Hi Pablo,
Welcome to the group.
Regarding:
#1 - Yes G32 G76 and the Wizards in Mach3 all work
#2 - Yes both are possible. As you may know closing the position loop around glass scales increases accuracy by eliminating errors caused by most mechanical components by placing them in the loop, however it can be difficult to make a fast stable loop because these mechanical elements add things like backlash and delay to the response. To help stabilize the loop, velocity feedback from the motor shaft is very helpful.
Steppers with closed loop scales is fairly straightforward. Steppers are position devices and inherently provide feedback at the motor shaft. As the Stepper's pole position is advanced to generate torque as soon as the torque overcomes friction and the motor begins to move the torque drops. This is not the case with a servo (torque) motor.
The simplest way to use servo motors with KFLOP and glass scale feedback is to make use of an amplifier that handles the feedback at the motor shaft by either an encoder or tachometer on the motor. KFLOP+Kanalog then simply outputs an analog velocity command to the amplifier. This type of position loop is fairly easy to tune as P (proportional) gain alone will make a stable system (in a position feedback loop where torque is commanded O gain alone will be unstable).
To use our SnapAmp with glass scales if feedback from the scales alone is not sufficient to meet your performance requirements you may need to use encoder feedback from the motor in a "dual loop" configuration. Currently to do this in KFLOP requires using two axis channels (of the 8 available) connected together. This actually works out nice as you can tune each loop independently and have all the axis parameters (PID + Filters etc) for each loop.
#3 - It sounds like you have a misconception of how typical incremental glass scales work. They basically put out pulses telling the system how far it has moved - not where you are at. So it needs to be referenced one time and then it can keep track of the position forever until there is a loss of power. KFLOP can keep track of the position even though the motor may stall, or you turn off the amplifiers, or exit Mach3, but not after a power cycle.
#4 - sorry no. The C programs are very simple, short, and mostly like script making high level calls so you shouldn't be afraid of them.
#5 - not really that I can think of. You can download our software, Run the KMotion application, use the C Programs screen to view, change, and compile some example programs, but you can not execute them without a KFLOP.
I don't really have any specific motor recommendations. The main limitation of SnapAmp is the Max 80V supply. There were some surplus Sanyo/Denski P5 motors (300~1000W) floating around that work well with SnapAmp but I don't think you will get 3000 RPM.
HTH
Regards
TK
| Group: DynoMotion |
Message: 2347 |
From: Pablo Untroib |
Date: 11/20/2011 |
| Subject: Re: CNC Lathe |
Re: [DynoMotion] CNC Lathe
Thanks for the reply Tom, much appreciated!
Friday, November 18, 2011, 2:32:37 PM, Tom Kerekes wrote:
|
Hi Pablo,
Welcome to the group.
Regarding:
#1 - Yes G32 G76 and the Wizards in Mach3 all work
#2 - Yes both are possible. As you may know closing the position loop around glass scales increases accuracy by eliminating errors caused by most mechanical components by placing them in the loop, however it can be difficult to make a fast stable loop because these mechanical elements add things like backlash and delay to the response. To help stabilize the loop, velocity feedback from the motor shaft is very helpful.
Steppers with closed loop scales is fairly straightforward. Steppers are position devices and inherently provide feedback at the motor shaft. As the Stepper's pole position is advanced to generate torque as soon as the torque overcomes friction and the motor begins to move the torque drops. This is not the case with a servo (torque) motor.
The simplest way to use servo motors with KFLOP and glass scale feedback is to make use of an amplifier that handles the feedback at the motor shaft by either an encoder or tachometer on the motor. KFLOP+Kanalog then simply outputs an analog velocity command to the amplifier. This type of position loop is fairly easy to tune as P (proportional) gain alone will make a stable system (in a position feedback loop where torque is commanded O gain alone will be unstable).
To use our SnapAmp with glass scales if feedback from the scales alone is not sufficient to meet your performance requirements you may need to use encoder feedback from the motor in a "dual loop" configuration. Currently to do this in KFLOP requires using two axis channels (of the 8 available) connected together. This actually works out nice as you can tune each loop independently and have all the axis parameters (PID + Filters etc) for each loop.
#3 - It sounds like you have a misconception of how typical incremental glass scales work. They basically put out pulses telling the system how far it has moved - not where you are at. So it needs to be referenced one time and then it can keep track of the position forever until there is a loss of power. KFLOP can keep track of the position even though the motor may stall, or you turn off the amplifiers, or exit Mach3, but not after a power cycle.
#4 - sorry no. The C programs are very simple, short, and mostly like script making high level calls so you shouldn't be afraid of them.
#5 - not really that I can think of. You can download our software, Run the KMotion application, use the C Programs screen to view, change, and compile some example programs, but you can not execute them without a KFLOP.
I don't really have any specific motor recommendations. The main limitation of SnapAmp is the Max 80V supply. There were some surplus Sanyo/Denski P5 motors (300~1000W) floating around that work well with SnapAmp but I don't think you will get 3000 RPM.
HTH
Regards
TK
From: Pablo <pablo@...>
To: DynoMotion@yahoogroups.com
Sent: Friday, November 18, 2011 5:14 AM
Subject: [DynoMotion] CNC Lathe
Hi, 1st post to the group.
I asked this at the mach3 yahoo list but I think I will receive a more accurate answer here.
We are in the developing stages of a mid size small parts gang tooled lathe and I'm looking for motion control alternatives.
Mach3 is going to be the user interface and I'm considering Dynomotion products among a couple others for the motion control.
I have a limitation regarding C programming, I don't program it.
I would like to know if I can configure a lathe to:
1- Do threading using mach3 canned cycles
2- Use stepper motors with some kind of encoder feedback, rotary or glass scale, or use dynomotion servoamps + servos w/encoders plus linear glass scales.
3 - Is there a way to avoid referencing? I mean for example using glass scales the machine could always know where it is and where it can or can't go.
4 - Can I do all this without C programming?
5 - Is there a way to prototype/simulate the solution before actually committing to purchase hardware?
