Mori MVJR Build Log

Looks like FF1 still needs dialing in. 0.8 is better but now increase
it by maybe .01 increments, then .001 increments to null out the following error

Notice that the phase of the following error reverses if you have too little (0.8 is too little)rather than too much (1.0 is too much) FF1

That is the green trace goes up first if you have too little, but down first if you have too much

PCW wrote:

Looks like FF1 still needs dialing in. 0.8 is better but now increase
it by maybe .01 increments, then .001 increments to null out the following error

Notice that the phase of the following error reverses if you have too little (0.8 is too little)rather than too much (1.0 is too much) FF1

That is the green trace goes up first if you have too little, but down first if you have too much

Thanks, that's a great tuning tip!

One thing I noticed is that as following error is reduced by adjusting FF1 the PID error increases.

So if I understand correctly, the tuning process goes in a cycle where you rough in PID, then get FF1 close then go back and do PID again and repeat.

Are there any similar tips for knowing what to look for about P, I, D, FF0 and FF2?

That I do not understand at all I would think they should be the same!

Most people tune for minimum FERROR and this is what EMC looks at to flag following errors so lacking any better information, I would not look at PID.error and tune for minimum FERROR
(perhaps PID-error is not what the manual says)

Maybe you can contact a EMC PID guru on the mailing list and ask why these are different

Progress report:

I have the PID close enough to test I think. The only thing that is concerning me now is the drive motors make a low humming, almost a growl, nothing I change in PID seems to make it go away completely but very high P makes it worse.

The spindle is almost working. It turns, but slowly (about 5 RPM) and doesn't change when I set the speed higher or lower.

I have the drive connected to the 7i48 on pins 23 (GND5) and 24(AOUT5)

HAL looks like this:

HALMETER tells me the following:

motion.spindle-on (responding as expected)
motion.spindle-speed-out (responding as expected)
motion.spindle-forward (responding as expected)
motion.spindle-speed-cms-rps (responding as expected)
motion.spindle-at-speed always FALSE (even when moving)
hm2_5i23.0.pwmgen.05.enable (FALSE, No change)
hm2_5i23.0.pwmgen.pwm.frequency 24000 (0x00005DC0) doesn't change with speed adjustment
hm2_5i23.0.pwmgen.pdm.frequency 20000 (0x00004E20) doesn't change with speed adjustment

Does anyone know of something else to check or try?

Note you cannot use 7I48 enable outputs this way as they are not independent (7I48 manual page 9)
and with standard firmware, all enables are driven by pwmgen enable 0
If you want an independent spindle enable, I would use a 7I64 output.

As far as the growling is concerned, do you have a reasonable deadzone setup? say the equivalent of 1 to 2 encoder counts

Also do you have any integral term? You can often trade off some P for I to make a quieter system
(but with somewhat poorer dynamics)

Another thing that may help is using the better velocity estimate from the encoder counters for the PID loop (instead of the default DP/DT)

He is only using the 7i48 for the -+10 V output not the enables.

Thanks
Cory

I tried this:

net spindle-drive-enable motion.spindle-on => hm2_5i23.0.pwmgen.05.enable

Now in the manual mode the spindle turns forward about 100 RPM but doesn't respond to changes to the speed.
In reverse, it still goes slowly.

This is progress.

I don't know if I an using the proper output type.

setp hm2_5i23.0.pwmgen.05.output-type 1

Maybe I am missing some output that tells what RPM to run?

first, output type _must_ be 2 for 7I48

I would then verify that you can get 0-10V on the spindle output (unconnected to drive)

Do this by setting the output value by hand by setting hm2_5i23.0.pwmgen.05.value to 0 and then to 6000 and checking the 7I48s analog output voltage for 0 and +10V. I'm not absolutely sure about the 7I48 you may get 0 to -10V in which case you would need to invert the PWM scale number (-6000)