Re: following error

[Jon Elson <jmelson-at-artsci.wustl.edu>]

garrett beaubien wrote:

> Hello,
>
> I have recently installed emc on my machine, and have
> been playing aroud with it for a little while.  What I
> am hoping to do is to get it working with a little
> single axis setup (servo motor, simple homebrew amp,
> quadrature counter, and DAC interfaced to ISA bus)
> that I have thrown togther.  When I fire up emc, and
> try to "home" the axis, and jog it, I get a follow
> error, and the machine goes into E-stop.  I assume
> this is because my PID parameters are way out of wack.

Maybe not.

> Is there some line in the code that I can comment out
> that will prevent this error (and subsequent E-Stop)?
> So that I can tune the loop properly?

What you want to change is in the .ini file, and you can
actually change it when EMC is running, with the settings
menu.  It is the FERROR and MIN_FERROR parameters.
MIN_FERROR sets a minimum following error that will
be tolerated.  FERROR is scaled by commanded velocity,
so that the full amount specified will be used at the max
velocity set in the [TRAJ] section.  So, if you will tolerate
.1" of following error at the max velocity, use that value.
When running at 1/2 max velocity, the actual error bound
would be .05".  When the error bound becomes so small
(at low velocity) that it would cause needless errors,
the MIN_FERROR parameter can be set to prevent
this.  Perhaps .01" would be a good setting. See
 http://www.isd.mel.nist.gov/projects/emc/emcsoft.html
for some more details on this parameter.

Note that some versions of EMC had a problem with these
fields being cleared to zero every time EMC is shut down
and restarted.  This would cause it to trip every time.

> Also, does anyone have any idea what I might use as
> good starting PID parameters?  right now, I just have
> the porportional gain set to 1.

As of recent history, the I and D terms didn't work too well,
as they assumed a steady state condition for all history,
which is certainly not what a CNC machine experiences.
Will Shackleford has made some changes, but I haven't
had a chance to test them, yet.  (I hate to try out new
stuff on a working system.)

As for best parameters, just set P to about 100 and see what
happens.  Too much P gain will cause VIOLENT rocking
of the axis.  When you have this stable and with sufficient
gain that the following error goes to about zero when idle,
you can turn the FF1 parameter to about 5.0 to improve
transient response.

You can check following error at idle by flipping back and
forth between commanded and actual position displays.

> One other thing I am curious about.  When I tried it
> the first time, it went into E-Stop and I reset.  Then
> I was able to turn the shaft and have the X axis show
> a position change on the EMC display.  I have not been
> able to reproduce this again.  Does anyone have any
> ideas?

I have had a couple experiences where the STG card got
incorrectly setup, and wouldn't work.  What are you using
for your quadrature counters?  (Of course, the obvious
things to check are encoder power and signal connections.)

Jon