > > > > > > > > > > > > > > > > > > > > Best Regards > > > > > > > > > > > > Albert > > > > PÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'‚¦ÃÆ'Æ'Æ'ÃÆ'‚¢ÃÆ'Æ'¢ÃÆ'¢â‚¬Å¡ÃÆ'‚¬ÃÆ'Æ'…ÃÆ'‚¡atek > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > 2012/1/13 Tom Kerekes tk@ > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > Hi Albert, > > > > > > > > > > > > > > > > > > > > > > > > > > Yes please send the file. Is the feedrate set high > > > > 300ipm? > > > > > > > > > > > > > > > > > > > > > > > > > > What do you mean by "path generation slows down"? > > > > > > > > > > > > > > > > > > > > > > > > > > Thanks > > > > > > > > > > > > > TK > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > *From:* albertplatek albertplatek86@ > > > > > > > > > > > > > *To:* DynoMotion@yahoogroups.com > > > > > > > > > > > > > *Sent:* Thursday, January 12, 2012 3:35 PM > > > > > > > > > > > > > *Subject:* [DynoMotion] Re: Coord buffer overflow again > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > Hi Tom, > > > > > > > > > > > > > > > > > > > > > > > > > > My settings: > > > > > > > > > > > > > > > > > > > > > > > > > > Break angle 30 > > > > > > > > > > > > > Look ahead 3 > > > > > > > > > > > > > Collinear tolerance 0.001 > > > > > > > > > > > > > Corner tolerance 0.005 > > > > > > > > > > > > > Facet angle 2.0 (if I go to 1.5, i get cord buffer > > > > overflow > > > > > > > > > almost from > > > > > > > > > > > > > program start) > > > > > > > > > > > > > > > > > > > > > > > > > > I use Tau = 0.01; (with facet angle 2.0 machine movements > > > > > > > > > are fairly > > > > > > > > > > > > > smooth) > > > > > > > > > > > > > > > > > > > > > > > > > > All 3 axis has same Vel/Accel parameters: > > > > > > > > > > > > > Vel 5.0 > > > > > > > > > > > > > Accel 40 > > > > > > > > > > > > > > > > > > > > > > > > > > Toolpath is parallel along X-axis (contain a lot of small > > > > > > > > > arcs, I will > > > > > > > > > > > > > send file tomorrow) > > > > > > > > > > > > > > > > > > > > > > > > > > With this settings my first router with old laptop works > > > > ok > > > > > > > > > (Dell, core 2 > > > > > > > > > > > > > duo, ati graphics card), > > > > > > > > > > > > > but on second router (same settings) with new laptop dont > > > > > > > > > work, i often > > > > > > > > > > > > > get buffer overflows. (laptop Dell, i7 cpu, Win7 64bit, > > > > > > > > > Nvidia Gt555) > > > > > > > > > > > > > > > > > > > > > > > > > > (Btw. when GViewer is open and I move view by mouse path > > > > > > > > > generation slows > > > > > > > > > > > > > down, is it normal?) > > > > > > > > > > > > > > > > > > > > > > > > > > Best Regards > > > > > > > > > > > > > Albert PÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚³atek > > > > > > > > > > > > > > > > > > > > > > > > > > --- In DynoMotion@yahoogroups.com, Tom Kerekes <tk@> > > > > wrote: > > > > > > > > > > > > > > > > > > > > > > > > > > > > Hi Albert, > > > > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'†'ÃÆ'Æ'Æ'ÃÆ'‚¢ÃÆ'Æ'¢ÃÆ'¢â‚¬Å¡ÃÆ'‚¬ÃÆ'Æ'…ÃÆ'‚¡ > > > > > > > > > > > > > > Could you send me the Gcode and all your > > > > settings?ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'†'ÃÆ'Æ'Æ' > > > > > > > > > ÃÆ'‚¢ÃÆ'Æ'¢ÃÆ'¢â‚¬Å¡ÃÆ'‚¬ÃÆ'Æ'…ÃÆ'‚¡ I don't understand > > > > > > > > > > > > > the collinear tolerance of > > > > 1?ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'†'ÃÆ'Æ'Æ'ÃÆ'‚¢ÃÆ'Æ'¢ÃÆ'¢â‚¬Å¡ÃÆ'‚¬ÃÆ'Æ'…ÃÆ'‚¡ That > > > > > > > > > seems huge.ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'†'ÃÆ'Æ'Æ'ÃÆ'‚¢ÃÆ'Æ'¢ÃÆ'¢â‚¬Å¡ÃÆ'‚¬ÃÆ'Æ'…ÃÆ'‚¡ Did you > > > > mean 0.001? > > > > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'†'ÃÆ'Æ'Æ'ÃÆ'‚¢ÃÆ'Æ'¢ÃÆ'¢â‚¬Å¡ÃÆ'‚¬ÃÆ'Æ'…ÃÆ'‚¡ > > > > > > > > > > > > > > A corner rounding facet angle too small might be > > > > > > > > > generating too many > > > > > > > > > > > > > small > > > > segments.ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'†'ÃÆ'Æ'Æ'ÃÆ'‚¢ÃÆ'Æ'¢ÃÆ'¢â‚¬Å¡ÃÆ'‚¬ÃÆ'Æ'…ÃÆ'‚¡ Are you using the > > > > > > > > > Coordinated Motion Low Pass Filter > > > > > > > > > > > > > Smoothing > > > > option?ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'†'ÃÆ'Æ'Æ'ÃÆ'‚¢ÃÆ'Æ'¢ÃÆ'¢â‚¬Å¡ÃÆ'‚¬ÃÆ'Æ'…ÃÆ'‚¡ That might help make > > > > > > > > > as smooth or smoother motion with > > > > > > > > > > > > > larger facets. > > > > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'†'ÃÆ'Æ'Æ'ÃÆ'‚¢ÃÆ'Æ'¢ÃÆ'¢â‚¬Å¡ÃÆ'‚¬ÃÆ'Æ'…ÃÆ'‚¡ > > > > > > > > > > > > > > The USB throughput can vary from one machine to the > > > > other. > > > > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'†'ÃÆ'Æ'Æ'ÃÆ'‚¢ÃÆ'Æ'¢ÃÆ'¢â‚¬Å¡ÃÆ'‚¬ÃÆ'Æ'…ÃÆ'‚¡ > > > > > > > > > > > > > > No data for the GViewer comes from KFLOP it all comes > > > > from > > > > > > > > > the > > > > > > > > > > > > > trajectory planner on the PC. > > > > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'†'ÃÆ'Æ'Æ'ÃÆ'‚¢ÃÆ'Æ'¢ÃÆ'¢â‚¬Å¡ÃÆ'‚¬ÃÆ'Æ'…ÃÆ'‚¡ > > > > > > > > > > > > > > Regards > > > > > > > > > > > > > > TK > > > > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'†'ÃÆ'Æ'Æ'ÃÆ'‚¢ÃÆ'Æ'¢ÃÆ'¢â‚¬Å¡ÃÆ'‚¬ÃÆ'Æ'…ÃÆ'‚¡ > > > > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'†'ÃÆ'Æ'Æ'ÃÆ'‚¢ÃÆ'Æ'¢ÃÆ'¢â‚¬Å¡ÃÆ'‚¬ÃÆ'Æ'…ÃÆ'‚¡ > > > > > > > > > > > > > > > > > > > > > > > > > > > > From: albertplatek <albertplatek86@> > > > > > > > > > > > > > > To: DynoMotion@yahoogroups.com > > > > > > > > > > > > > > Sent: Thursday, January 12, 2012 11:11 AM > > > > > > > > > > > > > > Subject: [DynoMotion] Coord buffer overflow again > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'†'ÃÆ'Æ'Æ'ÃÆ'‚¢ÃÆ'Æ'¢ÃÆ'¢â‚¬Å¡ÃÆ'‚¬ÃÆ'Æ'…ÃÆ'‚¡ > > > > > > > > > > > > > > Hi Tom, > > > > > > > > > > > > > > > > > > > > > > > > > > > > 1. Why on 1 pc kmotioncnc works ok, but on second not, > > > > > > > > > with same > > > > > > > > > > > > > settings? > > > > > > > > > > > > > > (collinear tollerance 1) > > > > > > > > > > > > > > same file, same accelerations etc. > > > > > > > > > > > > > > on second I have only usb 3.0 ports. > > > > > > > > > > > > > > > > > > > > > > > > > > > > 2. How much data come to Gviewer from kflop? if a lot > > > > is > > > > > > > > > it possible > > > > > > > > > > > > > > to bypass this if I dont need to use G viewer? > > > > > > > > > > > > > > > > > > > > > > > > > > > > I make a lot of reliefs in wood, with kinda high > > > > > > > > > acceleleration 40, if > > > > > > > > > > > > > > will be any way to get more performance from kflop with > > > > > > > > > smoothing > > > > > > > > > > > > > feature will be very good. > > > > > > > > > > > > > > > > > > > > > > > > > > > > Best Regards > > > > > > > > > > > > > > Albert > > > > PÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'†'ÃÆ'Æ'Æ'ÃÆ'‚¢ÃÆ'Æ'¢ÃÆ'¢â‚¬Å¡ÃÆ'‚¬ÃÆ'Æ'…ÃÆ'‚¡ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚³atek > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > ________________________________ > > > > From: albertplatek <albertplatek86@> > > To: DynoMotion@yahoogroups.com > > Sent: Wednesday, February 8, 2012 8:03 AM > > Subject: [DynoMotion] Re: Coord buffer overflow again > > > Group: DynoMotion Message: 3817 From: Tom Kerekes Date: 2/9/2012 Subject: Re: Coord buffer overflow again Hi Albert,   Thanks for reporting back.  I just wish I understood what was going on.   TK Group: DynoMotion Message: 3818 From: Brad Murry Date: 2/9/2012 Subject: Re: Coord buffer overflow again Hello Albert,   You might consider a free trial of http://www.usblyzer.com/download.htm  if you have some extra time on your hands.   I just used it to find a traffic issue and it worked well.   It might show some evidence of USB cross talk or something else weird going on, maybe worth a shot.   -Brad Murry   From: DynoMotion@yahoogroups.com [mailto:DynoMotion@yahoogroups.com] On Behalf Of albertplatek Sent: Thursday, February 09, 2012 12:04 PM To: DynoMotion@yahoogroups.com Subject: [DynoMotion] Re: Coord buffer overflow again     Hi Tom, I made test with gamepad conencted to pc, and after 20min I got this jerk of acceleration. After this I again made test without gamepad and it was working fine. Gamepad is not important for me, so I will not use it in future. Best Regards Albert P³atek --- In DynoMotion@yahoogroups.com, "albertplatek" <albertplatek86@...> wrote: > > Hi Tom, > "Have you tried re-connecting the GamPad and pushing the buttons to see if you can force errors?" > - no nothing happen, accelerations are normal. > I am happy that it is finally working and I finished this big reliefs. I can test tomorrow same gcode with gamepad connected to pc and check if it will happen again. > > Best Regards > Albert P³atek > > > --- In DynoMotion@yahoogroups.com, Tom Kerekes <tk@> wrote: > > > > Hi Albert, > >   > > Wow that is strange.  I just double checked and V4.28 is in fact testing the Gampad enable option.  So I have no explaination how the Gampad could make a difference.  Possibly it somehow corrupts the USB communication - but that should generate an error.  Maybe it causes Coordinated Motion Segments to just get completely lost (skipped). > >   > > That is something I couldn't understand about the problem, there should be no randomness.  Everything we are doing should be completely deterministic.  The same GCode and the same Settings should generate the exact same motion every time. > >   > > Have you tried re-connecting the GamPad and pushing the buttons to see if you can force errors? > > > > Regards > > TK > >   > >   > > > >   > > Hi Tom, > > > > I think that in 99% I found bug, it was gamepad connected into usb. > > Even if it was disabled in KMmotionCNC, it was causing randomly this > > crazy jerk of acceleration. Today I tested this big gcode program 3 times without gamepad and everything is fine! :) > > > > Best Regards > > Albert P³atek > > > > --- In DynoMotion@yahoogroups.com, Albert P³atek <albertplatek86@> wrote: > > > > > > Hi Tom, > > > "So are you saying that it did not report any errors until after 10 > > > minutes then printed those?" > > > - It was printing accel peak around 60in/sec2 sometimes from start and > > > after 10 min this big error. Second job execute was ok by about 20min then > > > again error. > > > Smoothing was Tau= 0.008. > > > > > > > > > I just made new test on other pc with file "test_tolerance_0.005.ngc" . And > > > I got *1166in/sec2* acceleration peak... and it stalled X motor. > > > > > > It is same program as before but with toolpath tolerance of 0.005. I > > > thought that maybe this will help. But it didnt. (old file Relief14.ngc had > > > toolpath tolerance of 0.01) > > > Small programs works even with extream acceleration of 60, but this program > > > and other bigger programs fail even with acceleration of 20. > > > > > > Here all my settings: > > > > > > *Trajectory planner* > > > > > > Break Angle 30 degrees > > > Look Ahead 3 sec > > > Collinear Tolerance 0.0005 in > > > Corner Tolerance 0.01 in > > > Facet Angle 1.5 > > > > > > *Axis Parameters* > > > > > > X Resolution 254000 Vel 5 Accel 40 > > > Y Resolution 127000 Vel 4 Accel 10 > > > Z Resolution 254000 Vel 5 Accel 40 > > > A Resolution 7055.5555 Vel 100 Accel 15 (I dont use this axis in programs > > > at moment) > > > > > > Smoothing Tau = 0.01 > > > > > > Gamepad disabled > > > Lathe disabled > > > I dont use MPG and it is not connected. > > > > > > Gcode program (test_tolerance_0.005.ngc) started from first line, I