Now another question regarding servo motors for the snapamp:
What brushless motor do you guys recommend or know to work OK in the 5~10Nm 3000RPM range
I'm sure there will be more questions, but this post is a good start :)
Thanks a lot!
Pablo
|
--
Pablo E. Untroib
-------
Suipacha 834 5to. "A"
(1008) Buenos Aires, Argentina
Tel. (+54-11) 4328-7500 / 7400
Phone USA: (260) UNTROIB
Twitter @PEU_AR
PGP Key ID: : 0xFC7177AF
|
|
| Group: DynoMotion |
Message: 2410 |
From: Pablo Untroib |
Date: 11/24/2011 |
| Subject: Re: CNC Lathe |
Re: [DynoMotion] CNC Lathe
Tom, let me ask you again regarding this reply of yours:
Friday, November 18, 2011, 2:32:37 PM, Tom Kerekes wrote:
|
#2 - Yes both are possible. As you may know closing the position loop around glass scales increases accuracy by eliminating errors caused by most mechanical components by placing them in the loop, however it can be difficult to make a fast stable loop because these mechanical elements add things like backlash and delay to the response. To help stabilize the loop, velocity feedback from the motor shaft is very helpful.
Steppers with closed loop scales is fairly straightforward. Steppers are position devices and inherently provide feedback at the motor shaft. As the Stepper's pole position is advanced to generate torque as soon as the torque overcomes friction and the motor begins to move the torque drops. |
To close the loop you say that a linear scale is not fast enough to keep the loop stable so you suggest an encoder, ton of machines uses this setup everyday, what I want to know is: if I use an encoder attached to a stepper I stll cannot compensate for backslash since the play is between the nut and the screw/motor, my idea for using the glass scale is to add this extra feedback to the loop to eliminate any backslash.
New questions:
1) Is this practical or I should consider purchasing far more expensive ballscrews? you know... prices of rolled vs ground ballscrews are veeery different.
2) In your experience how fast an axis can move for the position to be accurately tracked by a glass scale?
3) what brand/model of encoder you know to work OK, without the need of expensive cabling, with 1.8deg steppers?
Thanks Again!
--
Pablo E. Untroib
-------
Suipacha 834 5to. "A"
(1008) Buenos Aires, Argentina
Tel. (+54-11) 4328-7500 / 7400
Phone USA: (260) UNTROIB
Twitter @PEU_AR
PGP Key ID: : 0xFC7177AF
|
|
| Group: DynoMotion |
Message: 2412 |
From: oxforduniv |
Date: 11/24/2011 |
| Subject: Re: CNC Lathe |
Pablo, I think you will find that the machines you are referring to use systems
with dual feedback, an encoder on the motor with the scale on the final motion.
They typically split the PID loop up between PI on the Scale loop and the D on
the motor/encoder loop.
There is a on line video tutorial that explains this on the Galil site.
Al.
--- In DynoMotion@yahoogroups.com, Pablo Untroib <pablo@...> wrote:
>
>
> Tom, let me ask you again regarding this reply of yours:
>
> Friday, November 18, 2011, 2:32:37 PM, Tom Kerekes wrote:
>
>
> #2 - Yes both are possible. As you may know closing the position loop around glass scales increases accuracy by eliminating errors caused by most mechanical components by placing them in the loop, however it can be difficult to make a fast stable loop because these mechanical elements add things like backlash and delay to the response. To help stabilize the loop, velocity feedback from the motor shaft is very helpful.
> Steppers with closed loop scales is fairly straightforward. Steppers are position devices and inherently provide feedback at the motor shaft. As the Stepper's pole position is advanced to generate torque as soon as the torque overcomes friction and the motor begins to move the torque drops.
>
>
> PGP Key ID: : 0xFC7177AF
> |
|
| Group: DynoMotion |
Message: 2413 |
From: Tom Kerekes |
Date: 11/24/2011 |
| Subject: Re: CNC Lathe |
Hi Pablo,
The linear scales are very fast. That is not the problem. The problem is in your mechanics. There is no change or information in the glass scales until it is too late. I think if you do some research you will find all high end equipment make use of dual feedback from the motor shaft and the glass scales. Let me try and explain it this way. Imagine you are trying to steer a car down a road that has sloppy steering and a spring connection to the front wheels and all you can see is the position of the car. You cannot see the position of the steering wheel or the front tires. It will be very difficult to steer the car straight down the center of the road. The only solution is to make small changes to the steering wheel and wait to see if the position of the car changes as the front wheels may or may have even changed angle at all. If you just keep
turning the steering wheel rapidly until you see the car start to change direction it may be too late and you will end up weaving down the road. However if you then add additional information about when the front wheels actually start tuning, then you can make use of that information to slow down turning the steering wheel before you see any response of the position of the car. With this new information you will be able to make more rapid corrections to the steering wheel without weaving down the road. You should realize that no amount of intelligence or speed in the controller (pilot) will help if there is lack of information in what is happening in the system.
Obviously the the less backlash and the higher stiffness in your leadscrew the better everything will work. With Glass scale feedback the accuracy of the leadscrew is no longer important, but the amount of backlash and stiffness will determine how fast and accurately the system will be able to correct errors.
With regard to speed: KFLOP can track encoder counts up to 1MHz. So for example if the encoder resolution is 0.1um the max speed would be 0.1 m/s or 6000mm/min.
Most any encoders should work so long as they have digital quadrature output.
HTH
Regards
TK
| Group: DynoMotion |
Message: 2414 |
From: Pablo Untroib |
Date: 11/24/2011 |
| Subject: Re: CNC Lathe |
Re: [DynoMotion] Re: CNC Lathe
Thursday, November 24, 2011, 6:07:42 PM, oxforduniv wrote:
|
There is a on line video tutorial that explains this on the Galil site.
Al. |
Just finished listening to this presentation http://www.galilmc.com/learning/tutorial-display.php?tutorial=dual-loop-compensation-methods I guess its the one you directed me to.
Also, just in time I must say, just read Tom reply with the car example.
