On Thu, May 30, 2013 at 10:35 AM, Tom Kerekes <tk@...> wrote:

 

Hi Steve,

One obvious possibility is noise on the inputs (or the E-Stop Input/Output? itself).  KFLOP's Inputs are pretty fast they will register microsecond or even nanaoseconds pulses.  Watching the screen with your eyes may not detect them.  The screen only samples and updates a few times per second.  You really need to look at them with a scope.  One test would be to disconnect (and ground) the inputs and see if the problem goes away.  That would prove whether it really is noise or something else like a software bug.   Something like a 0.1uf ceramic capacitor can be added at the KFLOP inputs to eliminate any narrow noise spikes.

Something else to be aware of is that Mach3 Brains run as a Windows's User Application (the same as Mach3) so it is still at the mercy of MS Windows.  It will respond quickly 99.99% of the time, but sometimes it may take a second or more.  This would not cause your problem, but better to do this type of thing in KFLOP where real-time response is guaranteed (which is one of the main reasons to use KFLOP).

HTH

Regards

TK

Group: DynoMotion	Message: 7555	From: Tom Kerekes	Date: 5/30/2013
Subject: Re: E-Stop issue - with multiple E-Stops connected - MACH E-Stops ra

Hi Steve, KFLOP doesn't have a "debounce" setting.  Of course they can be implemented in software.  There are some examples for push button switches.  But I don't think they are a good idea for reducing noise.  In some cases it will just reduce the frequency of the problem and make it more difficult to find. It sounds like your problem might have been the way you were driving the inputs.  Most of KFLOP's inputs are high impedance MOSFET inputs (megohms).  The must be driven both high and low.  If you were just pulling them high and then leaving them float the results would be unpredictable.  But the first 8 IO on JP4 and JP6 do have strong 150ohm pull down resistors.  So those should have worked ok by just sourcing current to pull them high. BTW it isn't recommended to pull any of the KFLOP inputs above 3.8V.  The 5V Tolerant inputs will not be damaged by 5V but it is still not recommended. So to answer your question: yes pull down resistors are required on the other IO that don't have internal pull down resistors.  You might consider filter capacitors as well. You could certainly write a simple C program to "OR" your signals together as well as latch any one that triggered the error. HTH Regards TK Group: DynoMotion Message: 7556 From: Steve Simpson Date: 5/30/2013 Subject: Re: E-Stop issue - with multiple E-Stops connected - MACH E-Stops ra Thanks again TOM. Good info.   I am not putting 5V on any inputs. My 24V drive interfaces are opto isolated and run at either 5V (or now also 3.3V) on the CNC side, but the signals drive chips and in some cases LEDs and the 5V guys are only about 3.8 or therabout by the time the signal gets to the Kflop. I have them limited well below the Kflop amp spec.   Anyway, I'm good for now. I'll look into writing some code for the Kflop when I have more time to play.   - Steve On Thu, May 30, 2013 at 11:40 AM, Tom Kerekes <tk@...> wrote:   Hi Steve, KFLOP doesn't have a "debounce" setting.  Of course they can be implemented in software.  There are some examples for push button switches.  But I don't think they are a good idea for reducing noise.  In some cases it will just reduce the frequency of the problem and make it more difficult to find. It sounds like your problem might have been the way you were driving the inputs.  Most of KFLOP's inputs are high impedance MOSFET inputs (megohms).  The must be driven both high and low.  If you were just pulling them high and then leaving them float the results would be unpredictable.  But the first 8 IO on JP4 and JP6 do have strong 150ohm pull down resistors.  So those should have worked ok by just sourcing current to pull them high. BTW it isn't recommended to pull any of the KFLOP inputs above 3.8V.  The 5V Tolerant inputs will not be damaged by 5V but it is still not recommended. So to answer your question: yes pull down resistors are required on the other IO that don't have internal pull down resistors.  You might consider filter capacitors as well. You could certainly write a simple C program to "OR" your signals together as well as latch any one that triggered the error. HTH Regards TK Group: DynoMotion Message: 7557 From: William Carothers Date: 5/31/2013 Subject: Re: E-Stop issue - with multiple E-Stops connected - MACH E-Stops ra I've had my kflop controlling analog drives estop Mach every so often.  But it is always always on rapid retract move command for my z or a axis,  so in my case I think it's random appearing estop is actually the follow error triggering the estop because my z is some times heavy and can't accelerate (and keep up) every time.  