was > > > only capturing Acceleration > 58. > > > > > > Wt, lut 07, 2012, 12:42:31 KMotion Program Started > > > TAU = 0.010000 > > > ACCEL= 58.677936 > > > ACCEL= 58.927802 > > > ACCEL= 58.960033 > > > ACCEL= 60.769580 > > > ACCEL= 58.512884 > > > ACCEL= 58.924281 > > > ACCEL= 59.024541 > > > ACCEL= 58.133892 > > > ACCEL= 61.956729 > > > ACCEL= 59.006616 > > > ACCEL= 60.105317 > > > ACCEL= 58.828274 > > > ACCEL= 60.622193 > > > ACCEL= 58.060298 > > > ACCEL= 60.837834 > > > ACCEL= 58.185082 > > > ACCEL= 58.925759 > > > ACCEL= 59.444038 > > > ACCEL= 61.017190 > > > ACCEL= 59.082489 > > > ACCEL= 60.176545 > > > ACCEL= 58.677989 > > > ACCEL= 60.129486 > > > ACCEL= 59.890981 > > > ACCEL= 63.703736 > > > ACCEL= 58.300426 > > > ACCEL= 58.390017 > > > ACCEL= 64.994889 > > > ACCEL= 61.344673 > > > ACCEL= 61.264344 > > > ACCEL= 60.206659 > > > ACCEL= 61.110118 > > > ACCEL= 58.861575 > > > ACCEL= 62.844393 > > > ACCEL= 58.712601 > > > ACCEL= 62.743716 > > > ACCEL= 59.412413 > > > ACCEL= 59.090724 > > > ACCEL= 59.607079 > > > ACCEL= 58.240991 > > > ACCEL= 58.200542 > > > ACCEL= 62.400201 > > > ACCEL= 64.401300 > > > ACCEL= 60.829834 > > > ACCEL= 60.765056 > > > ACCEL= 58.584326 > > > ACCEL= 59.368296 > > > ACCEL= 58.845166 > > > ACCEL= 63.692201 > > > ACCEL= 59.398388 > > > ACCEL= 58.492754 > > > ACCEL= 58.201424 > > > ACCEL= 59.713704 > > > ACCEL= 59.488310 > > > ACCEL= 58.663588 > > > ACCEL= 62.887233 > > > ACCEL= 58.256548 > > > ACCEL= 58.483484 > > > ACCEL= 58.042246 > > > ACCEL= 58.518015 > > > ACCEL= 61.312055 > > > ACCEL= 60.224222 > > > ACCEL= 60.041371 > > > ACCEL= 61.019005 > > > ACCEL= 59.857207 > > > ACCEL= 60.667182 > > > ACCEL= 62.517183 > > > ACCEL= 62.978943 > > > ACCEL= 60.672158 > > > ACCEL= 62.764637 > > > ACCEL= 60.755186 > > > ACCEL= 58.177374 > > > ACCEL= 58.144752 > > > ACCEL= 62.349561 > > > ACCEL= 59.094084 > > > ACCEL= 59.707611 > > > ACCEL= 60.042771 > > > ACCEL= 59.493590 > > > ACCEL= 58.381740 > > > ACCEL= 58.552432 > > > ACCEL= 60.385809 > > > ACCEL= 61.290063 > > > ACCEL= 63.347741 > > > ACCEL= 59.724964 > > > ACCEL= 66.035414 > > > ACCEL= 58.827807 > > > ACCEL= 60.704297 > > > ACCEL= 60.619439 > > > ACCEL= 60.255062 > > > ACCEL= 59.246636 > > > ACCEL= 58.966244 > > > ACCEL= 62.119289 > > > ACCEL= 61.110753 > > > ACCEL= 64.018676 > > > ACCEL= 63.911028 > > > ACCEL= 60.811810 > > > ACCEL= 59.177635 > > > ACCEL= 61.196853 > > > ACCEL= 58.680083 > > > ACCEL= 59.527456 > > > ACCEL= 58.087839 > > > ACCEL= 59.119465 > > > ACCEL= 62.853732 > > > ACCEL= 59.093458 > > > ACCEL= 59.161929 > > > ACCEL= 61.429357 > > > ACCEL= 58.803188 > > > ACCEL= 59.175222 > > > ACCEL= 60.219244 > > > ACCEL= 58.596575 > > > ACCEL= 58.764846 > > > ACCEL= 58.426363 > > > ACCEL= 60.086656 > > > ACCEL= 59.610107 > > > ACCEL= 59.229815 > > > ACCEL= 58.044244 > > > ACCEL= 63.695621 > > > ACCEL= 58.131310 > > > ACCEL= 59.107058 > > > ACCEL= 58.927665 > > > ACCEL= 61.165485 > > > ACCEL= 58.315186 > > > ACCEL= 60.871648 > > > ACCEL= 60.176458 > > > ACCEL= 59.214217 > > > ACCEL= 59.208457 > > > ACCEL= 62.621824 > > > ACCEL= 58.398335 > > > ACCEL= 61.152436 > > > ACCEL= 58.604786 > > > ACCEL= 58.729217 > > > ACCEL= 59.923425 > > > ACCEL= 59.677849 > > > ACCEL= 58.123086 > > > ACCEL= 60.433701 > > > ACCEL= 58.291893 > > > ACCEL= 60.252449 > > > ACCEL= 60.417103 > > > ACCEL= 60.433561 > > > ACCEL= 61.127250 > > > ACCEL= 61.402805 > > > ACCEL= 58.202147 > > > ACCEL= 62.299667 > > > ACCEL= 62.973329 > > > ACCEL= 61.678133 > > > ACCEL= 58.407879 > > > ACCEL= 59.317560 > > > ACCEL= 58.968064 > > > ACCEL= 58.048711 > > > ACCEL= 58.288685 > > > ACCEL= 58.241909 > > > ACCEL= 59.930745 > > > ACCEL= 59.450933 > > > ACCEL= 59.439070 > > > ACCEL= 58.143918 > > > ACCEL= 58.259981 > > > ACCEL= 58.482531 > > > ACCEL= 58.101452 > > > ACCEL= 60.411654 > > > ACCEL= 60.631439 > > > ACCEL= 62.469302 > > > ACCEL= 60.847604 > > > ACCEL= 58.350368 > > > *ACCEL= 1166.694794 > > > ACCEL= 450.118253* > > > ACCEL= 186.931892 > > > ACCEL= 91.714558 > > > ACCEL= 60.190257 > > > ACCEL= 60.530275 > > > ACCEL= 59.028122 > > > > > > Here I stopped program line was N463261. > > > Smoothing TAU 0.01 > > > > > > In attechement INIT.c config program. > > > GCode program is in Yahoo Files folder "Albert CNC > > > files " (test_tolerance_0.005.part01.rar > > > and test_tolerance_0.005.part02.rar) > > > > > > Best Regards > > > Albert Platek > > > > > > > > > 2012/2/7 Tom Kerekes <tk@> > > > > > > > ** > > > > > > > > [Attachment(s)<https://mail.google.com/mail/html/compose/static_files/blank_quirks.html#1355587bcb0eb6ec_TopText>from Tom Kerekes included below] > > > > > > > > Hi Albert, > > > > > > > > Your program seems reasonable but the numbers are crazy. > > > > > > > > So are you saying that it did not report any errors until after 10 minutes > > > > then printed those? > > > > > > > > I made a similar program and ran it over night and didn't see such a > > > > problem. See attached. It displays the velocity and acceleration into > > > > unused axes 6 and 7 Destinations on the Axis Screen. It also captures the > > > > Max Velocity and Acceleration and displays those as 6 and 7 Positions. > > > > > > > > To be honest I see some tiny glitches and anomolies probably due to > > > > numerical rounding that I would like to better understand, but with the > > > > smoothing filter of 8ms it attenuates them way down. The biggest > > > > acceleration spike I see in the entire multi hour job is 90 in/sec2. > > > > > > > > Did you have the smoothing turned on? > > > > > > > > Is your Resolution 254,000 steps per inch? > > > > > > > > You don't have an MPG connected do you? > > > > > > > > I will try your program to see if I get similar results. > > > > > > > > Thanks > > > > TK > > > > > > > > *From:* albertplatek <albertplatek86@> > > > > *To:* DynoMotion@yahoogroups.com > > > > *Sent:* Monday, February 6, 2012 8:24 AM > > > > *Subject:* [DynoMotion] Re: Coord buffer overflow again > > > > > > > > > > > > Hi Tom, > > > > > > > > Sorry Tom but still this happens randomly :/ I was just lucky that after > > > > firmware flash 1 job was ok. > > > > I use ver 4.28. > > > > > > > > I dont know how to determine line numner of this jerk, but you can see > > > > huge acceleration! X motor stalled. > > > > It was runinng about 10min from program start. > > > > Acceleration was set 32in/sec2 > > > > Break Angle 30 > > > > Tau 0.008 > > > > Collinear Tolerance 0.0005 > > > > Corener Tolerance 0.01 > > > > Facet angle 0.5 > > > > Vel 4in/sec > > > > > > > > Here is my simple C program to check acceleration please check if it is > > > > correct: > > > > > > > > #include "KMotionDef.h" > > > > #define channel ch0 > > > > #define time 0.00018 > > > > main() > > > > { > > > > double X1,X2,speed1,speed2,accel; > > > > int k; > > > > > > > > while(1) > > > > { > > > > for (k=0; k<1; k++) WaitNextTimeSlice(); > > > > X1 = channel->Dest; > > > > for (k=0; k<1; k++) WaitNextTimeSlice(); > > > > X2 = channel->Dest; > > > > speed1=(X2-X1)/(time); > > > > > > > > for (k=0; k<1; k++) WaitNextTimeSlice(); > > > > X1 = channel->Dest; > > > > for (k=0; k<1; k++) WaitNextTimeSlice(); > > > > X2 = channel->Dest; > > > > speed2=(X2-X1)/(time); > > > > > > > > accel=(speed1-speed2)/(time); > > > > accel=(accel/254000)/2; > > > > if(accel>45) > > > > { > > > > printf("AccelError= %f\n",accel); > > > > } > > > > } > > > > } > > > > > > > > Here log: > > > > > > > > AccelError= 45.009219 > > > > AccelError= 45.229875 > > > > AccelError= 46.294106 > > > > AccelError= 45.055529 > > > > AccelError= 45.747547 > > > > AccelError= 45.070729 > > > > AccelError= 45.360541 > > > > AccelError= 45.765590 > > > > AccelError= 46.302076 > > > > AccelError= 46.775562 > > > > AccelError= 47.051896 > > > > AccelError= 45.225101 > > > > AccelError= 45.730674 > > > > AccelError= 46.758273 > > > > AccelError= 46.838768 > > > > AccelError= 45.698572 > > > > AccelError= 46.918682 > > > > AccelError= 45.204219 > > > > AccelError= 45.900150 > > > > AccelError= 47.023519 > > > > AccelError= 47.551550 > > > > AccelError= 48.022541 > > > > AccelError= 46.395295 > > > > AccelError= 81526.314840 > > > > AccelError= 72853.056587 > > > > AccelError= 65101.872298 > > > > AccelError= 58175.992747 > > > > AccelError= 51986.611437 > > > > AccelError= 46456.046360 > > > > AccelError= 41513.718492 > > > > AccelError= 37097.300673 > > > > AccelError= 33150.797142 > > > > AccelError= 29622.770379 > > > > AccelError= 26468.350534 > > > > AccelError= 23652.797849 > > > > AccelError= 21136.323622 > > > > AccelError= 18888.213633 > > > > AccelError= 16879.065077 > > > > AccelError= 15083.474245 > > > > AccelError= 13479.186003 > > > > AccelError= 12045.566722 > > > > AccelError= 10764.204608 > > > > AccelError= 9619.483512 > > > > AccelError= 8596.141212 > > > > AccelError= 7682.015002 > > > > AccelError= 6866.662006 > > > > AccelError= 6139.961139 > > > > AccelError= 5490.446488 > > > > AccelError= 4910.042286 > > > > AccelError= 4391.574405 > > > > AccelError= 3928.114510 > > > > AccelError= 3514.000485 > > > > AccelError= 3144.002655 > > > > AccelError= 2813.316136 > > > > AccelError= 2517.500643 > > > > AccelError= 2253.490792 > > > > AccelError= 2017.432428 > > > > AccelError= 1806.266152 > > > > AccelError= 1617.706032 > > > > AccelError= 1448.982225 > > > > AccelError= 1298.398178 > > > > AccelError= 1163.621865 > > > > AccelError= 1043.472200 > > > > AccelError= 935.821407 > > > > AccelError= 839.766485 > > > > AccelError= 753.863074 > > > > AccelError= 677.101783 > > > > AccelError= 608.587862 > > > > AccelError= 547.209083 > > > > AccelError= 492.589836 > > > > AccelError= 443.554631 > > > > AccelError= 399.735924 > > > > AccelError= 360.639632 > > > > AccelError= 325.643422 > > > > AccelError= 294.370902 > > > > AccelError= 266.425861 > > > > AccelError= 241.478288 > > > > AccelError= 219.191672 > > > > AccelError= 199.246085 > > > > AccelError= 181.422761 > > > > AccelError= 165.532090 > > > > AccelError= 151.295897 > > > > AccelError= 138.574472 > > > > AccelError= 127.206640 > > > > AccelError= 117.084135 > > > > AccelError= 108.003075 > > > > AccelError= 99.888264 > > > > AccelError= 92.598183 > > > > AccelError= 86.122270 > > > > AccelError= 76.770777 > > > > AccelError= 65.225363 > > > > AccelError= 54.908134 > > > > AccelError= 45.688617 > > > > AccelError= 45.070435 > > > > AccelError= 45.411744 > > > > AccelError= 45.770036 > > > > AccelError= 45.871485 > > > > AccelError= 45.239885 > > > > AccelError= 45.631836 > > > > AccelError= 45.001237 > > > > AccelError= 45.373033 > > > > AccelError= 45.577907 > > > > AccelError= 46.322608 > > > > AccelError= 46.958820 > > > > AccelError= 46.038556 > > > > AccelError= 45.018088 > > > > > > > > If you test it on your PC and will be ok, then maybe something is wrong > > > > with my laptop. Then I will test on other PC. > > > > > > > > Best Regards > > > > Albert P³atek > > > > > > > > --- In DynoMotion@yahoogroups.com, "albertplatek" <albertplatek86@> > > > > wrote: > > > > > > > > > > Hi Tom, > > > > > > > > > > Problem solved (probably), weird but after re-flash firmware 4.28 > > > > > everything is working ok, no axis stall. > > > > > > > > > > btw. Experimental version KMotion429x.exe is not working, when I start > > > > > execute gcode I get message "G Code Error GCode Aborted", this error > > > > > occurs on any gcode program even on