SO :)
My plan is to use not the worst neither the best ballscrews, for budget purposes I need to settle in the middle, typical backslash for a Hiwin ballcrew is quoted at less or equal to 0.03mm every 300mm of screw. Not a lot but its easily measurable.
Is there a way, remember I do not program in C, to implement the improved feedback loop as described in the video presentation (page 11) by separating the PD for the motor and then the I for the load via the glass scales?
Thanks!!
--
Pablo E. Untroib
-------
Suipacha 834 5to. "A"
(1008) Buenos Aires, Argentina
Tel. (+54-11) 4328-7500 / 7400
Phone USA: (260) UNTROIB
Twitter @PEU_AR
PGP Key ID: : 0xFC7177AF
|
|
| Group: DynoMotion |
Message: 2416 |
From: Tom Kerekes |
Date: 11/25/2011 |
| Subject: Re: CNC Lathe |
Hi Pablo,
Yes Dual Loop with KFLOP makes use of an axis channel for each loop. This means that all PID + filters + others are available for each loop. So if you wish to only use I for the outer loop like Galil does set the P and D and filter gains for that loop to zero. To use only P an D gains for the inner loop set the I and filters to zero. Having a separate axis channel for each loop is nice in that you can plot, test, and tune each of the loops separately rather than only looking at the overall result and not being able to see the performance of the inner loop. It is quite possible that for example using P gain in both loops will give a slightly higher performance system.
Regarding your ball screw choice: Backlash is normally a fixed amount and a function of the "nut" itself. Where lead errors and run out as a function of length are more related to the length of the screw. So I don't really understand why the backlash would be specified as a function of length. In systems with glass scale feedback accuracy is determined by the glass scale so you wouldn't benefit from a more accurate lead screw. The only requirement on the leadscrew is that the motion is smooth (without backlash).
Regards
TK
| Group: DynoMotion |
Message: 2419 |
From: Pablo Untroib |
Date: 11/25/2011 |
| Subject: Re: CNC Lathe |
Re: [DynoMotion] Re: CNC Lathe
Thanks Tom, Im looking for options to bring one kflop+kanalog to Argentina no start testing, hopefully next week will have the logistics arranged. Do you guys sell 2000ppr encoders? besides the encoder I already have everything to make a test bed axis.
Thanks!
Friday, November 25, 2011, 2:29:24 PM, Tom Kerekes wrote:
|
Hi Pablo,
Yes Dual Loop with KFLOP makes use of an axis channel for each loop. This means that all PID + filters + others are available for each loop. So if you wish to only use I for the outer loop like Galil does set the P and D and filter gains for that loop to zero. To use only P an D gains for the inner loop set the I and filters to zero. Having a separate axis channel for each loop is nice in that you can plot, test, and tune each of the loops separately rather than only looking at the overall result and not being able to see the performance of the inner loop. It is quite possible that for example using P gain in both loops will give a slightly higher performance system.
Regarding your ball screw choice: Backlash is normally a fixed amount and a function of the "nut" itself. Where lead errors and run out as a function of length are more related to the length of the screw. So I don't really understand why the backlash would be specified as a function of length. In systems with glass scale feedback accuracy is determined by the glass scale so you wouldn't benefit from a more accurate lead screw. The only requirement on the leadscrew is that the motion is smooth (without backlash).
Regards
TK
From: Pablo Untroib <pablo@...>
To: oxforduniv <DynoMotion@yahoogroups.com>
Sent: Thursday, November 24, 2011 2:03 PM
Subject: Re: [DynoMotion] Re: CNC Lathe
Thursday, November 24, 2011, 6:07:42 PM, oxforduniv wrote:
|
There is a on line video tutorial that explains this on the Galil site.
Al. |
Just finished listening to this presentation http://www.galilmc.com/learning/tutorial-display.php?tutorial=dual-loop-compensation-methods I guess its the one you directed me to.
Also, just in time I must say, just read Tom reply with the car example.
SO :)
My plan is to use not the worst neither the best ballscrews, for budget purposes I need to settle in the middle, typical backslash for a Hiwin ballcrew is quoted at less or equal to 0.03mm every 300mm of screw. Not a lot but its easily measurable.
Is there a way, remember I do not program in C, to implement the improved feedback loop as described in the video presentation (page 11) by separating the PD for the motor and then the I for the load via the glass scales?
Thanks!!
--
Pablo E. Untroib
-------
Suipacha 834 5to. "A"
(1008) Buenos Aires, Argentina
Tel. (+54-11) 4328-7500 / 7400
Phone USA: (260) UNTROIB
Twitter @PEU_AR
PGP Key ID: : 0xFC7177AF
|
--
Pablo E. Untroib
-------
Suipacha 834 5to. "A"
(1008) Buenos Aires, Argentina
Tel. (+54-11) 4328-7500 / 7400
Phone USA: (260) UNTROIB
Twitter @PEU_AR
PGP Key ID: : 0xFC7177AF
|
|
| Group: DynoMotion |
Message: 2420 |
From: Tom Kerekes |
Date: 11/25/2011 |
| Subject: Re: CNC Lathe |
Hi Pablo,
No we don't sell encoders. Here might be one source:
If possible get the differential output option.
Regards
TK
| Group: DynoMotion |
Message: 2427 |
From: Pablo Untroib |
Date: 11/25/2011 |
| Subject: Re: CNC Lathe |
Re: [DynoMotion] Re: CNC Lathe
Im configuring the E6 model, it says if you use 6 feet or less the single ended model is commonly used. Is there a non obvious reason to use the differential version?
I recall this thread at cnczone regarding encoders http://www.cnczone.com/forums/gecko_drives/89079-new_gecko_g320x_startup_problems.html and geckodrives, does the same applies to kflop?
Friday, November 25, 2011, 3:59:15 PM, Tom Kerekes wrote:
|
Hi Pablo,
No we don't sell encoders. Here might be one source:
http://www.usdigital.com/
If possible get the differential output option.
