Re: [CAD_CAM_EDM_DRO] Re: EMC problem/question

currinh@... wrote:

>
> I just tried this on a 450MHz PIII. At values of PERIOD between 16 and
> 40 us I could get motion but it didn't seem right. It wasn't as smooth
> as steppermod and would stop, buzz and lose steps at times. Some
> values of PERIOD seem better but I didn't get a feeling a value could be
> found to give "good" performance.
>
> I tried a value of 10 us which nearly locked the machine. Once the
> start-up hit "loading freqmod.o" it slowed to a snails pace. I closed
> that window but this didn't speed things up, it was like the real time
> process was still going. I was able to do a normal shutdown but even
> this was very slow, all the way to power down.

Oh, of course! Freqmod.o simulates a servo, as I mentioned in a previous
comment. Well, that means all the PID servo calculations are performed,
and loop stability is an issue. If the P (proportional gain) is too high,
unstable
performance will be seen. that is what I think is happening there. Turn P
down to around 100 and try some motion. If that works OK, turn it up
in small increments (you'll get a bit snappier performance) until either
the steppers lose steps or the motors start to sound rough, then go back
down some safe interval. You may have to set a larger following error
limit as lower P causes more lag.

Jon