only few G01 lines. > > > > > > > > > > Tom if you have C program that watch for too high of acceleration > > > > > please post it. If again in future something strange will > > > > > happen I will try find out the problem. > > > > > I am still not sure what was the problem in my system that axis > > > > > were stalling. > > > > > > > > > > Best Regards > > > > > Albert P³atek > > > > > > > > > > > > > > > > > > > > --- In DynoMotion@yahoogroups.com, Albert P³atek <albertplatek86@> > > > > wrote: > > > > > > > > > > > > Hi Tom, > > > > > > > > > > > > Thank you for help. > > > > > > Mostly stall X axis, and sometimes Z. > > > > > > Y axis seem to be ok. > > > > > > I uploaded file as an attachment (zip). > > > > > > > > > > > > Best Regards > > > > > > Albert Platek > > > > > > > > > > > > 2012/2/5 Tom Kerekes <tk@> > > > > > > > > > > > > > ** > > > > > > > > > > > > > > > > > > > > > Hi Albert, > > > > > > > > > > > > > > Thanks for the very complete problem description, I can't tell you > > > > how > > > > > > > much I appreciate that, but I can't think of anything that would > > > > cause > > > > > > > this. Anything with large files I might expect some processing issue > > > > but > > > > > > > that would result in a buffer underflow and an error message not > > > > incorrect > > > > > > > motion that would result in a stall. That delay I described in your > > > > email > > > > > > > only happens when "Execute" is pushed so I don't think it is related. > > > > > > > > > > > > > > There is an experimental version KMotion429x.exe on the download site > > > > > > > that has this delay eliminated and also faster USB throughput, but I > > > > don't > > > > > > > think either of those are related to this problem. > > > > > > > > > > > > > > Can you post the file somewhere? I'm thinking to write a C program > > > > that > > > > > > > would run in KFLOP and watch for too high of acceleration or > > > > velocity by > > > > > > > monitoring how ch0->Dest changes. If it detects something bad that > > > > could > > > > > > > be used as a trigger to help us track it down. > > > > > > > > > > > > > > Which axes loose position? > > > > > > > > > > > > > > Regards > > > > > > > TK > > > > > > > > > > > > > > > > > > > > > *From:* albertplatek <albertplatek86@> > > > > > > > *To:* DynoMotion@yahoogroups.com > > > > > > > *Sent:* Sunday, February 5, 2012 5:27 AM > > > > > > > > > > > > > > *Subject:* [DynoMotion] Re: Coord buffer overflow again > > > > > > > > > > > > > > > > > > > > > Hi Tom, > > > > > > > > > > > > > > My settings: > > > > > > > Break angle 30 deg > > > > > > > Look ahead 3 sec > > > > > > > Collinear tolerance 0.001 in > > > > > > > Corner tolerance 0.003 in > > > > > > > Facet angle 1.5 deg > > > > > > > Tau = 0.008 sec > > > > > > > XYZ Accelarations 40 in/sec2 > > > > > > > XYZ Velocity 4.5 in/sec > > > > > > > In gcode file I use max Vel 270 in/min (4.5 in/sec) > > > > > > > > > > > > > > I have strange problem, everything was ok until I started machining > > > > > > > bigger reliefs. > > > > > > > At half of the job my stepper motors lost position > > > > > > > probably becouse of high jerk of acceleration. > > > > > > > Firstly I thought that maybe axis accelerations for my system are to > > > > > > > high(40in/sec2). So I changed it to 20in/sec2 for test. > > > > > > > But didnt help, still stepper motors stall at half of the gcode file. > > > > > > > > > > > > > > This file contain 548389 lines, my steppers stall between 226000 - > > > > 274000 > > > > > > > line but not always in same place. > > > > > > > > > > > > > > I tried to find line where the problem occurs, but when I found line > > > > with > > > > > > > problem and tried to execute 500 lines before, it was ok. Motors > > > > didnt > > > > > > > stall. > > > > > > > I found that motors stall when KMotionCNC start execute at least > > > > > > > 100 000 lines before line number 274000. > > > > > > > > > > > > > > The problem disappear when I set break angle 5deg. > > > > > > > > > > > > > > "This diagnostic pointed out a side problem I hadn't noticed before. > > > > There > > > > > > > is an annoying ~30 second delay when it runs the second time. The > > > > > > > Interpreter keeps a record of all its state changes which allow > > > > Halts to go > > > > > > > back to any previous state. In this test of 100x500 subroutine calls > > > > > > > results in about 1 million state changes. Ironically creating the > > > > data, > > > > > > > making use of the data all happens in the blink of an eye, but > > > > discarding > > > > > > > the data (CList.RemoveAll) takes 30 seconds! We will find a better > > > > way in > > > > > > > the next version. For now just don't be surprised if there is an > > > > annoying > > > > > > > delay on a second run of a big job." > > > > > > > > > > > > > > Can this "delay" make that my stepper motors stall? > > > > > > > > > > > > > > Best Regards > > > > > > > Albert P³atek > > > > > > > > > > > > > > --- In DynoMotion@yahoogroups.com, Albert <albertplatek86@> wrote: > > > > > > > > > > > > > > > > Hi Tom, > > > > > > > > > > > > > > > > Thank you. > > > > > > > > Btw. I am trying to find out what changes make > > > > > > > > Fidelizer in Windows. If I will find something > > > > > > > > important I will let you know. > > > > > > > > Best regards > > > > > > > > Albert Platek > > > > > > > > > > > > > > > > > > > > > > > > Dnia 2012-01-18 o godz. 23:47 Tom Kerekes <tk@> napisaÃÆ'…‚ > > > > > > > > (a): > > > > > > > > > > > > > > > > > Hi Albert, > > > > > > > > > > > > > > > > > > Just wanted to mention that I now do notice an improvement with > > > > > > > > > Fidelizer. I don't understand why I didn't before. I might have > > > > > > > > > made a mistake or something changed. > > > > > > > > > > > > > > > > > > I sent an email to FTDI (USB chip we use) regarding it and the > > > > USB > > > > > > > > > 3.0 port issue to see if they have any answers. I figure it is > > > > > > > > > worth the cost of an email :} > > > > > > > > > > > > > > > > > > Still working on some optimization... > > > > > > > > > > > > > > > > > > Regards > > > > > > > > > TK > > > > > > > > > > > > > > > > > > > > > > > > > > > From: albertplatek <albertplatek86@> > > > > > > > > > To: DynoMotion@yahoogroups.com > > > > > > > > > Sent: Monday, January 16, 2012 12:01 PM > > > > > > > > > Subject: [DynoMotion] Re: Coord buffer overflow again > > > > > > > > > > > > > > > > > > > > > > > > > > > Hi Tom, > > > > > > > > > Thank you for making changes in new version to increase the > > > > segment > > > > > > > > > rate. > > > > > > > > > > > > > > > > > > Btw: > > > > > > > > > I installed on second pc Windows 7 (Core2Duo cpu), and aslo > > > > tested > > > > > > > > > with fidelizer. Strange but didn't make difference, as in your > > > > > > > > > computer. > > > > > > > > > > > > > > > > > > I cant get why in this particular case (pc with WIN7 i7 cpu) it > > > > helps. > > > > > > > > > Maybe it depends on type of host controller... > > > > > > > > > > > > > > > > > > Pc with WIN7 64bit (Core2Duo) has > > > > > > > > > SiS 7001 PCI to USB Open Host Controller > > > > > > > > > > > > > > > > > > Pc with WIN7 64bit (i7) has > > > > > > > > > Intel(R) 6 Series/C200 Series Chipset Family USB Enhanced Host > > > > > > > > > Controller > > > > > > > > > > > > > > > > > > This computer has aslo another usb ports: Renesas Electronics USB > > > > > > > > > 3.0 Host Controller > > > > > > > > > which works very bad with KFlop. And on this port fidelizer make > > > > no > > > > > > > > > difference. > > > > > > > > > > > > > > > > > > Best Regards > > > > > > > > > Albert Platek > > > > > > > > > > > > > > > > > > --- In DynoMotion@yahoogroups.com, Tom Kerekes <tk@> wrote: > > > > > > > > > > > > > > > > > > > > Hi Albert, > > > > > > > > > > ÃÆ'Æ'‚ > > > > > > > > > > Thanks for that interesting information. > > > > > > > > > > ÃÆ'Æ'‚ > > > > > > > > > > I tried the Fidelizer on my i7 W7 32 machine and it didn't > > > > seem to > > > > > > > > > make any difference. > > > > > > > > > > ÃÆ'Æ'‚ > > > > > > > > > > I'm trying some things to increase the segment rate by sending > > > > > > > > > blocks of segments and optimizing how they are handled in > > > > KFLOP.ÃÆ'Æ'‚ I > > > > > > > > > f this bears fruit I'll let you know. > > > > > > > > > > ÃÆ'Æ'‚ > > > > > > > > > > This diagnostic pointed out a side problem I hadn't noticed > > > > before > > > > > > > > > .ÃÆ'Æ'‚ There is an annoying ~30 second delay when it runs the > > > > second ti > > > > > > > > > me.ÃÆ'Æ'‚ The Interpreter keeps a record of all its state changes > > > > which > > > > > > > > > allow Halts to go back to any previous state.ÃÆ'Æ'‚ In this test > > > > of 100x > > > > > > > > > 500 subroutine calls results in about 1 million state > > > > changes.ÃÆ'Æ'‚ Iro > > > > > > > > > nically creating the data, making use of the data all happens in > > > > the > > > > > > > > > blink of an eye, but discarding the data (CList.RemoveAll) takes > > > > 30 > > > > > > > > > seconds!ÃÆ'Æ'‚ We will find a better way in the next > > > > version.ÃÆ'Æ'‚ For now > > > > > > > > > just don't be surprisedÃÆ'Æ'‚ if there is an annoying delay on a > > > > second > > > > > > > > > run of a big job. > > > > > > > > > > ÃÆ'Æ'‚ > > > > > > > > > > Regards > > > > > > > > > > TK > > > > > > > > > > ÃÆ'Æ'‚ > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > ________________________________ > > > > > > > > > > From: albertplatek <albertplatek86@> > > > > > > > > > > To: DynoMotion@yahoogroups.com > > > > > > > > > > Sent: Sunday, January 15, 2012 8:37 AM > > > > > > > > > > Subject: [DynoMotion] Re: Coord buffer overflow again > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > ÃÆ'Æ'‚ > > > > > > > > > > Hi Tom, > > > > > > > > > > First thank you Tom for the test file. > > > > > > > > > > I made few tests and I got interesting results :) > > > > > > > > > > > > > > > > > > > > PC No1: > > > > > > > > > > > > > > > > > > > > Intel Pentium 4 2.6Ghz WinXP 32bit port USB2.0 > > > > > > > > > > Max segments: 110ÃÆ'Æ'‚ > > > > > > > > > > Time: 31s > > > > > > > > > > 100 circles x 110 segments/circle / 31 seconds =354 segments > > > > per > > > > > > > > > second. > > > > > > > > > > > > > > > > > > > > PC No2: > > > > > > > > > > > > > > > > > > > > Intel Core2Duo 2.0 Ghz WinXP 32bit port USB2.0 > > > > > > > > > > Max segments: 140ÃÆ'Æ'‚ > > > > > > > > > > Time: 31s > > > > > > > > > > 100 circles x 140 segments/circle / 31 seconds = 451 segments > > > > per > > > > > > > > > second. > > > > > > > > > > > > > > > > > > > > PC No3: > > > > > > > > > > > > > > > > > > > > i7 2670QM Win7 64bit port USB3.0 > > > > > > > > > > Max segments:30ÃÆ'Æ'‚ > > > > > > > > > > Time:31s > > > > > > > > > > 100 circles x 60 segments/circle / 31 seconds = 193 segments > > > > per > > > > > > > > > second. > > > > > > > > > > (thats why I had problem here, usb 3.0 fail) > > > > > > > > > > > > > > > > > > > > (Luckily I had 1 USB 2.0 port) > > > > > > > > > > > > > > > > > > > > i7 2670QM Win7 64bit port USB2.0 > > > > > > > > > > Max segments:170 > > > > > > > > > > Time:31s > > > > > > > > > > 100 circles x 170 segments/circle / 31 seconds = 548 segments > > > > per > > > > > > > > > second. > > > > > > > > > > > > > > > > > > > > i7 2670QM Win7 64bit port