Re: [Emc-users] Trajectory planning and other topics from a, EMC(LinuxCNC) newbie (TheNewbie)

[Viesturs L. <vie...@gm...>]

2012/3/31 cogoman <co...@op...>:
> From: Viesturs L?cis <vie...@gm...> Subject: Re: [Emc-users]
>> This will run in a little off-topic by me...
>> You can run heavy duty machines with steppers, but then the
>> performance will suffer - You will have to leave big safety margin in
>> terms of load to motors (either move slower or use way much larger
>> motors) as currently there is no safe and reliable way to recover "on
>> the fly" from lost steps.
>> Last year I had a discussion with Festo people in Hannouver Messe
>> about this topic - they are offering pseido-servo control system,
>> which consists of:
>> 1) stepper motors
>> 2) encoders on motors for feedback
>> 3) controller, in which error in feedback loop is fixed in classical
>> manner - add throttle, when motor is falling behind
>>
>> I managed to left their representative without any arguments as this
>> approach totally sucks - if the load on motor is large enough for it
>> to start losing steps, then there is no sense to increase the RPM - it
>> will continue to lose steps.
>> It has been proposed on this mailing list that completely new approach
>> would required - all the remaining joints should slow down so that the
>> one with lost steps remains in the correct place with respect to
>> other.
>> I guess some special drive, that could increase the current for a
>> short time to increase motor torque, would allow to use the classical
>> approach, but I will leave that for people, who actually do understand
>> something about stepper drives and the principles, how they work.
> Seems to me the only way to practically set up a stepper system with
> feedback would be to run linuxcnc as if it were driving servos, but set
> up the stepgen as velocity mode, then when a feedback error occurs
> instead of increasing the velocity signal, it would feed back to the
> slider that controls percentage of commanded rate, and slow down the
> commanded rate, while lighting an obnoxious light on the front panel to
> alert you that you might have a tool that's about to snap.

Hmmm, I think that this idea is worth testing. But I guess that it
should combine both - traditional approach of adding some more
movement to catch up for lost steps and this new suggestion of
adjusting feed override value to slow down the other joints.

I guess it would not be that easy to implement in HAL, probably with
some custom HAL component to watch following error values of the
joints and then adjust feed override, if following error is exceeding
certain threshold.
I also do not know, what would be appropriate approach to tune PID
loops for steppers. Somehow I think that it would need only P
parameter, but am not sure.


>   I don't know the inner workings well enough to say this is possible,
> but it seems like it should be with linuxcnc, and if someone were to
> implement it and document it, that would be a feather in the cap of
> linuxcnc.

I am not sure that it would be a magic cure for one of the biggest
drawbacks of stepper systems, because sudden slowdown of the movement
adds another risky situation for any of joints to lose some steps.

I guess that there is only one way to find out for sure, if this
approach makes some sense.

Viesturs