View Single Post
Old September 8th, 2012, 05:23 PM   #19
Lifetime Supporting Member
United States

OkiePC is online now
OkiePC's Avatar
Join Date: Mar 2005
Location: ENE of Nowhere Oklahoma
Posts: 9,452
Originally Posted by jonney View Post
I checked 5 IMCs with total 13 GML program. The Transducer/Mtr revolutions and Motor type in original program are being set no uniform.
I don't understand that last part. Are you installing a new motor for which you have other examples to follow or is this your first attempt with this motor/amp combination?

Originally Posted by jonney
I remember we have changed some encoders from 2048PPR to 1024PPR. And we never changed the Transducer/Mtr revolutions parameter,just changed K constant, but the motor is running OK.
If the gains are properly set, you can get away with this sometimes.

K-constant. that's the word I was looking for earlier. The K-constant is encoder pulses to engineering uints. The motor definitions page should match reality, but how it is wired can affect that. If you are using a quadrature encoder with 1024PPR, and the software is set for quadrature encoder it should match.

The signal itself should be verified when you tune.

You can do a "battery box test" in commander. I think it might be called something different in GML 3.x.

Basically it ignores the feedback and runs the amplifier/motor "open loop" you set the command voltage, and the servo follows it. During this test, you should see your k-constant and resulting position in engineering units match the behavior of the axis. If this is a linear axis, be very careful with this test. You can slam a motor to full speed and have it running into an end limit faster than the safety circuit can shut her down, all with a careless click...

If you do the marker test, you can verify your encoder is there and making signals. It doesn't prove the settings for motor rpm are correct, so your crazy acting motor may be simply due to this. I would set this up, run the battery box test and ensure that all is well and smooth and happy with the motor at several different speeds, then go back and tune the drive.

Autotuning is tricky with A/B Gml. If you enter the wrong values for the tests it will fail miserably and make a perfectly gently servo into a nervous wreck.

I had several systems that simply could not be autotuned with any tuning values...we simply had to turn the thing down to a crawl, and manually tune the machine over several hours, then record those values and they could be manually entered as good starting points on subsequent machines.

Hope this helps.
It's not all the variables I am most concerned with, it's the undiscovered constants.
  Reply With Quote