USB2.0 (with 1 core affinity for > > > > > > > > > KmotionCNC and KmotionServer) > > > > > > > > > > Max segments:200 > > > > > > > > > > Time:31s > > > > > > > > > > 100 circles x 200 segments/circle / 31 seconds =645 segments > > > > per > > > > > > > > > second. > > > > > > > > > > > > > > > > > > > > I found on net some interesting software called Fidelizer, its > > > > > > > > > software to optimize win7/vista for audio workstation, but it > > > > aslo > > > > > > > > > tweak somehow usb throughput. > > > > > > > > > > More detail here: /www.windowsxlive.net/fidelizer > > > > > > > > > > > > > > > > > > > > i7 2670QM Win7 64bit port USB2.0 (Fidelizer) > > > > > > > > > > Max segments:410 > > > > > > > > > > Time:31s > > > > > > > > > > 100 circles x 410 segments/circle / 31 seconds = 1320 segments > > > > per > > > > > > > > > second. (wow! tested few times with same result) > > > > > > > > > > > > > > > > > > > > Be aware of new USB 3.0 ports they may not work with good > > > > > > > > > performance with KFlop, maybe becouse of bug in drivers. Windows > > > > 7 > > > > > > > > > dont see them without additional drivers. > > > > > > > > > > Fidelizer software only improved performance on pc with ÃÆ'Æ'‚ > > > > Win7. > > > > > > > > > > On WinXP seem to not work. I will soon try install on "PC No2" > > > > > > > > > Windows 7 and then try fidelizer also. > > > > > > > > > > > > > > > > > > > > Best regards > > > > > > > > > > Albert PÃÆ'Æ'‚ÃÆ'‚³atek > > > > > > > > > > --- In DynoMotion@yahoogroups.com, Tom Kerekes <tk@> wrote: > > > > > > > > > > > > > > > > > > > > > > Hi Albert, > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'‚ > > > > > > > > > > > Your application/system is quite demanding :} > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'‚ > > > > > > > > > > > That portion of GCode you sent is 100,000 lines of all G1 > > > > > > > > > movements with typical tiny lengths of only > > > > 50~150um.ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'‚ The > > > > > > > > > collinear tolerance of 0.001 (25um) will cause many > > > > blocksÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'‚ to > > > > > > > > > be combined together so the real required vector rate will be > > > > more d > > > > > > > > > ependent on the path curvature rather than the original > > > > GCode.ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'‚ > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'‚ > > > > > > > > > > > On my system (i7 W7 32-bit) it runs ok with a facet angle 0f > > > > 0.4 > > > > > > > > > degrees butÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'‚ can't keep up at 0.3 degrees. > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'‚ > > > > > > > > > > > I did some calculations for a system like yours with 40 > > > > in/sec2 > > > > > > > > > acceleration.ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'‚ Here is what I find: > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'‚ > > > > > > > > > > > That acceleration allows going around a 0.1 inch radius > > > > circle > > > > > > > > > at 120ipm (which is about 3 circles per second) > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'‚ > > > > > > > > > > > With a facet angle setting of 1.5 degrees that requires 240 > > > > > > > > > segments per circle which would require 720 segments per second > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'‚ > > > > > > > > > > > I created a diagnostic test GCode file that programmatically > > > > > > > > > generates an N sided Polygon 100 timesÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'‚ with > > > > radius 0.1 inches > > > > > > > > > and allows you to specify the number of polygon sides (it also > > > > gradu > > > > > > > > > ally shifts the "circle" to make the plot more > > > > interesting).ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'‚ > > > > > > > > > (see attachment). ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'‚ When using the diagnostic > > > > make sure to set > > > > > > > > > the Collinear tolerance to zero and the Corner Rounding to zero > > > > so t > > > > > > > > > he original segments are used.ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'‚ By changing N > > > > (number of sides > > > > > > > > > we can determine the segment download rate). > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'‚ > > > > > > > > > > > For example on my system it works with N = 200 and fails > > > > with N > > > > > > > > > = 240.ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'‚ The total time to do all 100 circles > > > > takes about 30 > > > > > > > > > seconds.ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'‚ So: > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'‚ > > > > > > > > > > > 100 circles x 200 segments/circle / 30 seconds = 666 segments > > > > > > > > > per second. > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'‚ > > > > > > > > > > > In simulation mode the Interpreter+Graphics finishes the job > > > > in > > > > > > > > > ~ 5 seconds so it isn't a significant factor > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'‚ > > > > > > > > > > > With the GCode Window closed it simulates in ~ 4 seconds so > > > > the > > > > > > > > > Graphical Display is not aÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'‚ significant factor. > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'‚ > > > > > > > > > > > You might run the tests on your two systems to see if you get > > > > > > > > > drastically different results. > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'‚ > > > > > > > > > > > But to try to answer your original question of why one CPU > > > > would > > > > > > > > > be significantly different I don't know.ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'‚ Google > > > > finds some > > > > > > > > > benchmarking tools but they areÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'‚ mainly for High > > > > Speed USB > > > > > > > > > devices like flash drives.ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'‚ Some sites state if > > > > an older USB > > > > > > > > > 1.1 device is attached it slows everything down.ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'‚ > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'‚ > > > > > > > > > > > Regards > > > > > > > > > > > TKÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'‚ > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'‚ > > > > > > > > > > > > > > > > > > > > > > From: albertplatek albertplatek86@ > > > > > > > > > > > To: DynoMotion@yahoogroups.com > > > > > > > > > > > Sent: Saturday, January 14, 2012 11:02 AM > > > > > > > > > > > Subject: [DynoMotion] Re: Coord buffer overflow again > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'‚ > > > > > > > > > > > Hi Tom, > > > > > > > > > > > How to determine USB throughput on pc? or what pc/system is > > > > best > > > > > > > > > for KFlop? > > > > > > > > > > > > > > > > > > > > > > I have still sometimes cord buffer overflow on second pc > > > > even I > > > > > > > > > > > set lower accelaration 25 and bigger facet angle 2.5 it > > > > sometimes > > > > > > > > > > > overflow. Its very strange pc has windows xp clean > > > > installation, > > > > > > > > > > > Integrated Intel Graphics. > > > > > > > > > > > On first pc works perfect even with high acceleration 40, > > (Message over 64 KB, truncated) Group: DynoMotion Message: 3828 From: albertplatek Date: 2/9/2012 Subject: Re: Coord buffer overflow again Hi Brad, Thanks I will try it. Best Regards Albert Platek --- In DynoMotion@yahoogroups.com, Brad Murry <bradodarb@...> wrote: > > Hello Albert, > > > > You might consider a free trial of http://www.usblyzer.com/download.htm if you have some extra time on your hands. > > > > I just used it to find a traffic issue and it worked well. > > > > It might show some evidence of USB cross talk or something else weird going on, maybe worth a shot. > > > > -Brad Murry > > > > From: DynoMotion@yahoogroups.com [mailto:DynoMotion@yahoogroups.com] On Behalf Of albertplatek > Sent: Thursday, February 09, 2012 12:04 PM > To: DynoMotion@yahoogroups.com > Subject: [DynoMotion] Re: Coord buffer overflow again > > > > > > Hi Tom, > I made test with gamepad conencted to pc, and after 20min I got this > jerk of acceleration. > After this I again made test without gamepad and it was working fine. > > Gamepad is not important for me, so I will not use it in future. > > Best Regards > Albert P³atek > > --- In DynoMotion@yahoogroups.com <mailto:DynoMotion%40yahoogroups.com> , "albertplatek" <albertplatek86@> wrote: > > > > Hi Tom, > > "Have you tried re-connecting the GamPad and pushing the buttons to see if you can force errors?" > > - no nothing happen, accelerations are normal. > > I am happy that it is finally working and I finished this big reliefs. I can test tomorrow same gcode with gamepad connected to pc and check if it will happen again. > > > > Best Regards > > Albert P³atek > > > > > > --- In DynoMotion@yahoogroups.com <mailto:DynoMotion%40yahoogroups.com> , Tom Kerekes <tk@> wrote: > > > > > > Hi Albert, > > >  > > > Wow that is strange. I just double checked and V4.28 is in fact testing the Gampad enable option. So I have no explaination how the Gampad could make a difference. Possibly it somehow corrupts the USB communication - but that should generate an error. Maybe it causes Coordinated Motion Segments to just get completely lost (skipped). > > >  > > > That is something I couldn't understand about the problem, there should be no randomness. Everything we are doing should be completely deterministic. The same GCode and the same Settings should generate the exact same motion every time. > > >  > > > Have you tried re-connecting the GamPad and pushing the buttons to see if you can force errors? > > > > > > Regards > > > TK > > >  > > >  > > > > > >  > > > Hi Tom, > > > > > > I think that in 99% I found bug, it was gamepad connected into usb. > > > Even if it was disabled in KMmotionCNC, it was causing randomly this > > > crazy jerk of acceleration. Today I tested this big gcode program 3 times without gamepad and everything is fine! :) > > > > > > Best Regards > > > Albert P³atek > > > > > > --- In DynoMotion@yahoogroups.com <mailto:DynoMotion%40yahoogroups.com> , Albert P³atek <albertplatek86@> wrote: > > > > > > > > Hi Tom, > > > > "So are you saying that it did not report any errors until after 10 > > > > minutes then printed those?" > > > > - It was printing accel peak around 60in/sec2 sometimes from start and > > > > after 10 min this big error. Second job execute was ok by about 20min then > > > > again error. > > > > Smoothing was Tau= 0.008. > > > > > > > > > > > > I just made new test on other pc with file "test_tolerance_0.005.ngc" . And > > > > I got *1166in/sec2* acceleration peak... and it stalled X motor. > > > > > > > > It is same program as before but with toolpath tolerance of 0.005. I > > > > thought that maybe this will help. But it didnt. (old file Relief14.ngc had > > > > toolpath tolerance of 0.01) > > > > Small programs works even with extream acceleration of 60, but this program > > > > and other bigger programs fail even with acceleration of 20. > > > > > > > > Here all my settings: > > > > > > > > *Trajectory planner* > > > > > > > > Break Angle 30 degrees > > > > Look Ahead 3 sec > > > > Collinear Tolerance 0.0005 in > > > > Corner Tolerance 0.01 in > > > > Facet Angle 1.5 > > > > > > > > *Axis Parameters* > > > > > > > > X Resolution 254000 Vel 5 Accel 40 > > > > Y Resolution 127000 Vel 4 Accel 10 > > > > Z Resolution 254000 Vel 5 Accel 40 > > > > A Resolution 7055.5555 Vel 100 Accel 15 (I dont use this axis in programs > > > > at moment) > > > > > > > > Smoothing Tau = 0.01 > > > > > > > > Gamepad disabled > > > > Lathe disabled > > > > I dont use MPG and it is not connected. > > > > > > > > Gcode program (test_tolerance_0.005.ngc) started from first line, I was > > > > only capturing Acceleration > 58. > > > > > > > > Wt, lut 07, 2012, 12:42:31 KMotion Program Started > > > > TAU = 0.010000 > > > > ACCEL= 58.677936 > > > > ACCEL= 58.927802 > > > > ACCEL= 58.960033 > > > > ACCEL= 60.769580 > > > > ACCEL= 58.512884 > > > > ACCEL= 58.924281 > > > > ACCEL= 59.024541 > > > > ACCEL= 58.133892 > > > > ACCEL= 61.956729 > > > > ACCEL= 59.006616 > > > > ACCEL= 60.105317 > > > > ACCEL= 58.828274 > > > > ACCEL= 60.622193 > > > > ACCEL= 58.060298 > > > > ACCEL= 60.837834 > > > > ACCEL= 58.185082 > > > > ACCEL= 58.925759 > > > > ACCEL= 59.444038 > > > > ACCEL= 61.017190 > > > > ACCEL= 59.082489 > > > > ACCEL= 60.176545 > > > > ACCEL= 58.677989 > > > > ACCEL= 60.129486 > > > > ACCEL= 59.890981 > > > > ACCEL= 63.703736 > > > > ACCEL= 58.300426 > > > > ACCEL= 58.390017 > > > > ACCEL= 64.994889 > > > > ACCEL= 61.344673 > > > > ACCEL= 61.264344 > > > > ACCEL= 60.206659 > > > > ACCEL= 61.110118 > > > > ACCEL= 58.861575 > > > > ACCEL= 62.844393 > > > > ACCEL= 58.712601 > > > > ACCEL= 62.743716 > > > > ACCEL= 59.412413 > > > > ACCEL= 59.090724 > > > > ACCEL= 59.607079 > > > > ACCEL= 58.240991 > > > > ACCEL= 58.200542 > > > > ACCEL= 62.400201 > > > > ACCEL= 64.401300 > > > > ACCEL= 60.829834 > > > > ACCEL= 60.765056 > > > > ACCEL= 58.584326 > > > > ACCEL= 59.368296 > > > > ACCEL= 58.845166 > > > > ACCEL= 63.692201 > > > > ACCEL= 59.398388 > > > > ACCEL= 58.492754 > > > > ACCEL= 58.201424 > > > > ACCEL= 59.713704 > > > > ACCEL= 59.488310 > > > > ACCEL= 58.663588 > > > > ACCEL= 62.887233 > > > > ACCEL= 58.256548 > > > > ACCEL= 58.483484 > > > > ACCEL= 58.042246 > > > > ACCEL= 58.518015 > > > > ACCEL= 61.312055 > > > > ACCEL= 60.224222 > > > > ACCEL= 60.041371 > > > > ACCEL= 61.019005 > > > > ACCEL= 59.857207 > > > > ACCEL= 60.667182 > > > > ACCEL= 62.517183 > > > > ACCEL= 62.978943 > > > > ACCEL= 60.672158 > > > > ACCEL= 62.764637 > > > > ACCEL= 60.755186 > > > > ACCEL= 58.177374 > > > > ACCEL= 58.144752 > > > > ACCEL= 62.349561 > > > > ACCEL= 59.094084 > > > > ACCEL= 59.707611 > > > > ACCEL= 60.042771 > > > > ACCEL= 59.493590 > > > > ACCEL= 58.381740 > > > > ACCEL= 58.552432 > > > > ACCEL= 60.385809 > > > > ACCEL= 61.290063 > > > > ACCEL= 63.347741 > > > > ACCEL= 59.724964 > > > > ACCEL= 66.035414 > > > > ACCEL= 58.827807 > > > > ACCEL= 60.704297 > > > > ACCEL= 60.619439 > > > > ACCEL= 60.255062 > > > > ACCEL= 59.246636 > > > > ACCEL= 58.966244 > > > > ACCEL= 62.119289 > > > > ACCEL= 61.110753 > > > > ACCEL= 64.018676 > > > > ACCEL= 63.911028 > > > > ACCEL= 60.811810 > > > > ACCEL= 59.177635 > > > > ACCEL= 61.196853 > > > > ACCEL= 58.680083 > > > > ACCEL= 59.527456 > > > > ACCEL= 58.087839 > > > > ACCEL= 59.119465 > > > > ACCEL= 62.853732 > > > > ACCEL= 59.093458 > > > > ACCEL= 59.161929 > > > > ACCEL= 61.429357 > > > > ACCEL= 58.803188 > > > > ACCEL= 59.175222 > > > > ACCEL= 60.219244 > > > > ACCEL= 58.596575 > > > > ACCEL= 58.764846 > > > > ACCEL= 58.426363 > > > > ACCEL= 60.086656 > > > > ACCEL= 59.610107 > > > > ACCEL= 59.229815 > > > > ACCEL= 58.044244 > > > > ACCEL= 63.695621 > > > > ACCEL= 58.131310 > > > > ACCEL= 59.107058 > > > > ACCEL= 58.927665 > > > > ACCEL= 61.165485 > > > > ACCEL= 58.315186 > > > > ACCEL= 60.871648 > > > > ACCEL= 60.176458 > > > > ACCEL= 59.214217 > > > > ACCEL= 59.208457 > > > > ACCEL= 62.621824 > > > > ACCEL= 58.398335 > > > > ACCEL= 61.152436 > > > > ACCEL= 58.604786 > > > > ACCEL= 58.729217 > > > > ACCEL= 59.923425 > > > > ACCEL= 59.677849 > > > > ACCEL= 58.123086 > > > > ACCEL= 60.433701 > > > > ACCEL= 58.291893 > > > > ACCEL= 60.252449 > > > > ACCEL= 60.417103 > > > > ACCEL= 60.433561 > > > > ACCEL= 61.127250 > > > > ACCEL= 61.402805 > > > > ACCEL= 58.202147 > > > > ACCEL= 62.299667 > > > > ACCEL= 62.973329 > > > > ACCEL= 61.678133 > > > > ACCEL= 58.407879 > > > > ACCEL= 59.317560 > > > > ACCEL= 58.968064 > > > > ACCEL= 58.048711 > > > > ACCEL= 58.288685 > > > > ACCEL= 58.241909 > > > > ACCEL= 59.930745 > > > > ACCEL= 59.450933 > > > > ACCEL= 59.439070 > > > > ACCEL= 58.143918 > > > > ACCEL= 58.259981 > > > > ACCEL= 58.482531 > > > > ACCEL= 58.101452 > > > > ACCEL= 60.411654 > > > > ACCEL= 60.631439 > > > > ACCEL= 62.469302 > > > > ACCEL= 60.847604 > > > > ACCEL= 58.350368 > > > > *ACCEL= 1166.694794 > > > > ACCEL= 450.118253* > > > > ACCEL= 186.931892 > > > > ACCEL= 91.714558 > > > > ACCEL= 60.190257 > > > > ACCEL= 60.530275 > > > > ACCEL= 59.028122 > > > > > > > > Here I stopped program line was N463261. > > > > Smoothing TAU 0.01 > > > > > > > > In attechement INIT.c config program. > > > > GCode program is in Yahoo Files folder "Albert CNC > > > > files " (test_tolerance_0.005.part01.rar > > > > and test_tolerance_0.005.part02.rar) > > > > > > > > Best Regards > > > > Albert Platek > > > > > > > > > > > > 2012/2/7 Tom Kerekes <tk@> > > > > > > > > > ** > > > > > > > > > > [Attachment(s)<https://mail.google.com/mail/html/compose/static_files/blank_quirks.html#1355587bcb0eb6ec_TopText>from Tom Kerekes included below] > > > > > > > > > > Hi Albert, > > > > > > > > > > Your program seems reasonable but the numbers are crazy. > > > > > > > > > > So are you saying that it did not report any errors until after 10 minutes > > > > > then printed those? > > > > > > > > > > I made a similar program and ran it over night and didn't see such a > > > > > problem. See attached. It displays the velocity and acceleration into > > > > > unused axes 6 and 7 Destinations on the Axis Screen. It also captures the > > > > > Max Velocity and Acceleration and displays those as 6 and 7 Positions. > > > > > > > > > > To be honest I see some tiny glitches and anomolies probably due to > > > > > numerical rounding that I would like to better understand, but with the > > > > > smoothing filter of 8ms it attenuates them way down. The biggest > > > > > acceleration spike I see in the entire multi hour job is 90 in/sec2. > > > > > > > > > > Did you have the smoothing turned on? > > > > > > > > > > Is your Resolution 254,000 steps per inch? > > > > > > > > > > You don't have an MPG connected do you? > > > > > > > > > > I will try your program to see if I get similar results. > > > > > > > > > > Thanks > > > > > TK > > > > > > > > > > *From:* albertplatek <albertplatek86@> > > > > > *To:* DynoMotion@yahoogroups.com <mailto:DynoMotion%40yahoogroups.com> > > > > > *Sent:* Monday, February 6, 2012 8:24 AM > > > > > *Subject:* [DynoMotion] Re: Coord buffer overflow again > > > > > > > > > > > > > > > Hi Tom, > > > > > > > > > > Sorry Tom but still this happens randomly :/ I was just lucky that after > > > > > firmware flash 1 job was ok. > > > > > I use ver 4.28. > > > > > > > > > > I dont know how to determine line numner of this jerk, but you can see > > > > > huge acceleration! X motor stalled. > > > > > It was runinng about 10min from program start. > > > > > Acceleration was set 32in/sec2 > > > > > Break Angle 30 > > > > > Tau 0.008 > > > > > Collinear Tolerance 0.0005 > > > > > Corener Tolerance 0.01 > > > > > Facet angle 0.5 > > > > > Vel 4in/sec > > > > > > > > > > Here is my simple C program to check acceleration please check if it is > > > > > correct: > > > > > > > > > > #include "KMotionDef.h" > > > > > #define channel ch0 > > > > > #define time 0.00018 > > > > > main() > > > > > { > > > > > double X1,X2,speed1,speed2,accel; > > > > > int k; > > > > > > > > > > while(1) > > > > > { > > > > > for (k=0; k<1; k++) WaitNextTimeSlice(); > > > > > X1 = channel->Dest; > > > > > for (k=0; k<1; k++) WaitNextTimeSlice(); > > > > > X2 = channel->Dest; > > > > > speed1=(X2-X1)/(time); > > > > > > > > > > for (k=0; k<1; k++) WaitNextTimeSlice(); > > > > > X1 = channel->Dest; > > > > > for (k=0; k<1; k++) WaitNextTimeSlice(); > > > > > X2 = channel->Dest; > > > > > speed2=(X2-X1)/(time); > > > > > > > > > > accel=(speed1-speed2)/(time); > > > > > accel=(accel/254000)/2; > > > > > if(accel>45) > > > > > { > > > > > printf("AccelError= %f\n",accel); > > > > > } > > > > > } > > > > > } > > > > > > > > > > Here log: > > > > > > > > > > AccelError= 45.009219 > > > > > AccelError= 45.229875 > > > > > AccelError= 46.294106 > > > > > AccelError= 45.055529 > > > > > AccelError= 45.747547 > > > > > AccelError= 45.070729 > > > > > AccelError= 45.360541 > > > > > AccelError= 45.765590 > > > > > AccelError= 46.302076 > > > > > AccelError= 46.775562 > > > > > AccelError= 47.051896 > > > > > AccelError= 45.225101 > > > > > AccelError= 45.730674 > > > > > AccelError= 46.758273 > > > > > AccelError= 46.838768 > > > > > AccelError= 45.698572 > > > > > AccelError= 46.918682 > > > > > AccelError= 45.204219 > > > > > AccelError= 45.900150 > > > > > AccelError= 47.023519 > > > > > AccelError= 47.551550 > > > > > AccelError= 48.022541 > > > > > AccelError= 46.395295 > > > > > AccelError= 81526.314840 > > > > > AccelError= 72853.056587 > > > > > AccelError= 65101.872298 > > > > > AccelError= 58175.992747 > > > > > AccelError= 51986.611437 > > > > > AccelError= 46456.046360 > > > > > AccelError= 41513.718492 > > > > > AccelError= 37097.300673 > > > > > AccelError= 33150.797142 > > > > > AccelError= 29622.770379 > > > > > AccelError= 26468.350534 > > > > > AccelError= 23652.797849 > > > > > AccelError= 21136.323622 > > > > > AccelError= 18888.213633 > > > > > AccelError= 16879.065077 > > > > > AccelError= 15083.474245 > > > > > AccelError= 13479.186003 > > > > > AccelError= 12045.566722 > > > > > AccelError= 10764.204608 > > > > > AccelError= 9619.483512 > > > > > AccelError= 8596.141212 > > > > > AccelError= 7682.015002 > > > > > AccelError= 6866.662006 > > > > > AccelError= 6139.961139 > > > > > AccelError= 5490.446488 > > > > > AccelError= 4910.042286 > > > > > AccelError= 4391.574405 > > > > > AccelError= 3928.114510 > > > > > AccelError= 3514.000485 > > > > > AccelError= 3144.002655 > > > > > AccelError= 2813.316136 > > > > > AccelError= 2517.500643 > > > > > AccelError= 2253.490792 > > > > > AccelError= 2017.432428 > > > > > AccelError= 1806.266152 > > > > > AccelError= 1617.706032 > > > > > AccelError= 1448.982225 > > > > > AccelError= 1298.398178 > > > > > AccelError= 1163.621865 > > > > > AccelError= 1043.472200 > > > > > AccelError= 935.821407 > > > > > AccelError= 839.766485 > > > > > AccelError= 753.863074 > > > > > AccelError= 677.101783 > > > > > AccelError= 608.587862 > > > > > AccelError= 547.209083 > > > > > AccelError= 492.589836 > > > > > AccelError= 443.554631 > > > > > AccelError= 399.735924 > > > > > AccelError= 360.639632 > > > > > AccelError= 325.643422 > > > > > AccelError= 294.370902 > > > > > AccelError= 266.425861 > > > > > AccelError= 241.478288 > > > > > AccelError= 219.191672 > > > > > AccelError= 199.246085 > > > > > AccelError= 181.422761 > > > > > AccelError= 165.532090 > > > > > AccelError= 151.295897 > > > > > AccelError= 