Regards
TK
From: Pablo Untroib <pablo@...>
To: Tom Kerekes <DynoMotion@yahoogroups.com>
Sent: Friday, November 25, 2011 10:35 AM
Subject: Re: [DynoMotion] Re: CNC Lathe
Thanks Tom, Im looking for options to bring one kflop+kanalog to Argentina no start testing, hopefully next week will have the logistics arranged. Do you guys sell 2000ppr encoders? besides the encoder I already have everything to make a test bed axis.
Thanks!
Friday, November 25, 2011, 2:29:24 PM, Tom Kerekes wrote:
|
Hi Pablo,
Yes Dual Loop with KFLOP makes use of an axis channel for each loop. This means that all PID + filters + others are available for each loop. So if you wish to only use I for the outer loop like Galil does set the P and D and filter gains for that loop to zero. To use only P an D gains for the inner loop set the I and filters to zero. Having a separate axis channel for each loop is nice in that you can plot, test, and tune each of the loops separately rather than only looking at the overall result and not being able to see the performance of the inner loop. It is quite possible that for example using P gain in both loops will give a slightly higher performance system.
Regarding your ball screw choice: Backlash is normally a fixed amount and a function of the "nut" itself. Where lead errors and run out as a function of length are more related to the length of the screw. So I don't really understand why the backlash would be specified as a function of length. In systems with glass scale feedback accuracy is determined by the glass scale so you wouldn't benefit from a more accurate lead screw. The only requirement on the leadscrew is that the motion is smooth (without backlash).
Regards
TK
From: Pablo Untroib <pablo@...>
To: oxforduniv <DynoMotion@yahoogroups.com>
Sent: Thursday, November 24, 2011 2:03 PM
Subject: Re: [DynoMotion] Re: CNC Lathe
Thursday, November 24, 2011, 6:07:42 PM, oxforduniv wrote:
|
There is a on line video tutorial that explains this on the Galil site.
Al. |
Just finished listening to this presentation http://www.galilmc.com/learning/tutorial-display.php?tutorial=dual-loop-compensation-methods I guess its the one you directed me to.
Also, just in time I must say, just read Tom reply with the car example.
SO :)
My plan is to use not the worst neither the best ballscrews, for budget purposes I need to settle in the middle, typical backslash for a Hiwin ballcrew is quoted at less or equal to 0.03mm every 300mm of screw. Not a lot but its easily measurable.
Is there a way, remember I do not program in C, to implement the improved feedback loop as described in the video presentation (page 11) by separating the PD for the motor and then the I for the load via the glass scales?
Thanks!!
--
Pablo E. Untroib
-------
Suipacha 834 5to. "A"
(1008) Buenos Aires, Argentina
Tel. (+54-11) 4328-7500 / 7400
Phone USA: (260) UNTROIB
Twitter @PEU_AR
PGP Key ID: : 0xFC7177AF |
--
Pablo E. Untroib
-------
Suipacha 834 5to. "A"
(1008) Buenos Aires, Argentina
Tel. (+54-11) 4328-7500 / 7400
Phone USA: (260) UNTROIB
Twitter @PEU_AR
PGP Key ID: : 0xFC7177AF
|
--
Pablo E. Untroib
-------
Suipacha 834 5to. "A"
(1008) Buenos Aires, Argentina
Tel. (+54-11) 4328-7500 / 7400
Phone USA: (260) UNTROIB
Twitter @PEU_AR
PGP Key ID: : 0xFC7177AF
|
|
| Group: DynoMotion |
Message: 2429 |
From: Tom Kerekes |
Date: 11/25/2011 |
| Subject: Re: CNC Lathe |
Hi Pablo,
Differential signals are much less noise sensitive than single ended signals. So if you plan to use our Kanalog or SnapAmp boards which have differential inputs it would be worth a few extra $ to go that route. Those E6 models appear to have very weak single ended drive capability which makes going single ended an even poorer choice. It seems the Gecko G320Xs don't accept differential inputs so they recommend a pull up resistor to help with the single ended drive.
Regards
TK
| Group: DynoMotion |
Message: 2430 |
From: oxforduniv |
Date: 11/25/2011 |
| Subject: Re: CNC Lathe |
Differential encoders typically use RS422 drivers, The standard for RS422 is Data rates of up to 100K bits/second and distances up to 4000 Ft.
Al.
--- In DynoMotion@yahoogroups.com, Pablo Untroib <pablo@...> wrote:
>
>
>
> Im configuring the E6 model, it says if you use 6 feet or less the single ended model is commonly used. Is there a non obvious reason to use the differential version?
>
> I recall this thread at cnczone regarding encoders http://www.cnczone.com/forums/gecko_drives/89079-new_gecko_g320x_startup_problems.html and geckodrives, does the same applies to kflop?
>> Pablo E. Untroib
> -------
> Suipacha 834 5to. "A"
> (1008) Buenos Aires, Argentina
> Tel. (+54-11) 4328-7500 / 7400
> Phone USA: (260) UNTROIB
> Twitter @PEU_AR
>
> PGP Key ID: : 0xFC7177AF
> |
|
| Group: DynoMotion |
Message: 2434 |
From: Pablo Untroib |
Date: 11/25/2011 |
| Subject: Re: CNC Lathe |
Re: [DynoMotion] Re: CNC Lathe
I declare myself absolutelly ignorant regarding encoders, Im a plain stepper guy :)
So differential it is, do these require custom cabling, or I can wire them myself? I see they offer cables.
Friday, November 25, 2011, 5:55:37 PM, Tom Kerekes wrote:
|
Hi Pablo,
Differential signals are much less noise sensitive than single ended signals. So if you plan to use our Kanalog or SnapAmp boards which have differential inputs it would be worth a few extra $ to go that route. Those E6 models appear to have very weak single ended drive capability which makes going single ended an even poorer choice. It seems the Gecko G320Xs don't accept differential inputs so they recommend a pull up resistor to help with the single ended drive.
Regards
TK
From: Pablo Untroib <pablo@...>
To: Tom Kerekes <DynoMotion@yahoogroups.com>
Sent: Friday, November 25, 2011 12:19 PM
Subject: Re: [DynoMotion] Re: CNC Lathe
Im configuring the E6 model, it says if you use 6 feet or less the single ended model is commonly used. Is there a non obvious reason to use the differential version?