Does not happen often enough to cause me trouble but I suspect if I added a couple ticks of follow error alowed ( I know the actual parameter is called something different) it would completely go away. .1 uf caps are also gold B. On May 30, 2013, at 11:35 AM, Tom Kerekes <tk@...> wrote:   Hi Steve, One obvious possibility is noise on the inputs (or the E-Stop Input/Output? itself).  KFLOP's Inputs are pretty fast they will register microsecond or even nanaoseconds pulses.  Watching the screen with your eyes may not detect them.  The screen only samples and updates a few times per second.  You really need to look at them with a scope.  One test would be to disconnect (and ground) the inputs and see if the problem goes away.  That would prove whether it really is noise or something else like a software bug.   Something like a 0.1uf ceramic capacitor can be added at the KFLOP inputs to eliminate any narrow noise spikes. Something else to be aware of is that Mach3 Brains run as a Windows's User Application (the same as Mach3) so it is still at the mercy of MS Windows.  It will respond quickly 99.99% of the time, but sometimes it may take a second or more.  This would not cause your problem, but better to do this type of thing in KFLOP where real-time response is guaranteed (which is one of the main reasons to use KFLOP). HTH Regards TK Group: DynoMotion Message: 7558 From: Michael Rosenfield Date: 5/31/2013 Subject: Re: E-Stop issue - with multiple E-Stops connected - MACH E-Stops ra The following error is a very good point! I had the same issue with my Z and knee. These are relatively low performance servos, and don't accelerate rapidly. Of course, they don't need to. But I had to adjust the following error and lower the jerk settings, otherwise I'd get estops after some motion. Works fine now!   Regards, Michael Rosenfield   Group: DynoMotion Message: 7559 From: Steve Simpson Date: 5/31/2013 Subject: Re: E-Stop issue - with multiple E-Stops connected - MACH E-Stops ra You might look at counterbalancing the head with an air spring (or equiv). The mass will stay the same, but you remove the 'weight' and the servo has a much smaller job to do. In fact, if you use a 'strut' with higher force than the head weight, you will have eliminated all backlash in the Z axis for most operations.   - Steve On Fri, May 31, 2013 at 10:00 AM, Michael Rosenfield <mrosenfield@...> wrote:   The following error is a very good point! I had the same issue with my Z and knee. These are relatively low performance servos, and don't accelerate rapidly. Of course, they don't need to. But I had to adjust the following error and lower the jerk settings, otherwise I'd get estops after some motion. Works fine now!   Regards, Michael Rosenfield   To: DynoMotion@yahoogroups.com From: wcarrothers@... Date: Fri, 31 May 2013 05:51:36 -0400 Subject: Re: [DynoMotion] E-Stop issue - with multiple E-Stops connected - MACH E-Stops randomly   I've had my kflop controlling analog drives estop Mach every so often.  But it is always always on rapid retract move command for my z or a axis,  so in my case I think it's random appearing estop is actually the follow error triggering the estop because my z is some times heavy and can't accelerate (and keep up) every time.  Does not happen often enough to cause me trouble but I suspect if I added a couple ticks of follow error alowed ( I know the actual parameter is called something different) it would completely go away. .1 uf caps are also gold B. On May 30, 2013, at 11:35 AM, Tom Kerekes <tk@...> wrote:   Hi Steve, One obvious possibility is noise on the inputs (or the E-Stop Input/Output? itself).  KFLOP's Inputs are pretty fast they will register microsecond or even nanaoseconds pulses.  Watching the screen with your eyes may not detect them.  The screen only samples and updates a few times per second.  You really need to look at them with a scope.  One test would be to disconnect (and ground) the inputs and see if the problem goes away.  That would prove whether it really is noise or something else like a software bug.   Something like a 0.1uf ceramic capacitor can be added at the KFLOP inputs to eliminate any narrow noise spikes. Something else to be aware of is that Mach3 Brains run as a Windows's User Application (the same as Mach3) so it is still at the mercy of MS Windows.  It will respond quickly 99.99% of the time, but sometimes it may take a second or more.  This would not cause your problem, but better to do this type of thing in KFLOP where real-time response is guaranteed (which is one of the main reasons to use KFLOP). HTH Regards TK