Re: [CAD_CAM_EDM_DRO] Question tuning G320 servoamps.

Erik Reikes wrote:

> Hello all,
> I turned on my G320's last night and was playing around optimizing
> the settings. I'm not sure that I have a concise question, but I
> thought I'd share what I'm seeing so that others who might have had
> the same symptoms might comment.
>
> Some background : I am using some leeson 1/8HP DC motors with
> USDigital 100cpr encoders. It will hook up with a 2-1 reduction to a
> 10tpi leadscrew on my 3-1 machine. So, there are 8000 counts per
> inch which in theory will get me roughly 60ipm with Master5. The
> power supply is a Camtronics.
>
> First, is it realistic to try to setup the servos with no load
> attached? It seems to me that the load of the lead screws will
> significantly affect the dynamics of the feedback loop (i.e. it will
> change the transfer function and therefore the step and impulse
> responses significantly).
>
> Keep in mind I've just gotten started dialing them in and haven't
> attacked it with my Oscope yet. I didn't get it wired until about
> 2am last night.
>
> I'm noticing that when I give a jog command in Master5 that when
> the servo is accelerating it sort of does it in fits and starts. It
> basically starts moving then violently stops then continues
> acceleration. It does this a couple of times before it reaches
> speed.

This is likely due to the motor not having a load, and the compensation
loop not being able to handle the sudden acceleration the motor is
capable of.

> I suspect that it is the acceleration profile that Master5 is
> outputting. It seemed to get slightly better as I lowered the
> acceleration rate from Master5. Adjusting the gain and damping had
> some effect but even at the minimum settings it still diplayed this
> symptom.
>

The acceleration or rough acceleration ramp will excite any system
resonances you have. A higher-resolution encoder, or the provision
of a tachometer would help this problem. I WAS able to tune my analog
servos (fitted with DC tachs) to work fine with no load. They did require
some tuning when the motors were attached to the machine.

>
> When it is up to speed it is smooth. Tonight I will break out the
> scope and look for things like voltage droop on the input voltage
> line, and try to get a feel for what the pulse stream looks like. I
> played with the Master5 clibration tool, but I think I really need
> the whole machine hooked up for that to be useful.

I think the Geckos have an internal test point that allows you to see its
idea of instantaneous position error.

Jon