I recall this thread at cnczone regarding encoders http://www.cnczone.com/forums/gecko_drives/89079-new_gecko_g320x_startup_problems.html and geckodrives, does the same applies to kflop?
Friday, November 25, 2011, 3:59:15 PM, Tom Kerekes wrote:
|
Hi Pablo,
No we don't sell encoders. Here might be one source:
http://www.usdigital.com/
If possible get the differential output option.
Regards
TK
From: Pablo Untroib <pablo@...>
To: Tom Kerekes <DynoMotion@yahoogroups.com>
Sent: Friday, November 25, 2011 10:35 AM
Subject: Re: [DynoMotion] Re: CNC Lathe
Thanks Tom, Im looking for options to bring one kflop+kanalog to Argentina no start testing, hopefully next week will have the logistics arranged. Do you guys sell 2000ppr encoders? besides the encoder I already have everything to make a test bed axis.
Thanks!
Friday, November 25, 2011, 2:29:24 PM, Tom Kerekes wrote:
|
Hi Pablo,
Yes Dual Loop with KFLOP makes use of an axis channel for each loop. This means that all PID + filters + others are available for each loop. So if you wish to only use I for the outer loop like Galil does set the P and D and filter gains for that loop to zero. To use only P an D gains for the inner loop set the I and filters to zero. Having a separate axis channel for each loop is nice in that you can plot, test, and tune each of the loops separately rather than only looking at the overall result and not being able to see the performance of the inner loop. It is quite possible that for example using P gain in both loops will give a slightly higher performance system.
Regarding your ball screw choice: Backlash is normally a fixed amount and a function of the "nut" itself. Where lead errors and run out as a function of length are more related to the length of the screw. So I don't really understand why the backlash would be specified as a function of length. In systems with glass scale feedback accuracy is determined by the glass scale so you wouldn't benefit from a more accurate lead screw. The only requirement on the leadscrew is that the motion is smooth (without backlash).
Regards
TK
From: Pablo Untroib <pablo@...>
To: oxforduniv <DynoMotion@yahoogroups.com>
Sent: Thursday, November 24, 2011 2:03 PM
Subject: Re: [DynoMotion] Re: CNC Lathe
Thursday, November 24, 2011, 6:07:42 PM, oxforduniv wrote:
|
There is a on line video tutorial that explains this on the Galil site.
Al. |
Just finished listening to this presentation http://www.galilmc.com/learning/tutorial-display.php?tutorial=dual-loop-compensation-methods I guess its the one you directed me to.
Also, just in time I must say, just read Tom reply with the car example.
SO :)
My plan is to use not the worst neither the best ballscrews, for budget purposes I need to settle in the middle, typical backslash for a Hiwin ballcrew is quoted at less or equal to 0.03mm every 300mm of screw. Not a lot but its easily measurable.
Is there a way, remember I do not program in C, to implement the improved feedback loop as described in the video presentation (page 11) by separating the PD for the motor and then the I for the load via the glass scales?
Thanks!!
--
Pablo E. Untroib
-------
Suipacha 834 5to. "A"
(1008) Buenos Aires, Argentina
Tel. (+54-11) 4328-7500 / 7400
Phone USA: (260) UNTROIB
Twitter @PEU_AR
PGP Key ID: : 0xFC7177AF |
--
Pablo E. Untroib
-------
Suipacha 834 5to. "A"
(1008) Buenos Aires, Argentina
Tel. (+54-11) 4328-7500 / 7400
Phone USA: (260) UNTROIB
Twitter @PEU_AR
PGP Key ID: : 0xFC7177AF |
--
Pablo E. Untroib
-------
Suipacha 834 5to. "A"
(1008) Buenos Aires, Argentina
Tel. (+54-11) 4328-7500 / 7400
Phone USA: (260) UNTROIB
Twitter @PEU_AR
PGP Key ID: : 0xFC7177AF
|
--
Pablo E. Untroib
-------
Suipacha 834 5to. "A"
(1008) Buenos Aires, Argentina
Tel. (+54-11) 4328-7500 / 7400
Phone USA: (260) UNTROIB
Twitter @PEU_AR
PGP Key ID: : 0xFC7177AF
|
|
| Group: DynoMotion |
Message: 2435 |
From: Tom Kerekes |
Date: 11/25/2011 |
| Subject: Re: CNC Lathe |
Hi Pablo,
Our end with Kanalog has screw terminals. But the encoder has a 10 pin connector. Looks like if you select the "Standard" differential pinout then this cable should go with it:
TK
| Group: DynoMotion |
Message: 2543 |
From: Pablo Untroib |
Date: 12/2/2011 |
| Subject: Re: CNC Lathe |
Re: [DynoMotion] Re: CNC Lathe
everything ordered, kflop/kanalog/encoders/slotted sensors/etc can't wait to get them here. Expect my flood of questions by Christmas :)
Thanks Tom!
Friday, November 25, 2011, 8:17:50 PM, Tom Kerekes wrote:
|
Hi Pablo,
Our end with Kanalog has screw terminals. But the encoder has a 10 pin connector. Looks like if you select the "Standard" differential pinout then this cable should go with it:
http://www.usdigital.com/products/cables-connectors/cables/10-pin/ca-c10-sh-nc
TK
From: Pablo Untroib <pablo@...>
To: Tom Kerekes <DynoMotion@yahoogroups.com>
Sent: Friday, November 25, 2011 2:59 PM
Subject: Re: [DynoMotion] Re: CNC Lathe
I declare myself absolutelly ignorant regarding encoders, Im a plain stepper guy :)
So differential it is, do these require custom cabling, or I can wire them myself? I see they offer cables.
