Re: phase stepping



Hi Ray

Thans for your help, I've now made some progress.
I grabbed the BDI-209 distribution to get newer code.
That did not install on my poor old P-pro 150, so I manually 
installed rpm's for utils, kernel, rtlinux, rcslib and emc, ending up
with emc-2.07-rtl3.0.i586.rpm onto redhat 6.2

This did not actually change anything much :(

So I read and tweaked everything I could find.

My main problem was output scale, which I had left at the original
value of 1.   getting that to match the input scales (of ~300) 
resulted in P values behaving in the region of 10.
Everything else seems to tweak ok.  Period works at 50us (not 16us)

now - only issue seems to be poor output waveforms, which are not
square. this severely limits the max speed. I'm seeing 3:1 
mark space ratio on the motor.
investigating - there is a slew problem in the driver - I'll 
fix that!  Then onto the welder, make the mechanicals.

A distribution ini file with more comments and unit explainations
might help.   Can the items only used for simulation be moved into a
different file?   Its a little long, can you have multiple [AXIS_0] 
sections? Also if output scale is supposed to match input scale, why
does it exist?

thanks, john



On Fri, Sep 07, 2001 at 10:19:31AM -0400, Ray wrote:
> 
> 
> Hello John.
> 
> As I understand freqmod, it figures out how many steps it has to go in what
> length of time and sets a rate generating routine.  Steps that can't be
> spit out immediately get put into a buffer.  To many pulses in this buffer
> and you get the following error.  In the software this looks a lot like a
> servo system and in fact uses the PID tuning stuff.
> 
> 1 - I suspect that since your problem is with several quick starts that your
> acceleration is to high.  Try setting it between 2-5 and see if that makes
> a difference.
> 
> 2 - EMC seems to have a problem with scheduling pulses that it's placed in
> the buffer back into the pulse stream.  I tend to think of it this way.  It
> may not be completely true. Some combination of the [EMCMOT] PERIOD, and
> [TRAJ] MAX_VELOCITY,  parameters sets the upper bound of the frequency
> generator.  Anytime the trajectory planner wants to go faster than that
> upper bound, it starts piling up pulses in the buffer.  Then when the
> frequency generator has slack time, it begins to feed these pulses back
> into the stream.  I believe that you can create some slack overhead by
> setting [TRAJ] MAX_VELOCITY larger by a few percent than you set the same
> variable in each of the axis definitions. 
> 
> 3 - You can also rather easily overrun the frequency generator by demanding
> a lot of pulses per unit with INPUT_SCALE.  This problem becomes one of the
> mechanical/electrical design of the device you want to run with the EMC. 
> Design a high speed wood or foam router with much fewer pulses per unit
> than you would an EDM machine.
> 
> BTW - Since the release of January, we have found a bug in the code that
> sometimes trashes the EMC when you first issue a motion command.  I'm sure
> that someone on the list can describe exactly where in the source code the
> changes need to be made to fix this release.  This was more of a problem
> with steppermod than freqmod but you might watch for it.
> 
> Hope this helps
> 
> Ray
> 
> 
> On Fri, 07 Sep 2001, John wrote:
> > Hi
> > I'm a hardware designer trying to build a router table.
> > 
> > I've set up redhat 6.2, rtlinux 2.2a, and the January emc 
> > sourceforge distribution. 
> > 
> > I have the simulations running ok, now added  several 
> > steppers running  phase stepping using freqmod.o  
> > (phase step both matches my stepper driver, and seems
> >  more noiseproof and sensible).
> > 
> > I'm having trouble in two areas:
> > 
> > when jogging with the arrow keys, hitting the key several times close
> > together generates a "following error" and reverts to ESTOP reset.
> > Can someone explain (or point to documenation) to help me understand 
> > which ini file veriables are relevant to this mode, and what they do?
> > There are kind of a lot, most seem to be servomotor related.
> > 
> > secondly, I'm using the minimillio as EMCIO.  Is this wise? I have no 
> > parallel port hardware at 0x278. 
> > (but this could change - as a PS1 video game controller handpiece looks 
> >  an interesting option here..) 
> > 
> > regards, john
> 
> 



Date Index | Thread Index | Back to archive index | Back to Mailing List Page

Problems or questions? Contact