Re: [CAD_CAM_EDM_DRO] E.M.C. PID Tuning - HELP !!!!

ja_erickson wrote:

> evening all,
>
> i have a few questions and oddities that have arisen during a tuning
> attempt of e.m.c. for my bridgeport series 2 retrofit project.the
> first thing that i have set is the 'p' value, and a value of 150
> seems to be the smallest reliable value that will work. i arrived at
> this number by starting at 50 then trying to run an axis. when using
> this number the axis operated very smoothly at feedrates of 1 to 5
> i.p.m. which is very desirable for me, unfortunately when approaching
> feedrates above this feed the motors lost counts and carried on which
> is no good for me. in order to provide a realistic value for
> feedrates and rapids i kept increasing the 'p' value and finally
> attained reliable operation at the 150 value. i also noted that by
> adding a value of 5 for the 'i' variable it seemed to help smooth out
> the motors stepping. by setting up the software with the above values
> the feedrates below 6 i.p.m have become very steppy. does anybody
> know of a better approach to tuning in order to smooth the slower
> feedrates while keeping the abilities of the higher 'p' variable?
> i must add that i have altered the acceleration rates to values that
> are conservative by most comparisons to other c.n.c. machines.what
> would be ideal for me is to have a dynamically adjusting 'p' value
> based on feedrates, is this possible?

Essentially, yes. That is what FF1 does, it adds more "boost" when
acceleration is requested. I have used this in a servo machine to
get the commanded and actual position traces to overlap extremely
well.

> the oddity that i have
> encountered happened when i added a backlash value into the axis.
> when i added a value the motors kept on stepping after the command
> was finished, they had no real direction or distance they were
> wandering in , they just looked like an epileptic having an orgasm?

The I term may have odd effects with the backlash compensation on.
You might try turning I back to zero.

>
> there are also variables such as
> deadband,cycletime,bias,ff0,ff1,ff2,ferror, and min_ferror that i
> cannot find an explaination for,can anybody tell me what these
> variables do and how to implement them into tuning this motor?

Deadband is a small distance over which no effort will be expended
to reduce position error to zero. Placing a small value in deadband
will stop "hunting" behavior, but the machine may not be exactly
where it is commanded. FF1 causes the requested velocity to be
increased when accelerating, to compensate for lag in the system.
FERROR and min_ferror are related to distance between commanded
and actual position, and are not really much use in open-loop systems.
If I remember right, FERROR is scaled by the velocity in user units/sec,
and MIN_FERROR is a static value. So, at a commanded velocity of
0.5 Inch/Sec, for MIN_FERROR was .050 and FERROR was .002,
then a difference between commanded and actual position of .050 + .001
= .051" would be permitted before causing a following error fault.
If you have encoders, this is much more important, as you can actually
have the system stop if any axis lags too far behind.

Most of this is documented in the .../emc/docs directory.

Jon