Friday, November 25, 2011, 5:55:37 PM, Tom Kerekes wrote:
|
Hi Pablo,
Differential signals are much less noise sensitive than single ended signals. So if you plan to use our Kanalog or SnapAmp boards which have differential inputs it would be worth a few extra $ to go that route. Those E6 models appear to have very weak single ended drive capability which makes going single ended an even poorer choice. It seems the Gecko G320Xs don't accept differential inputs so they recommend a pull up resistor to help with the single ended drive.
Regards
TK
From: Pablo Untroib <pablo@...>
To: Tom Kerekes <DynoMotion@yahoogroups.com>
Sent: Friday, November 25, 2011 12:19 PM
Subject: Re: [DynoMotion] Re: CNC Lathe
Im configuring the E6 model, it says if you use 6 feet or less the single ended model is commonly used. Is there a non obvious reason to use the differential version?
I recall this thread at cnczone regarding encoders http://www.cnczone.com/forums/gecko_drives/89079-new_gecko_g320x_startup_problems.html and geckodrives, does the same applies to kflop?
Friday, November 25, 2011, 3:59:15 PM, Tom Kerekes wrote:
|
Hi Pablo,
No we don't sell encoders. Here might be one source:
http://www.usdigital.com/
If possible get the differential output option.
Regards
TK
From: Pablo Untroib <pablo@...>
To: Tom Kerekes <DynoMotion@yahoogroups.com>
Sent: Friday, November 25, 2011 10:35 AM
Subject: Re: [DynoMotion] Re: CNC Lathe
Thanks Tom, Im looking for options to bring one kflop+kanalog to Argentina no start testing, hopefully next week will have the logistics arranged. Do you guys sell 2000ppr encoders? besides the encoder I already have everything to make a test bed axis.
Thanks!
Friday, November 25, 2011, 2:29:24 PM, Tom Kerekes wrote:
|
Hi Pablo,
Yes Dual Loop with KFLOP makes use of an axis channel for each loop. This means that all PID + filters + others are available for each loop. So if you wish to only use I for the outer loop like Galil does set the P and D and filter gains for that loop to zero. To use only P an D gains for the inner loop set the I and filters to zero. Having a separate axis channel for each loop is nice in that you can plot, test, and tune each of the loops separately rather than only looking at the overall result and not being able to see the performance of the inner loop. It is quite possible that for example using P gain in both loops will give a slightly higher performance system.
Regarding your ball screw choice: Backlash is normally a fixed amount and a function of the "nut" itself. Where lead errors and run out as a function of length are more related to the length of the screw. So I don't really understand why the backlash would be specified as a function of length. In systems with glass scale feedback accuracy is determined by the glass scale so you wouldn't benefit from a more accurate lead screw. The only requirement on the leadscrew is that the motion is smooth (without backlash).
Regards
TK
From: Pablo Untroib <pablo@...>
To: oxforduniv <DynoMotion@yahoogroups.com>
Sent: Thursday, November 24, 2011 2:03 PM
Subject: Re: [DynoMotion] Re: CNC Lathe
Thursday, November 24, 2011, 6:07:42 PM, oxforduniv wrote:
|
There is a on line video tutorial that explains this on the Galil site.
Al. |
Just finished listening to this presentation http://www.galilmc.com/learning/tutorial-display.php?tutorial=dual-loop-compensation-methods I guess its the one you directed me to.
Also, just in time I must say, just read Tom reply with the car example.
SO :)
My plan is to use not the worst neither the best ballscrews, for budget purposes I need to settle in the middle, typical backslash for a Hiwin ballcrew is quoted at less or equal to 0.03mm every 300mm of screw. Not a lot but its easily measurable.
Is there a way, remember I do not program in C, to implement the improved feedback loop as described in the video presentation (page 11) by separating the PD for the motor and then the I for the load via the glass scales?
Thanks!!
--
Pablo E. Untroib
-------
Suipacha 834 5to. "A"
(1008) Buenos Aires, Argentina
Tel. (+54-11) 4328-7500 / 7400
Phone USA: (260) UNTROIB
Twitter @PEU_AR
PGP Key ID: : 0xFC7177AF |
--
Pablo E. Untroib
-------
Suipacha 834 5to. "A"
(1008) Buenos Aires, Argentina
Tel. (+54-11) 4328-7500 / 7400
Phone USA: (260) UNTROIB
Twitter @PEU_AR
PGP Key ID: : 0xFC7177AF |
--
Pablo E. Untroib
-------
Suipacha 834 5to. "A"
(1008) Buenos Aires, Argentina
Tel. (+54-11) 4328-7500 / 7400
Phone USA: (260) UNTROIB
Twitter @PEU_AR
PGP Key ID: : 0xFC7177AF |
--
Pablo E. Untroib
-------
Suipacha 834 5to. "A"
(1008) Buenos Aires, Argentina
Tel. (+54-11) 4328-7500 / 7400
Phone USA: (260) UNTROIB
Twitter @PEU_AR
PGP Key ID: : 0xFC7177AF
|
--
Pablo E. Untroib
-------
Suipacha 834 5to. "A"
(1008) Buenos Aires, Argentina
Tel. (+54-11) 4328-7500 / 7400
Phone USA: (260) UNTROIB
Twitter @PEU_AR
PGP Key ID: : 0xFC7177AF
|
|
| Group: DynoMotion |
Message: 2549 |
From: Tom Kerekes |
Date: 12/2/2011 |
| Subject: Re: CNC Lathe |
Happy Holidays!
| Group: DynoMotion |
Message: 2710 |
From: Pablo Untroib |
Date: 12/13/2011 |
| Subject: Re: CNC Lathe |
Re: [DynoMotion] Re: CNC Lathe
Dear Tom, I spent the whole day cutting/welding/drilling/assembling a test bed for kflop with linear guides, ballscrew and glass scales, to have them ready for when the boards/encoders arrive. I have another question:
Once I have an axis properly configured, tested and verified, duplicating the config to have two axis do the same is just a matter of copyng the config for the new axis, adjust travels and be done or there are other interaxis interactions that need to be considered too?
Thanks!