138.574472 > > > > > AccelError= 127.206640 > > > > > AccelError= 117.084135 > > > > > AccelError= 108.003075 > > > > > AccelError= 99.888264 > > > > > AccelError= 92.598183 > > > > > AccelError= 86.122270 > > > > > AccelError= 76.770777 > > > > > AccelError= 65.225363 > > > > > AccelError= 54.908134 > > > > > AccelError= 45.688617 > > > > > AccelError= 45.070435 > > > > > AccelError= 45.411744 > > > > > AccelError= 45.770036 > > > > > AccelError= 45.871485 > > > > > AccelError= 45.239885 > > > > > AccelError= 45.631836 > > > > > AccelError= 45.001237 > > > > > AccelError= 45.373033 > > > > > AccelError= 45.577907 > > > > > AccelError= 46.322608 > > > > > AccelError= 46.958820 > > > > > AccelError= 46.038556 > > > > > AccelError= 45.018088 > > > > > > > > > > If you test it on your PC and will be ok, then maybe something is wrong > > > > > with my laptop. Then I will test on other PC. > > > > > > > > > > Best Regards > > > > > Albert P³atek > > > > > > > > > > --- In DynoMotion@yahoogroups.com <mailto:DynoMotion%40yahoogroups.com> , "albertplatek" <albertplatek86@> > > > > > wrote: > > > > > > > > > > > > Hi Tom, > > > > > > > > > > > > Problem solved (probably), weird but after re-flash firmware 4.28 > > > > > > everything is working ok, no axis stall. > > > > > > > > > > > > btw. Experimental version KMotion429x.exe is not working, when I start > > > > > > execute gcode I get message "G Code Error GCode Aborted", this error > > > > > > occurs on any gcode program even on only few G01 lines. > > > > > > > > > > > > Tom if you have C program that watch for too high of acceleration > > > > > > please post it. If again in future something strange will > > > > > > happen I will try find out the problem. > > > > > > I am still not sure what was the problem in my system that axis > > > > > > were stalling. > > > > > > > > > > > > Best Regards > > > > > > Albert P³atek > > > > > > > > > > > > > > > > > > > > > > > > --- In DynoMotion@yahoogroups.com <mailto:DynoMotion%40yahoogroups.com> , Albert P³atek <albertplatek86@> > > > > > wrote: > > > > > > > > > > > > > > Hi Tom, > > > > > > > > > > > > > > Thank you for help. > > > > > > > Mostly stall X axis, and sometimes Z. > > > > > > > Y axis seem to be ok. > > > > > > > I uploaded file as an attachment (zip). > > > > > > > > > > > > > > Best Regards > > > > > > > Albert Platek > > > > > > > > > > > > > > 2012/2/5 Tom Kerekes <tk@> > > > > > > > > > > > > > > > ** > > > > > > > > > > > > > > > > > > > > > > > > Hi Albert, > > > > > > > > > > > > > > > > Thanks for the very complete problem description, I can't tell you > > > > > how > > > > > > > > much I appreciate that, but I can't think of anything that would > > > > > cause > > > > > > > > this. Anything with large files I might expect some processing issue > > > > > but > > > > > > > > that would result in a buffer underflow and an error message not > > > > > incorrect > > > > > > > > motion that would result in a stall. That delay I described in your > > > > > email > > > > > > > > only happens when "Execute" is pushed so I don't think it is related. > > > > > > > > > > > > > > > > There is an experimental version KMotion429x.exe on the download site > > > > > > > > that has this delay eliminated and also faster USB throughput, but I > > > > > don't > > > > > > > > think either of those are related to this problem. > > > > > > > > > > > > > > > > Can you post the file somewhere? I'm thinking to write a C program > > > > > that > > > > > > > > would run in KFLOP and watch for too high of acceleration or > > > > > velocity by > > > > > > > > monitoring how ch0->Dest changes. If it detects something bad that > > > > > could > > > > > > > > be used as a trigger to help us track it down. > > > > > > > > > > > > > > > > Which axes loose position? > > > > > > > > > > > > > > > > Regards > > > > > > > > TK > > > > > > > > > > > > > > > > > > > > > > > > *From:* albertplatek <albertplatek86@> > > > > > > > > *To:* DynoMotion@yahoogroups.com <mailto:DynoMotion%40yahoogroups.com> > > > > > > > > *Sent:* Sunday, February 5, 2012 5:27 AM > > > > > > > > > > > > > > > > *Subject:* [DynoMotion] Re: Coord buffer overflow again > > > > > > > > > > > > > > > > > > > > > > > > Hi Tom, > > > > > > > > > > > > > > > > My settings: > > > > > > > > Break angle 30 deg > > > > > > > > Look ahead 3 sec > > > > > > > > Collinear tolerance 0.001 in > > > > > > > > Corner tolerance 0.003 in > > > > > > > > Facet angle 1.5 deg > > > > > > > > Tau = 0.008 sec > > > > > > > > XYZ Accelarations 40 in/sec2 > > > > > > > > XYZ Velocity 4.5 in/sec > > > > > > > > In gcode file I use max Vel 270 in/min (4.5 in/sec) > > > > > > > > > > > > > > > > I have strange problem, everything was ok until I started machining > > > > > > > > bigger reliefs. > > > > > > > > At half of the job my stepper motors lost position > > > > > > > > probably becouse of high jerk of acceleration. > > > > > > > > Firstly I thought that maybe axis accelerations for my system are to > > > > > > > > high(40in/sec2). So I changed it to 20in/sec2 for test. > > > > > > > > But didnt help, still stepper motors stall at half of the gcode file. > > > > > > > > > > > > > > > > This file contain 548389 lines, my steppers stall between 226000 - > > > > > 274000 > > > > > > > > line but not always in same place. > > > > > > > > > > > > > > > > I tried to find line where the problem occurs, but when I found line > > > > > with > > > > > > > > problem and tried to execute 500 lines before, it was ok. Motors > > > > > didnt > > > > > > > > stall. > > > > > > > > I found that motors stall when KMotionCNC start execute at least > > > > > > > > 100 000 lines before line number 274000. > > > > > > > > > > > > > > > > The problem disappear when I set break angle 5deg. > > > > > > > > > > > > > > > > "This diagnostic pointed out a side problem I hadn't noticed before. > > > > > There > > > > > > > > is an annoying ~30 second delay when it runs the second time. The > > > > > > > > Interpreter keeps a record of all its state changes which allow > > > > > Halts to go > > > > > > > > back to any previous state. In this test of 100x500 subroutine calls > > > > > > > > results in about 1 million state changes. Ironically creating the > > > > > data, > > > > > > > > making use of the data all happens in the blink of an eye, but > > > > > discarding > > > > > > > > the data (CList.RemoveAll) takes 30 seconds! We will find a better > > > > > way in > > > > > > > > the next version. For now just don't be surprised if there is an > > > > > annoying > > > > > > > > delay on a second run of a big job." > > > > > > > > > > > > > > > > Can this "delay" make that my stepper motors stall? > > > > > > > > > > > > > > > > Best Regards > > > > > > > > Albert PÃÆ'‚³atek > > > > > > > > > > > > > > > > --- In DynoMotion@yahoogroups.com <mailto:DynoMotion%40yahoogroups.com> , Albert <albertplatek86@> wrote: > > > > > > > > > > > > > > > > > > Hi Tom, > > > > > > > > > > > > > > > > > > Thank you. > > > > > > > > > Btw. I am trying to find out what changes make > > > > > > > > > Fidelizer in Windows. If I will find something > > > > > > > > > important I will let you know. > > > > > > > > > Best regards > > > > > > > > > Albert Platek > > > > > > > > > > > > > > > > > > > > > > > > > > > Dnia 2012-01-18 o godz. 23:47 Tom Kerekes <tk@> napisaÃÆ'Æ'…ÃÆ'¢â‚¬Å¡ > > > > > > > > > (a): > > > > > > > > > > > > > > > > > > > Hi Albert, > > > > > > > > > > > > > > > > > > > > Just wanted to mention that I now do notice an improvement with > > > > > > > > > > Fidelizer. I don't understand why I didn't before. I might have > > > > > > > > > > made a mistake or something changed. > > > > > > > > > > > > > > > > > > > > I sent an email to FTDI (USB chip we use) regarding it and the > > > > > USB > > > > > > > > > > 3.0 port issue to see if they have any answers. I figure it is > > > > > > > > > > worth the cost of an email :} > > > > > > > > > > > > > > > > > > > > Still working on some optimization... > > > > > > > > > > > > > > > > > > > > Regards > > > > > > > > > > TK > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > From: albertplatek <albertplatek86@> > > > > > > > > > > To: DynoMotion@yahoogroups.com <mailto:DynoMotion%40yahoogroups.com> > > > > > > > > > > Sent: Monday, January 16, 2012 12:01 PM > > > > > > > > > > Subject: [DynoMotion] Re: Coord buffer overflow again > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > Hi Tom, > > > > > > > > > > Thank you for making changes in new version to increase the > > > > > segment > > > > > > > > > > rate. > > > > > > > > > > > > > > > > > > > > Btw: > > > > > > > > > > I installed on second pc Windows 7 (Core2Duo cpu), and aslo > > > > > tested > > > > > > > > > > with fidelizer. Strange but didn't make difference, as in your > > > > > > > > > > computer. > > > > > > > > > > > > > > > > > > > > I cant get why in this particular case (pc with WIN7 i7 cpu) it > > > > > helps. > > > > > > > > > > Maybe it depends on type of host controller... > > > > > > > > > > > > > > > > > > > > Pc with WIN7 64bit (Core2Duo) has > > > > > > > > > > SiS 7001 PCI to USB Open Host Controller > > > > > > > > > > > > > > > > > > > > Pc with WIN7 64bit (i7) has > > > > > > > > > > Intel(R) 6 Series/C200 Series Chipset Family USB Enhanced Host > > > > > > > > > > Controller > > > > > > > > > > > > > > > > > > > > This computer has aslo another usb ports: Renesas Electronics USB > > > > > > > > > > 3.0 Host Controller > > > > > > > > > > which works very bad with KFlop. And on this port fidelizer make > > > > > no > > > > > > > > > > difference. > > > > > > > > > > > > > > > > > > > > Best Regards > > > > > > > > > > Albert Platek > > > > > > > > > > > > > > > > > > > > --- In DynoMotion@yahoogroups.com <mailto:DynoMotion%40yahoogroups.com> , Tom Kerekes <tk@> wrote: > > > > > > > > > > > > > > > > > > > > > > Hi Albert, > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ > > > > > > > > > > > Thanks for that interesting information. > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ > > > > > > > > > > > I tried the Fidelizer on my i7 W7 32 machine and it didn't > > > > > seem to > > > > > > > > > > make any difference. > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ > > > > > > > > > > > I'm trying some things to increase the segment rate by sending > > > > > > > > > > blocks of segments and optimizing how they are handled in > > > > > KFLOP.ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ I > > > > > > > > > > f this bears fruit I'll let you know. > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ > > > > > > > > > > > This diagnostic pointed out a side problem I hadn't noticed > > > > > before > > > > > > > > > > .ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ There is an annoying ~30 second delay when it runs the > > > > > second ti > > > > > > > > > > me.ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ The Interpreter keeps a record of all its state changes > > > > > which > > > > > > > > > > allow Halts to go back to any previous state.ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ In this test > > > > > of 100x > > > > > > > > > > 500 subroutine calls results in about 1 million state > > > > > changes.ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ Iro > > > > > > > > > > nically creating the data, making use of the data all happens in > > > > > the > > > > > > > > > > blink of an eye, but discarding the data (CList.RemoveAll) takes > > > > > 30 > > > > > > > > > > seconds!ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ We will find a better way in the next > > > > > version.ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ For now > > > > > > > > > > just don't be surprisedÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ if there is an annoying delay on a > > > > > second > > > > > > > > > > run of a big job. > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ > > > > > > > > > > > Regards > > > > > > > > > > > TK > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > ________________________________ > > > > > > > > > > > From: albertplatek <albertplatek86@> > > > > > > > > > > > To: DynoMotion@yahoogroups.com <mailto:DynoMotion%40yahoogroups.com> > > > > > > > > > > > Sent: Sunday, January 15, 2012 8:37 AM > > > > > > > > > > > Subject: [DynoMotion] Re: Coord buffer overflow again > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ > > > > > > > > > > > Hi Tom, > > > > > > > > > > > First thank you Tom for the test file. > > > > > > > > > > > I made few tests and I got interesting results :) > > > > > > > > > > > > > > > > > > > > > > PC No1: > > > > > > > > > > > > > > > > > > > > > > Intel Pentium 4 2.6Ghz WinXP 32bit port USB2.0 > > > > > > > > > > > Max segments: 110ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ > > > > > > > > > > > Time: 31s > > > > > > > > > > > 100 circles x 110 segments/circle / 31 seconds =354 segments > > > > > per > > > > > > > > > > second. > > > > > > > > > > > > > > > > > > > > > > PC No2: > > > > > > > > > > > > > > > > > > > > > > Intel Core2Duo 2.0 Ghz WinXP 32bit port USB2.0 > > > > > > > > > > > Max segments: 140ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ > > > > > > > > > > > Time: 31s > > > > > > > > > > > 100 circles x 140 segments/circle / 31 seconds = 451 segments > > > > > per > > > > > > > > > > second. > > > > > > > > > > > > > > > > > > > > > > PC No3: > > > > > > > > > > > > > > > > > > > > > > i7 2670QM Win7 64bit port USB3.0 > > > > > > > > > > > Max segments:30ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ > > > > > > > > > > > Time:31s > > > > > > > > > > > 100 circles x 60 segments/circle / 31 seconds = 193 segments > > > > > per > > > > > > > > > > second. > > > > > > > > > > > (thats why I had problem here, usb 3.0 fail) > > > > > > > > > > > > > > > > > > > > > > (Luckily I had 1 USB 2.0 port) > > > > > > > > > > > > > > > > > > > > > > i7 2670QM Win7 64bit port USB2.0 > > > > > > > > > > > Max segments:170 > > > > > > > > > > > Time:31s > > > > > > > > > > > 100 circles x 170 segments/circle / 31 seconds = 548 segments > > > > > per > > > > > > > > > > second. > > > > > > > > > > > > > > > > > > > > > > i7 2670QM Win7 64bit port USB2.0 (with 1 core affinity for > > > > > > > > > > KmotionCNC and KmotionServer) > > > > > > > > > > > Max segments:200 > > > > > > > > > > > Time:31s > > > > > > > > > > > 100 circles x 200 segments/circle / 31 seconds =645 segments > > > > > per > > > > > > > > > > second. > > > > > > > > > > > > > > > > > > > > > > I found on net some interesting software called Fidelizer, its > > > > > > > > > > software to optimize win7/vista for audio workstation, but it > > > > > aslo > > > > > > > > > > tweak somehow usb throughput. > > > > > > > > > > > More detail here: /www.windowsxlive.net/fidelizer > > > > > > > > > > > > > > > > > > > > > > i7 2670QM Win7 64bit port USB2.0 (Fidelizer) > > > > > > > > > > > Max segments:410 > > > > > > > > > > > Time:31s > > > > > > > > > > > 100 circles x 410 segments/circle / 31 seconds = 1320 segments > > > > > per > > > > > > > > > > second. (wow! tested few times with same result) > > > > > > > > > > > > > > > > > > > > > > Be aware of new USB 3.0 ports they may not work with good > > > > > > > > > > performance with KFlop, maybe becouse of bug in drivers. Windows > > > > > 7 > > > > > > > > > > dont see them without additional drivers. > > > > > > > > > > > Fidelizer software only improved performance on pc with ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ > > > > > Win7. > > > > > > > > > > > On WinXP seem to not work. I will soon try install on "PC No2" > > > > > > > > > > Windows 7 and then try fidelizer also. > > > > > > > > > > > > > > > > > > > > > > Best regards > > > > > > > > > > > Albert PÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚³atek > > > > > > > > > > > --- In DynoMotion@yahoogroups.com <mailto:DynoMotion%40yahoogroups.com> , Tom Kerekes <tk@> wrote: > > > > > > > > > > > > > > > > > > > > > > > > Hi Albert, > > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ > > > > > > > > > > > > Your application/system is quite demanding :} > > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ > > > > > > > > > > > > That portion of GCode you sent is 100,000 lines of all G1 > > > > > > > > > > movements with typical tiny lengths of only > > > > > 50~150um.ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ The > > > > > > > > > > collinear tolerance of 0.001 (25um) will cause many > > > > > blocksÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ to > > > > > > > > > > be combined together so the real required vector rate will be > > > > > more d > > > > > > > > > > ependent on the path curvature rather than the original > > > > > GCode.ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ > > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ > > > > > > > > > > > > On my system (i7 W7 32-bit) it runs ok with a facet angle 0f > > > > > 0.4 > > > > > > > > > > degrees butÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ can't keep up at 0.3 degrees. > > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ > > > > > > > > > > > > I did some calculations for a system like yours with 40 > > > > > in/sec2 > > > > > > > > > > acceleration.ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ Here is what I find: > > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ > > > > > > > > > > > > That acceleration allows going around a 0.1 inch radius > > > > > circle > > > > > > > > > > at 120ipm (which is about 3 circles per second) > > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ > > > > > > > > > > > > With a facet angle setting of 1.5 degrees that requires 240 > > > > > > > > > > segments per circle which would require 720 segments per second > > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ > > > > > > > > > > > > I created a diagnostic test GCode file that programmatically > > > > > > > > > > generates an N sided Polygon 100 timesÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ with > > > > > radius 0.1 inches > > > > > > > > > > and allows you to specify the number of polygon sides (it also > > > > > gradu > > > > > > > > > > ally shifts the "circle" to make the plot more > > > > > interesting).ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ > > > > > > > > > > (see attachment). ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ When using the diagnostic > > > > > make sure to set > > > > > > > > > > the Collinear tolerance to zero and the Corner Rounding to zero > > > > > so t > > > > > > > > > > he original segments are used.ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ By changing N > > > > > (number of sides > > > > > > > > > > we can determine the segment download rate). > > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ > > > > > > > > > > > > For example on my system it works with N = 200 and fails > > > > > with N > > > > > > > > > > = 240.ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ The total time to do all 100 circles > > > > > takes about 30 > > > > > > > > > > seconds.ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ So: > > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ > > > > > > > > > > > > 100 circles x 200 segments/circle / 30 seconds = 666 segments > > > > > > > > > > per second. > > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ > > > > > > > > > > > > In simulation mode the Interpreter+Graphics finishes the job > > > > > in > > > > > > > > > > ~ 5 seconds so it isn't a significant factor > > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ > > > > > > > > > > > > With the GCode Window closed it simulates in ~ 4 seconds so > > > > > the > > > > > > > > > > Graphical Display is not aÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ significant factor. > > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ > > > > > > > > > > > > You might run the tests on your two systems to see if you get > > > > > > > > > > drastically different results. > > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ > > > > > > > > > > > > But to try to answer your original question of why one CPU > > > > > would > > > > > > > > > > be significantly different I don't know.ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ Google > > > > > finds some > > > > > > > > > > benchmarking tools but they areÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ mainly for High > > > > > Speed USB > > > > > > > > > > devices like flash drives.ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ Some sites state if > > > > > an older USB > > > > > > > > > > 1.1 device is attached it slows everything down.ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ > > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ > > > > > > > > > > > > Regards > > > > > > > > > > > > TKÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ > > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ > > > > > > > > > > > > > > > > > > > > > > > > From: albertplatek albertplatek86@ > > > > > > > > > > > > To: DynoMotion@yahoogroups.com <mailto:DynoMotion%40yahoogroups.com> > > > > > > > > > > > > Sent: Saturday, January 14, 2012 11:02 AM > > > > > > > > > > > > Subject: [DynoMotion] Re: Coord buffer overflow again > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ > > > > > > > > > > > > Hi Tom, > > > > > > > > > > > > How to determine USB throughput on pc? or what pc/system is > > > > > best > > > > > > > > > > for KFlop? > > > > > > > > > > > > > > > > > > > > > > > > I have still sometimes cord buffer overflow on second pc > > > > > even I > > > > > > > > > > > > set lower accelaration 25 and bigger facet angle 2.5 it > > > > > sometimes > > > > > > > > > > > > overflow. Its very strange pc has windows xp clean > > > > > installation, > > > > > > > > > > > > Integrated Intel Graphics. > > > > > > > > > > > > On first pc works perfect even with high acceleration 40, > > > > > facet > > > > > > > > > > angle 2.0 (also windows xp, but ati graphics) > > > > > > > > > > > > > > > > > > > > > > > > Break angle 30 > > > > > > > > > > > > Look ahead 3 > > > > > > > > > > > > Collinear tolerance 0.001 > > > > > > > > > > > > Corner tolerance 0.005 > > > > > > > > > > > > Vel 5 > > > > > > > > > > > > > > > > > > > > > > > > Best Regards > > > > > > > > > > > > Albert Platek > > > > > > > > > > > > > > > > > > > > > > > > --- In DynoMotion@yahoogroups.com <mailto:DynoMotion%40yahoogroups.com> , Albert > > > > > PÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚³atek > > > > > > > > > > albertplatek86@ wrote: > > > > > > > > > > > > > > > > > > > > > > > > > > Hi