Re: [CAD_CAM_EDM_DRO] Servo tunning?

Torsten wrote:

>I am just about done with my mill/drill conversion when I mounted
>up the pulley for the ametec servo motors 600 oz/in I did some
>test runs checking the strenght of the motors by holding on to
>the pulleys by hand when rottating slowly.
>I am very impressed with the tourqe on those motors I was not able
>to hold them by hand.
>When the motor is holding steady I was able to turn them about
>2-3 degree ether way before the tourqe gets up to full power.
>Logic would dictate that there has to be some movement but I would
>like to try to get this down some.
>

Are you using Gecko 320 or 340 drivers? I have observed this, too,
on a client's machine, but in his case, the motors deliver little torque
until they are moved 5 or more degrees!

>The motor is using 250 quadrature encoders is this part of what
>determends how much deflection one would have to put up with?
>

I assume that is a 250 cycle/rev encoder, giving 1000 quadrature counts
per rev? (250 counts/rev would be 62.5 cycles/rev, which can't exist, so
the first must be right.)

>There are 2 adjustment options on the driver board gain and torque
>limit, my question is will this be the right action to take and
>is there a procedure on how to do it?
>

You want the gain as high as you can get it without causing oscillation.
There ought to be a loop compensation adjustment (or damping). You
want the gain (basically a low-frequency open-loop gain setting) as
high as possible to keep the constant error as small as possible.
The loop compensation reduces high-frequency gain to keep the
servo loop from trying to get movement faster than the mechanical
inertia can respond.

>What amount of deflection would be normal for a well tuned system?
>Since this has about the same effect as backlash I wonder how much
>of it I will be left with?
>Do stepper motor have a advantage here it seams like they would not
>experience this problem?
>
>

I can say that my true servo system (encoder -> PC, PC -> DAC -> servo amp)
will hold within 2-3 encoder counts until you exceed the current limit of
the amp. So, that is .00005" x 3 = 0.00015". The Gecko drives are supposed
to work the same way, but I have seen results that indicate it may not in
some installations. We may still have problems in that setup that cause
this. The high resistance of the Ametek motors combined with low
power supply voltage may also be to blame.

Jon