Servo Tuning

Rather than reinvent the wheel, has anyone got a hal configuration/file file that sends a step train to the servo output ?
My Copley servo cards have a few adjustments on them I need to set those before I go anywhere near the CNC PID loops. I can use stepgen and map the pins but there may be something a bit more elaborate out there.

If anyone has used Copley (306AP) drives before, I presume that Current is the best ?

Thanks Wallace

as far as I am aware the Copley 300 Series are analog +10v or pwm input drives dependant on how the jumpers are configured
so your unable to use stepgen inputs .

what interface are you using Mesa ? with analog interface
in that case pncconf should setup fine , use 100% mode ( see Copley manual ) so you have 2 pins per axis analog input and Direction , is the easiest , along with an amplifier enable pin

to set the Copley drives is mostly painless with the analog input set at 10v for full speed , a shorted analog pin to the input ground should give you no movement , if it moves adjust the input balance pot to null the movement
( you may also have to re adjust this once you add your interface , due to imbalance of voltages etc ) most of this is covered in the manual .

I tend to use a test input signal using 2 x 1.5v dc batteries in series , so at least you don't have the axis flying at full speed !
and then use wire on the direction pin to either low or high logic to switch direction ..

their are other ways , just some are easier than others

I have the mesa servo drives board. I want to supply the drives with a 0.5 hz square wave so I can tune the drive. My signal generator does not go that low.

The servo drive seems very responsive. 1.5 volts swings the motor at full speed.

You can use the siggen component to output square/sine/triangle and sawtooth waveforms of programmable frequencies. You can just connect the siggen.square output to wherever you want to inject the signal

man siggen

Note that a square wave feed to a velocity mode drive is going to be very harsh and perhaps damaging to mechanics, a sine wave is probably preferable for tuning

What we are finding is that any voltage over about 0.2v puts the motor into full speed. I have reduced the gain of the servo drive by changing the 47K feedback resistor to 4k7 but it didn't make much difference.

The cards should work without velocity feedback, however, after quite a few hours of experimenting, I'm going to temporarily hook up the real drive that has a velocity output. Trouble is, the encoder is on the ballscrew not the motor so we have quite a few parts to clean and reassemble to do the testing.

My last CNC machine used Gecko drives and that was a synch to setup.

I can't remember but is there a servo setup in Ver 2.4 for the Mesa cards ? It might be worth a try. I do remember that you have to add the mesa driver in during the compile.

Peter - Working on a Sunday!!!

Sounds like you are tuning a torque mode drive or a velocity mode drive without feedback. A torque mode drives behavior will be totally different than a velocity mode drive (with local velocity feedback)

A velocity mode drive without velocity feedback connected will likely be untunable

We finally have it running. We added the tacho feedback into the drive and all runs well. Basically the LinuxCNC/7I48 combination is not quick enough to successfully run the motors without tacho feedback.

Both motors we tried are servo motors and run up to a max of 2000 rpm so the LinuxCNC/7I48 loop seems to be quite slow.

If I have to do this exercise again, I would choose drives with the encoder feedback built directly into them. Saying that, the encoder signals would still have to be inputted to LinuxCNC to cater for manual hand wheel control and updating of the DRO.

Finally I can now get on with the lathe rebuild.

We have many users successfully running torque loops with 7I33/7I48 and LinuxCNC, so there is no speed issue
with the hardware or LinuxCNC. Torque loop tuning _is_ more difficult however and since a torque mode PID loop
is responsible for the high frequency velocity control portion, you may have to use a faster servo thread (up to around 4 KHz)
for a combination of stability with sufficient D term gain.

For good velocity feedback you will also have to use the improved velocity estimate be connecting the encoder velocity output to the
PID loops feedback.deriv pin