Friday, November 25, 2011, 5:55:37 PM, Tom Kerekes wrote:
|
Hi Pablo,
Differential signals are much less noise sensitive than single ended signals. So if you plan to use our Kanalog or SnapAmp boards which have differential inputs it would be worth a few extra $ to go that route. Those E6 models appear to have very weak single ended drive capability which makes going single ended an even poorer choice. It seems the Gecko G320Xs don't accept differential inputs so they recommend a pull up resistor to help with the single ended drive.
Regards
TK
From: Pablo Untroib <pablo@...>
To: Tom Kerekes <DynoMotion@yahoogroups.com>
Sent: Friday, November 25, 2011 12:19 PM
Subject: Re: [DynoMotion] Re: CNC Lathe
Im configuring the E6 model, it says if you use 6 feet or less the single ended model is commonly used. Is there a non obvious reason to use the differential version?
I recall this thread at cnczone regarding encoders http://www.cnczone.com/forums/gecko_drives/89079-new_gecko_g320x_startup_problems.html and geckodrives, does the same applies to kflop?
Friday, November 25, 2011, 3:59:15 PM, Tom Kerekes wrote:
|
Hi Pablo,
No we don't sell encoders. Here might be one source:
http://www.usdigital.com/
If possible get the differential output option.
Regards
TK
From: Pablo Untroib <pablo@...>
To: Tom Kerekes <DynoMotion@yahoogroups.com>
Sent: Friday, November 25, 2011 10:35 AM
Subject: Re: [DynoMotion] Re: CNC Lathe
Thanks Tom, Im looking for options to bring one kflop+kanalog to Argentina no start testing, hopefully next week will have the logistics arranged. Do you guys sell 2000ppr encoders? besides the encoder I already have everything to make a test bed axis.
Thanks!
Friday, November 25, 2011, 2:29:24 PM, Tom Kerekes wrote:
|
Hi Pablo,
Yes Dual Loop with KFLOP makes use of an axis channel for each loop. This means that all PID + filters + others are available for each loop. So if you wish to only use I for the outer loop like Galil does set the P and D and filter gains for that loop to zero. To use only P an D gains for the inner loop set the I and filters to zero. Having a separate axis channel for each loop is nice in that you can plot, test, and tune each of the loops separately rather than only looking at the overall result and not being able to see the performance of the inner loop. It is quite possible that for example using P gain in both loops will give a slightly higher performance system.
Regarding your ball screw choice: Backlash is normally a fixed amount and a function of the "nut" itself. Where lead errors and run out as a function of length are more related to the length of the screw. So I don't really understand why the backlash would be specified as a function of length. In systems with glass scale feedback accuracy is determined by the glass scale so you wouldn't benefit from a more accurate lead screw. The only requirement on the leadscrew is that the motion is smooth (without backlash).
Regards
TK
From: Pablo Untroib <pablo@...>
To: oxforduniv <DynoMotion@yahoogroups.com>
Sent: Thursday, November 24, 2011 2:03 PM
Subject: Re: [DynoMotion] Re: CNC Lathe
Thursday, November 24, 2011, 6:07:42 PM, oxforduniv wrote:
|
There is a on line video tutorial that explains this on the Galil site.
Al. |
Just finished listening to this presentation http://www.galilmc.com/learning/tutorial-display.php?tutorial=dual-loop-compensation-methods I guess its the one you directed me to.
Also, just in time I must say, just read Tom reply with the car example.
SO :)
My plan is to use not the worst neither the best ballscrews, for budget purposes I need to settle in the middle, typical backslash for a Hiwin ballcrew is quoted at less or equal to 0.03mm every 300mm of screw. Not a lot but its easily measurable.
Is there a way, remember I do not program in C, to implement the improved feedback loop as described in the video presentation (page 11) by separating the PD for the motor and then the I for the load via the glass scales?
Thanks!!
--
Pablo E. Untroib
-------
Suipacha 834 5to. "A"
(1008) Buenos Aires, Argentina
Tel. (+54-11) 4328-7500 / 7400
Phone USA: (260) UNTROIB
Twitter @PEU_AR
PGP Key ID: : 0xFC7177AF |
--
Pablo E. Untroib
-------
Suipacha 834 5to. "A"
(1008) Buenos Aires, Argentina
Tel. (+54-11) 4328-7500 / 7400
Phone USA: (260) UNTROIB
Twitter @PEU_AR
PGP Key ID: : 0xFC7177AF |
--
Pablo E. Untroib
-------
Suipacha 834 5to. "A"
(1008) Buenos Aires, Argentina
Tel. (+54-11) 4328-7500 / 7400
Phone USA: (260) UNTROIB
Twitter @PEU_AR
PGP Key ID: : 0xFC7177AF
|
--
Pablo E. Untroib
-------
Suipacha 834 5to. "A"
(1008) Buenos Aires, Argentina
Tel. (+54-11) 4328-7500 / 7400
Phone USA: (260) UNTROIB
Twitter @PEU_AR
PGP Key ID: : 0xFC7177AF
|
|
| Group: DynoMotion |
Message: 2711 |
From: Tom Kerekes |
Date: 12/13/2011 |
| Subject: Re: CNC Lathe |
Hi Pablo,
The simplest way to duplicate an axis channel settings is to save the channel (Save Channel Button on the Config/Flash Screen).
Then switch to the new channel and load the channel (Load Channel Button on the Config/Flash Screen).
You will still need to change the Input/Output channels to the appropriate values and possibly Limit input bit numbers.
Regards,
TK
| Group: DynoMotion |
Message: 2712 |
From: Pablo Untroib |
Date: 12/13/2011 |
| Subject: Re: CNC Lathe |
Re: [DynoMotion] Re: CNC Lathe
cool, that's what I imagined, no extra configs inherent to having two axis instead of one. Thanks
Tuesday, December 13, 2011, 7:45:41 PM, Tom Kerekes wrote:
|
Hi Pablo,
The simplest way to duplicate an axis channel settings is to save the channel (Save Channel Button on the Config/Flash Screen).
Then switch to the new channel and load the channel (Load Channel Button on the Config/Flash Screen).
You will still need to change the Input/Output channels to the appropriate values and possibly Limit input bit numbers.
