Update on closed-loop to the computer with Gecko 320

Since several people were interested in this, I thought I should
send an update. The problem is that the Gecko 320 (and 340)
are closed-loop servo drives, but the standard connection doesn't
allow the CNC computer to see the encoder position. There are
several advantages to getting the position info back to the CNC
computer. One is to have more flexible following error limits,
such as a sliding scale proportional to velocity. Another is
to be able to make plots of commanded vs actual position during
various moves, to verify the accuracy of the servo movements.
Another advantage is that you can switch between manual and
CNC operation while maintaining the coordinate system, and
using EMC as a DRO when in manual mode.

I have verified that this works, and that EMC can control the servo
motor with very stable loop response. This was a concern in that
it adds several more layers of control loops to the existing servo
loops in the Gecko drive. I believe that I have actually gotten much
better servo response with this system, too. Since the I and D terms
are adjustable, and velocity and acceleration feed-forward are
also available, it is more flexible. I have made a sample plot
of a stop from moderate speed, using an Ametek servo motor and
a 36 V power supply.
http://jelinux.pico-systems.com/images/servo.png

The encoder was powered from my USC board, and I tried a
magnetically-coupled isolator to convey a copy of the encoder
signals to the Gecko drive. That did not work, as the encoder's
rise and fall times were apparently too slow for that isolator.
I will try it with a fast opto-isolator shortly.

But, the big news is that the entire scheme works, and the servo
response is stable, and the tuning is only slightly harder than usual.
(The more knobs, the harder anything is to set up. The Gecko 320
has 3 pots on it that interact with the software-set variables.)

Jon