Tom,

What I was suggesting was that if my knee servo was perhaps set for a bit too much P, not enough D, perhaps Mach3, since it just slams on the acceleration, would get away with working OK, but with the more gentle acceleration at the ends of a move that KFlop creates, the deficiencies in the turning become more of a problem. I'm kind of grasping at straws, but I do know in 3 years with Mach3, I never once lost position or faulted a servo without running into something. But it's happening with some frequency with the KFlop. On the knee, I was able to minimize it by re-tuning the servo, turning the gain and damping down. The faults are still happening, just less often. I just finished jury-rigging a way to get the encoders into KFlop, so I can see what's going on. I'll run that in the AM. I have done captures, and haven't seen anything unexpected, so this doesn't look like the KFlop is doing anything wrong - my machine just doesn't seem to like what it is doing.

My app, BTW, has performed flawlessly all day!

Regards,
Ray L.

--- In DynoMotion@yahoogroups.com, Tom Kerekes <tk@...> wrote:
>
> Hi Ray,
>  
> The S-curve acceleration should be smoother with less overshoot.  Was this where you modified the Interpreter to apply offsets in some non-standard manner?
>  
> Regardless, you probably should capture the data to see exactly what is happening.
>  
> Regards
> TK
>
> From: himykabibble <jagboy@...>
> To: DynoMotion@yahoogroups.com
> Sent: Monday, February 6, 2012 4:59 PM
> Subject: [DynoMotion] Re: What's Wrong With My Math?
>
>
>  
> Ah! Forgot about that! Yes, I had my init thread running, in addition to the capture thread. I knew there was a logical explanation.
>
> Things haven't gone quite as smoothly as I'd hoped today. Things are mostly working, but I'm getting random faults on the knee when toolcomp is applied. Does it make any sense that the servo would need to be re-tuned from the settings that worked well in Mach3? I'm wondering if perhaps I might have gotten away with too-high gain or perhaps too-low damping with Mach3's infinite jerk but the S-curve acceleration KFlop is doing might cause faults at the beginning and end of a move due to overshoot. That would be consistent with what I'm seeing and hearing - once an axis starts moving, it keeps moving just fine, but I'm seeing occasional random servo faults at the very start and very end of moves. I *never* had faults in Mach3.
>
> Regards,
> Ray L.
>
> --- In DynoMotion@yahoogroups.com, Tom Kerekes <tk@> wrote:
> >
> > Hi Ray,
> >  
> > I'm guessing you have two user Threads running.  The basic Time Slice for KFLOP is 90us.  With one System Thread (that always runs) and your two User threads then each thread runs every 3 time slices.
> >  
> > 80,000 counts/sec x 90us x 3 = 21.6 counts
> >  
> > See:
> >  
> > http://www.dynomotion.com/Help/Multitasking.htm
> >  
> > Regards
> > TK
> >
> > From: himykabibble <jagboy@>
> > To: DynoMotion@yahoogroups.com
> > Sent: Monday, February 6, 2012 4:40 PM
> > Subject: [DynoMotion] What's Wrong With My Math?
> >
> >
> >  
> > I'm doing some data captures from ch?->Dest, but when I try to calculate velocity from those numbers, I'm not getting a sensible result. My A axis (knee) is 96000/inch, velocity limit 50 IPM. But if I do a capture when the knee is moving at full speed, I see a difference of 21 counts between successive data points. By my calculation (assuming the capture rate is roughly 10,000 samples/second):
> >
> > MaxVel => 50 IPM => 50 * 96000 = 4,800,000 counts/minute = 80,000 counts/second = 8 counts/timeslice
> >
> > So why am I seeing 21 counts/timeslice and not 8?? Where am I going wrong?
> >
> > Regards,
> > Ray L.
> >
>

Tom,

Is there a reliable way to tell if an axis is moving in response to a G0, G43, G49, etc.? I modified CaptureXYZToFile.c to include the A axis, but the test for motion starting in the initial while loop did not work for G0s, G43 and G49. I tried using CheckDone(3), but that causes frequent false triggers, and I keep capturing whole buffers of data when the axis isn't moving.