Regards,
TK
From: Pablo Untroib <pablo@...>
To: Tom Kerekes <DynoMotion@yahoogroups.com>
Sent: Tuesday, December 13, 2011 2:23 PM
Subject: Re: [DynoMotion] Re: CNC Lathe
Dear Tom, I spent the whole day cutting/welding/drilling/assembling a test bed for kflop with linear guides, ballscrew and glass scales, to have them ready for when the boards/encoders arrive. I have another question:
Once I have an axis properly configured, tested and verified, duplicating the config to have two axis do the same is just a matter of copyng the config for the new axis, adjust travels and be done or there are other interaxis interactions that need to be considered too?
Thanks!
Friday, November 25, 2011, 5:55:37 PM, Tom Kerekes wrote:
|
Hi Pablo,
Differential signals are much less noise sensitive than single ended signals. So if you plan to use our Kanalog or SnapAmp boards which have differential inputs it would be worth a few extra $ to go that route. Those E6 models appear to have very weak single ended drive capability which makes going single ended an even poorer choice. It seems the Gecko G320Xs don't accept differential inputs so they recommend a pull up resistor to help with the single ended drive.
Regards
TK
From: Pablo Untroib <pablo@...>
To: Tom Kerekes <DynoMotion@yahoogroups.com>
Sent: Friday, November 25, 2011 12:19 PM
Subject: Re: [DynoMotion] Re: CNC Lathe
Im configuring the E6 model, it says if you use 6 feet or less the single ended model is commonly used. Is there a non obvious reason to use the differential version?
I recall this thread at cnczone regarding encoders http://www.cnczone.com/forums/gecko_drives/89079-new_gecko_g320x_startup_problems.html and geckodrives, does the same applies to kflop?
Friday, November 25, 2011, 3:59:15 PM, Tom Kerekes wrote:
|
Hi Pablo,
No we don't sell encoders. Here might be one source:
http://www.usdigital.com/
If possible get the differential output option.
Regards
TK
From: Pablo Untroib <pablo@...>
To: Tom Kerekes <DynoMotion@yahoogroups.com>
Sent: Friday, November 25, 2011 10:35 AM
Subject: Re: [DynoMotion] Re: CNC Lathe
Thanks Tom, Im looking for options to bring one kflop+kanalog to Argentina no start testing, hopefully next week will have the logistics arranged. Do you guys sell 2000ppr encoders? besides the encoder I already have everything to make a test bed axis.
Thanks!
Friday, November 25, 2011, 2:29:24 PM, Tom Kerekes wrote:
|
Hi Pablo,
Yes Dual Loop with KFLOP makes use of an axis channel for each loop. This means that all PID + filters + others are available for each loop. So if you wish to only use I for the outer loop like Galil does set the P and D and filter gains for that loop to zero. To use only P an D gains for the inner loop set the I and filters to zero. Having a separate axis channel for each loop is nice in that you can plot, test, and tune each of the loops separately rather than only looking at the overall result and not being able to see the performance of the inner loop. It is quite possible that for example using P gain in both loops will give a slightly higher performance system.
Regarding your ball screw choice: Backlash is normally a fixed amount and a function of the "nut" itself. Where lead errors and run out as a function of length are more related to the length of the screw. So I don't really understand why the backlash would be specified as a function of length. In systems with glass scale feedback accuracy is determined by the glass scale so you wouldn't benefit from a more accurate lead screw. The only requirement on the leadscrew is that the motion is smooth (without backlash).
Regards
TK
From: Pablo Untroib <pablo@...>
To: oxforduniv <DynoMotion@yahoogroups.com>
Sent: Thursday, November 24, 2011 2:03 PM
Subject: Re: [DynoMotion] Re: CNC Lathe
Thursday, November 24, 2011, 6:07:42 PM, oxforduniv wrote:
|
There is a on line video tutorial that explains this on the Galil site.
Al. |
Just finished listening to this presentation http://www.galilmc.com/learning/tutorial-display.php?tutorial=dual-loop-compensation-methods I guess its the one you directed me to.
Also, just in time I must say, just read Tom reply with the car example.
SO :)
My plan is to use not the worst neither the best ballscrews, for budget purposes I need to settle in the middle, typical backslash for a Hiwin ballcrew is quoted at less or equal to 0.03mm every 300mm of screw. Not a lot but its easily measurable.
Is there a way, remember I do not program in C, to implement the improved feedback loop as described in the video presentation (page 11) by separating the PD for the motor and then the I for the load via the glass scales?
Thanks!!
--
Pablo E. Untroib
-------
Suipacha 834 5to. "A"
(1008) Buenos Aires, Argentina
Tel. (+54-11) 4328-7500 / 7400
Phone USA: (260) UNTROIB
Twitter @PEU_AR
PGP Key ID: : 0xFC7177AF |
--
Pablo E. Untroib
-------
Suipacha 834 5to. "A"
(1008) Buenos Aires, Argentina
Tel. (+54-11) 4328-7500 / 7400
Phone USA: (260) UNTROIB
Twitter @PEU_AR
PGP Key ID: : 0xFC7177AF |
--
Pablo E. Untroib
-------
Suipacha 834 5to. "A"
(1008) Buenos Aires, Argentina
Tel. (+54-11) 4328-7500 / 7400
Phone USA: (260) UNTROIB
Twitter @PEU_AR
PGP Key ID: : 0xFC7177AF |
--
Pablo E. Untroib
-------
Suipacha 834 5to. "A"
(1008) Buenos Aires, Argentina
Tel. (+54-11) 4328-7500 / 7400
Phone USA: (260) UNTROIB
Twitter @PEU_AR
PGP Key ID: : 0xFC7177AF
|
--
Pablo E. Untroib
-------
Suipacha 834 5to. "A"
(1008) Buenos Aires, Argentina
Tel. (+54-11) 4328-7500 / 7400
Phone USA: (260) UNTROIB
Twitter @PEU_AR
PGP Key ID: : 0xFC7177AF
|
|
| | | | | | | | | | | | | | | |
