Re: handbook info -- autotuning

["Les Watts" <leswatts-at-rabun.net>]

Ray,

I have successfully used Zeigler-Nichols closed loop tuning with EMC. It
might be the best
method (there are many) because it requires minimum instrumentation.
Proportional gain
alone is  slowly increased until a small sustained oscillation is observed.
If one can simply
estimate that frequency that is enough (with the gain setting that caused
it) to get
calculated P,I, and D values.

Zeigler Nichols theory concerns linear systems- so a machine with lots of
non linearity
like stick slip or other non velocity related friction might not tune well.

It works well enough with my large ballscrew/linear rail machine.

It is also possible to get the parameters from a single bump or jog with any
stable machine
setting but you must acquire motion vs time. If the EMC data logging feature
worked
(at a proper sample rate) it could supply those parameters. I have tried
this but the
log file is always garbage for some reason.

A math description of ZN tuning is at:
http://www.shu.ac.uk/schools/eng/teaching/rw/pidtutorial.htm

It appears that EMC is actually fast enough to do some anti-vibration
control!
If the servo bandwidth is higher that some structural resonances the system
could (in theory) remain stable through and beyond them. I might play with
that later-but it's not pressing. I just always want to run the machines
faster
faster faster... until I run in to a thermal limit (burning up work).

Les
Leslie Watts
L M Watts Furniture
Tiger, Georgia USA
http://www.rabun.net/~leswatts/wattsfurniturewp.html
engineering page:
http://www.rabun.net/~leswatts/shop.html
----- Original Message -----
From: "Ray Henry" <rehenry-at-up.net>
To: "Multiple recipients of list" <emc-at-nist.gov>
Sent: Sunday, June 09, 2002 10:16 PM
Subject: handbook info -- autotuning


>
>
> List
>
> I found the following statement in the handbook.
>
> "There are several auto tuning and computational tuning procedures but
these
> have not been tested with EMC."
>
> Is this true?  Do any of you have further information on this subject.
>
> Let me know if you are willing to dig into this a bit further.
>
> Ray
>
>
>
>
>