I also tried putting the whole capture in a loop, so I could capture the start of all moves, but for some reason the thread nearly almost crashes after just a few captures - usually just 2 or 3, rarely as many as 4 or 5. Is there some interaction between the capture code and printfs, PCComm, or other things happening in other threads that would explain the crashes? It seems to be crashing in the middle of writing the file.

Regards,
Ray L.

--- In DynoMotion@yahoogroups.com, "himykabibble" <jagboy@...> wrote:
>
> Tom,
>
> What I was suggesting was that if my knee servo was perhaps set for a bit too much P, not enough D, perhaps Mach3, since it just slams on the acceleration, would get away with working OK, but with the more gentle acceleration at the ends of a move that KFlop creates, the deficiencies in the turning become more of a problem. I'm kind of grasping at straws, but I do know in 3 years with Mach3, I never once lost position or faulted a servo without running into something. But it's happening with some frequency with the KFlop. On the knee, I was able to minimize it by re-tuning the servo, turning the gain and damping down. The faults are still happening, just less often. I just finished jury-rigging a way to get the encoders into KFlop, so I can see what's going on. I'll run that in the AM. I have done captures, and haven't seen anything unexpected, so this doesn't look like the KFlop is doing anything wrong - my machine just doesn't seem to like what it is doing.
>
> My app, BTW, has performed flawlessly all day!
>
> Regards,
> Ray L.
>
>
> --- In DynoMotion@yahoogroups.com, Tom Kerekes <tk@> wrote:
> >
> > Hi Ray,
> >  
> > The S-curve acceleration should be smoother with less overshoot.  Was this where you modified the Interpreter to apply offsets in some non-standard manner?
> >  
> > Regardless, you probably should capture the data to see exactly what is happening.
> >  
> > Regards
> > TK
> >
> > From: himykabibble <jagboy@>
> > To: DynoMotion@yahoogroups.com
> > Sent: Monday, February 6, 2012 4:59 PM
> > Subject: [DynoMotion] Re: What's Wrong With My Math?
> >
> >
> >  
> > Ah! Forgot about that! Yes, I had my init thread running, in addition to the capture thread. I knew there was a logical explanation.
> >
> > Things haven't gone quite as smoothly as I'd hoped today. Things are mostly working, but I'm getting random faults on the knee when toolcomp is applied. Does it make any sense that the servo would need to be re-tuned from the settings that worked well in Mach3? I'm wondering if perhaps I might have gotten away with too-high gain or perhaps too-low damping with Mach3's infinite jerk but the S-curve acceleration KFlop is doing might cause faults at the beginning and end of a move due to overshoot. That would be consistent with what I'm seeing and hearing - once an axis starts moving, it keeps moving just fine, but I'm seeing occasional random servo faults at the very start and very end of moves. I *never* had faults in Mach3.
> >
> > Regards,
> > Ray L.
> >
> > --- In DynoMotion@yahoogroups.com, Tom Kerekes <tk@> wrote:
> > >
> > > Hi Ray,
> > >  
> > > I'm guessing you have two user Threads running.  The basic Time Slice for KFLOP is 90us.  With one System Thread (that always runs) and your two User threads then each thread runs every 3 time slices.
> > >  
> > > 80,000 counts/sec x 90us x 3 = 21.6 counts
> > >  
> > > See:
> > >  
> > > http://www.dynomotion.com/Help/Multitasking.htm
> > >  
> > > Regards
> > > TK
> > >
> > > From: himykabibble <jagboy@>
> > > To: DynoMotion@yahoogroups.com
> > > Sent: Monday, February 6, 2012 4:40 PM
> > > Subject: [DynoMotion] What's Wrong With My Math?
> > >
> > >
> > >  
> > > I'm doing some data captures from ch?->Dest, but when I try to calculate velocity from those numbers, I'm not getting a sensible result. My A axis (knee) is 96000/inch, velocity limit 50 IPM. But if I do a capture when the knee is moving at full speed, I see a difference of 21 counts between successive data points. By my calculation (assuming the capture rate is roughly 10,000 samples/second):
> > >
> > > MaxVel => 50 IPM => 50 * 96000 = 4,800,000 counts/minute = 80,000 counts/second = 8 counts/timeslice
> > >
> > > So why am I seeing 21 counts/timeslice and not 8?? Where am I going wrong?
> > >
> > > Regards,
> > > Ray L.
> > >
> >
>