Re: [CAD_CAM_EDM_DRO] Re: re:cleaning, lovejoy, black box tach, linear scales, making pulleys

Mariss Freimanis wrote:

> Jon, could you clear up something for me? Assuming a PID loop, it
> seems there is no simple way to keep the motor from dithering or
> bouncing between adjacent encoder edges due to the "infinite" DC gain
> of the integral term. Does your system exhibt this phenomena, and if
> it does, what is the frequency?

I have one axis that diters on rare occasions, I don't think this is a positioning
loop but a velocity loop oscillation, and I believe it is caused by a
mechanical resonance, as I can usually stop it by touching the helical
shaft couplers. The X axis never dithers at all. The Z axis does dither
at a 1-2 Hz rate due to the weight of the quill pulling down on the ballscrew.
You can see the dithering in the current monitor bar graph lights on the
servo amp, but as it is only one encoder count (= .000025") it doesn't
show up in the screen display, and you CERTAINLY can't feel it, even
touching the motor shaft.

>
> On an allied subject, quasi sine-cosine encoders are intriguing. If
> you strip the slope information for sine -45 deg to sine +45, it is
> pretty darn linear. Stitch these 90 degree segments together and you
> get continuous position feedback.

<snip>

> The result is an absolutely smooth, monotonic motion down to zero
> speed! On a much lesser note, the interface requires only a single
> signal line. I am going to pursue this further to see if it is
> practical.

HMMMMM! Now, that's an interesting thought! Wow! A derived tach with
more bandwidth than the encoder!

Jon