Tom, > > > > > > > > > > > > > GCode file is attached with this message. > > > > > > > > > > > > > Yes feedrate is 300ipm. > > > > > > > > > > > > > What do you mean by "path generation slows down"? - i mean > > > > > > > > > > that when I am > > > > > > > > > > > > > changing view perspective in GViewer > > > > > > > > > > > > > the "simulated tool" catching up end of buffered path. So > > > > > > > > > > thats why I > > > > > > > > > > > > > thought that maybe gviewer impact on performance somehow. > > > > > > > > > > > > > > > > > > > > > > > > > > Best Regards > > > > > > > > > > > > > Albert > > > > > PÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'â€Â 'ÃÆ'Æ'Æ'ÃÆ'‚¢ÃÆ'Æ'¢ÃÆ'¢â‚¬Å¡ÃÆ'‚¬ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚¦ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚¢ÃÆ'Æ'Æ'ÃÆ'‚¢ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚¬ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Â¦ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚¡atek > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > 2012/1/13 Tom Kerekes tk@ > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > Hi Albert, > > > > > > > > > > > > > > > > > > > > > > > > > > > > Yes please send the file. Is the feedrate set high > > > > > 300ipm? > > > > > > > > > > > > > > > > > > > > > > > > > > > > What do you mean by "path generation slows down"? > > > > > > > > > > > > > > > > > > > > > > > > > > > > Thanks > > > > > > > > > > > > > > TK > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > *From:* albertplatek albertplatek86@ > > > > > > > > > > > > > > *To:* DynoMotion@yahoogroups.com <mailto:DynoMotion%40yahoogroups.com> > > > > > > > > > > > > > > *Sent:* Thursday, January 12, 2012 3:35 PM > > > > > > > > > > > > > > *Subject:* [DynoMotion] Re: Coord buffer overflow again > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > Hi Tom, > > > > > > > > > > > > > > > > > > > > > > > > > > > > My settings: > > > > > > > > > > > > > > > > > > > > > > > > > > > > Break angle 30 > > > > > > > > > > > > > > Look ahead 3 > > > > > > > > > > > > > > Collinear tolerance 0.001 > > > > > > > > > > > > > > Corner tolerance 0.005 > > > > > > > > > > > > > > Facet angle 2.0 (if I go to 1.5, i get cord buffer > > > > > overflow > > > > > > > > > > almost from > > > > > > > > > > > > > > program start) > > > > > > > > > > > > > > > > > > > > > > > > > > > > I use Tau = 0.01; (with facet angle 2.0 machine movements > > > > > > > > > > are fairly > > > > > > > > > > > > > > smooth) > > > > > > > > > > > > > > > > > > > > > > > > > > > > All 3 axis has same Vel/Accel parameters: > > > > > > > > > > > > > > Vel 5.0 > > > > > > > > > > > > > > Accel 40 > > > > > > > > > > > > > > > > > > > > > > > > > > > > Toolpath is parallel along X-axis (contain a lot of small > > > > > > > > > > arcs, I will > > > > > > > > > > > > > > send file tomorrow) > > > > > > > > > > > > > > > > > > > > > > > > > > > > With this settings my first router with old laptop works > > > > > ok > > > > > > > > > > (Dell, core 2 > > > > > > > > > > > > > > duo, ati graphics card), > > > > > > > > > > > > > > but on second router (same settings) with new laptop dont > > > > > > > > > > work, i often > > > > > > > > > > > > > > get buffer overflows. (laptop Dell, i7 cpu, Win7 64bit, > > > > > > > > > > Nvidia Gt555) > > > > > > > > > > > > > > > > > > > > > > > > > > > > (Btw. when GViewer is open and I move view by mouse path > > > > > > > > > > generation slows > > > > > > > > > > > > > > down, is it normal?) > > > > > > > > > > > > > > > > > > > > > > > > > > > > Best Regards > > > > > > > > > > > > > > Albert PÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'â€Â 'ÃÆ'Æ'Æ'ÃÆ'‚¢ÃÆ'Æ'¢ÃÆ'¢â‚¬Å¡ÃÆ'‚¬ÃÆ'Æ'…ÃÆ'‚¡ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚³atek > > > > > > > > > > > > > > > > > > > > > > > > > > > > --- In DynoMotion@yahoogroups.com <mailto:DynoMotion%40yahoogroups.com> , Tom Kerekes <tk@> > > > > > wrote: > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > Hi Albert, > > > > > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'â€Â 'ÃÆ'Æ'Æ'ÃÆ'¢â‚¬ 'ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚¢ÃÆ'Æ'Æ'ÃÆ'‚¢ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚¬ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Â¦ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚¡ > > > > > > > > > > > > > > > Could you send me the Gcode and all your > > > > > settings?ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'â€Â 'ÃÆ'Æ'Æ'ÃÆ'¢â‚¬ 'ÃÆ'Æ'Æ'ÃÆ'†' > > > > > > > > > > ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚¢ÃÆ'Æ'Æ'ÃÆ'‚¢ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚¬ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Â¦ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚¡ I don't understand > > > > > > > > > > > > > > the collinear tolerance of > > > > > 1?ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'â€Â 'ÃÆ'Æ'Æ'ÃÆ'¢â‚¬ 'ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚¢ÃÆ'Æ'Æ'ÃÆ'‚¢ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚¬ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Â¦ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚¡ That > > > > > > > > > > seems huge.ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'â€Â 'ÃÆ'Æ'Æ'ÃÆ'¢â‚¬ 'ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚¢ÃÆ'Æ'Æ'ÃÆ'‚¢ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚¬ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Â¦ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚¡ Did you > > > > > mean 0.001? > > > > > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'â€Â 'ÃÆ'Æ'Æ'ÃÆ'¢â‚¬ 'ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚¢ÃÆ'Æ'Æ'ÃÆ'‚¢ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚¬ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Â¦ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚¡ > > > > > > > > > > > > > > > A corner rounding facet angle too small might be > > > > > > > > > > generating too many > > > > > > > > > > > > > > small > > > > > segments.ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'â€Â 'ÃÆ'Æ'Æ'ÃÆ'¢â‚¬ 'ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚¢ÃÆ'Æ'Æ'ÃÆ'‚¢ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚¬ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Â¦ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚¡ Are you using the > > > > > > > > > > Coordinated Motion Low Pass Filter > > > > > > > > > > > > > > Smoothing > > > > > option?ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'â€Â 'ÃÆ'Æ'Æ'ÃÆ'¢â‚¬ 'ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚¢ÃÆ'Æ'Æ'ÃÆ'‚¢ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚¬ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Â¦ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚¡ That might help make > > > > > > > > > > as smooth or smoother motion with > > > > > > > > > > > > > > larger facets. > > > > > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'â€Â 'ÃÆ'Æ'Æ'ÃÆ'¢â‚¬ 'ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚¢ÃÆ'Æ'Æ'ÃÆ'‚¢ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚¬ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Â¦ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚¡ > > > > > > > > > > > > > > > The USB throughput can vary from one machine to the > > > > > other. > > > > > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'â€Â 'ÃÆ'Æ'Æ'ÃÆ'¢â‚¬ 'ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚¢ÃÆ'Æ'Æ'ÃÆ'‚¢ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚¬ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Â¦ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚¡ > > > > > > > > > > > > > > > No data for the GViewer comes from KFLOP it all comes > > > > > from > > > > > > > > > > the > > > > > > > > > > > > > > trajectory planner on the PC. > > > > > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'â€Â 'ÃÆ'Æ'Æ'ÃÆ'¢â‚¬ 'ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚¢ÃÆ'Æ'Æ'ÃÆ'‚¢ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚¬ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Â¦ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚¡ > > > > > > > > > > > > > > > Regards > > > > > > > > > > > > > > > TK > > > > > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'â€Â 'ÃÆ'Æ'Æ'ÃÆ'¢â‚¬ 'ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚¢ÃÆ'Æ'Æ'ÃÆ'‚¢ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚¬ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Â¦ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚¡ > > > > > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'â€Â 'ÃÆ'Æ'Æ'ÃÆ'¢â‚¬ 'ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚¢ÃÆ'Æ'Æ'ÃÆ'‚¢ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚¬ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Â¦ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚¡ > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > From: albertplatek <albertplatek86@> > > > > > > > > > > > > > > > To: DynoMotion@yahoogroups.com <mailto:DynoMotion%40yahoogroups.com> > > > > > > > > > > > > > > > Sent: Thursday, January 12, 2012 11:11 AM > > > > > > > > > > > > > > > Subject: [DynoMotion] Coord buffer overflow again > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'â€Â 'ÃÆ'Æ'Æ'ÃÆ'¢â‚¬ 'ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚¢ÃÆ'Æ'Æ'ÃÆ'‚¢ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚¬ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Â¦ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚¡ > > > > > > > > > > > > > > > Hi Tom, > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > 1. Why on 1 pc kmotioncnc works ok, but on second not, > > > > > > > > > > with same > > > > > > > > > > > > > > settings? > > > > > > > > > > > > > > > (collinear tollerance 1) > > > > > > > > > > > > > > > same file, same accelerations etc. > > > > > > > > > > > > > > > on second I have only usb 3.0 ports. > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > 2. How much data come to Gviewer from kflop? if a lot > > > > > is > > > > > > > > > > it possible > > > > > > > > > > > > > > > to bypass this if I dont need to use G viewer? > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > I make a lot of reliefs in wood, with kinda high > > > > > > > > > > acceleleration 40, if > > > > > > > > > > > > > > > will be any way to get more performance from kflop with > > > > > > > > > > smoothing > > > > > > > > > > > > > > feature will be very good. > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > Best Regards > > > > > > > > > > > > > > > Albert > > > > > PÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'â€Â 'ÃÆ'Æ'Æ'ÃÆ'¢â‚¬ 'ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚¢ÃÆ'Æ'Æ'ÃÆ'‚¢ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚¬ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Â¦ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚¡ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'â€Â 'ÃÆ'Æ'Æ'ÃÆ'‚¢ÃÆ'Æ'¢ÃÆ'¢â‚¬Å¡ÃÆ'‚¬ÃÆ'Æ'…ÃÆ'‚¡ÃÆ'Æ'Æ'ÃÆ'†'ÃÆ'Æ'¢ÃÆ'Â¢Ã¢â‚¬à ¡Ã‚Â¬ÃÆ'…¡ÃÆ'Æ'Æ'ÃÆ'¢â‚¬Å¡ÃÆ'Æ'Ã¢â‚¬à ¡ÃÆ'‚³atek > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > ________________________________ > > > > > > From: albertplatek <albertplatek86@> > > > To: DynoMotion@yahoogroups.com <mailto:DynoMotion%40yahoogroups.com> > > > Sent: Wednesday, February 8, 2012 8:03 AM > > > Subject: [DynoMotion] Re: Coord buffer overflow again > > > > > >