Re: "Y" axis problem

["Pete Cook" <kf2qd-at-terraworld.net>]

Your machine sounds like a variant of the plasma& oxyfuel machines that I
work on for a living -

On some of the older machines they got along fine with rather low encoder
counts - the best they could do at the time was a 8 millisecond(or longer)
refresh on the servo loop. For a faster servo loop you need to get a few
counts in there or you don't generate enough error to correct for. If you
could run the encoder off the motor, or even gear it faster than the motor
(ore counts in less time) you will have a more responsive system. Also look
at how you tuned the outputs - if you have only the proportional gain set
the control will seem like a spring - it only responds to error by
increasing the output voltage until the error causes some output voltage and
the axis stabilzes at some torque output. The integral and derivative work
to correct the out of position by offsetting the output voltage based on
errors over time. Increasing the encoder counts gives the controller
something to respond to.

Pete

>
> List,
>   I'm working on a two axis wood routing machine. It has a stg8 model 2
> card. Each axis is driven by a rack and pinion. Each pinion is driven by a
> series of two cog belts.
>   I have an encoder with a resolution of 2048, but it has to travel about
> three and a half inches to complete one revolution. This gives me a
> resolution of about 585 per inch.
>   On the "x" axis the encoder is coupled with the cog belt drive. This
axis
> work fine.
>   On the "y" axis the encoder has it's own rack and pinion making it more
> accurate as far as the Poisson of the router. My observation is that emc
> sends the dac a signal and before the amp can ramp up and move the axis
> enough to get some encoder counts it gets get a fowling error. After that
> the emc releases control of the axis and there is a small hop of the axis.
>   I have hooked a meter to the dac so that I could watch the signal. I
have
> applied voltage in place of the computer to the dac and the axis and
encoder
> run smoothly.
>   This machine was designed to move quickly three horse power on each axis
> and with out a high degree of accuracy. I may be way off on these
> observations but in any case I think you can get the idea of the problem.
>   I would appreciate any input on this matter,I have no idea on what to
try
> next.
>                                               Joe
>
>