Re: [CAD_CAM_EDM_DRO] Re: CNC Pulse jitter?

"j.guenther" wrote:

> Tim,
>
> I agree that the problem does exist even though I don't see it as much with
> my setup as you would on a higher performance machine. I really believe
> that a major part of the reason I can not get reliable performance at higher
> speeds is due in part to the jitters. For example, rapids run fine at 20
> IPM for some period of time. I shutdown EMC and come back later, now when I
> start the machine the rapids fail at 20 IPM but will work at lower speeds.
> It has to be the jitters in the pulse train causing this since nothing else
> has changed. I suppose the real answer to most of this problem in an
> external hardware based pulse generator and a new module for EMC that can
> handle the external pulse generation, like what Jon Elson is working on. I
> wonder if the Gecko G210 with the step multiplier / regeneration capability
> can help this problem any? Maybe a simpler solution to the problem is a
> pulse regenerator that takes the incoming pulse stream, not caring about how
> smooth or regular it is and produces a nice smooth output stream.

Not necessarily. It is beginning to look like some motherboards may have
some BIOS function that uses the NMI interrupt for some kind of housekeeping
(possibly thermal safety or energy saving mode) that gets serviced ahead of
the real time scheduler. My board would mask this problem by generating smooth
velocities, but if the CPU is not regularly watching the motion system, then it
could lead to watchdog timeouts, poorly cut contours, passing the endpoints
of commanded motion, etc. Not a whole lot better!

If the amount of jitter you see varies from run to run, I suspect you may have
the problem I describe above.

Jon