Fanuc DC motor drives

[Scott Pancheau]

Setting up a lathe with Fanuc DC drives. I can use DAC commands on X ,Z, and spindle. They respond smoothly and correctly for direction and speed. I can use the move commands in tuning for the Z axis and get the expected results. The step command will only jerk a minor amount and back regardless of how I change the settings. The X axis just takes off in one direction or the other and runs until it hits the limit. Can the direction be wrong even if the Dac commands work correctly? What else should I check and what would you want too see to help with this?
Thanks for your help,
Scott

[TomKerekes]

Hi Scott,

The step command will only jerk a minor amount and back regardless of how I change the settings.

Use the Move command not the Step command. The Step command is usually not needed and would only be useful for testing very small steps movements. Servos do not normally respond well to large instantaneous errors.

The X axis just takes off in one direction or the other and runs until it hits the limit. Can the direction be wrong even if the Dac commands work correctly?

Yes the encoder feedback can be backwards. Try reversing the InputGain0 setting.

What else should I check and what would you want too see to help with this?

Follow the tuning process in the wiki. Post the resulting plots and all your settings used to make them (Config/Step Response/Filters Screens).

Its simpler to work on one axis at a time.

[Scott Pancheau]

Thank you Tom. It did work to change the input gain setting. I have spent several hours trying to get the X axis to be more responsive, but it seems to not like very much of anything as far as PID settings. It wants to oscillate and trip out with increases of any of these. I can get smooth moves, but cannot get much for speed. With DAC commands I can get high speeds, but not with tuning. I don't understand what to try with the filter setting to effect a better outcome.

[Scott Pancheau]

I made another run at tuning. I think this plot is better? The command and move are more together. I don't know how to try to get the red - commanded position to be less wide? The analog output was showing just over 9 volts so I think that would be maximum velocity that can be achieved? I also entered the suggested 500 freq on the third item of trajectory planner.
Thanks,
Scott

[TomKerekes]

Hi Scott,

You don't seem to be following the instructions and steps in the wiki.

Here are some comments:

Max limit output should always be 2047 or less for Kanalog DACs otherwise they will wrap around and cause weird behavior.

D gain of 0.7 is not likely to do anything as it is too small. Set it to 0 until you are ready to tune it properly

Plot: Command, Position, Output Command vs Time not Velocities. Note it is helpful to use the Save Data button and save the raw data and post it here. That way we can re-plot the data however we wish.

Test with a smaller move size

Increase the capture time to 3 seconds

Remove Feed forwards at this stage

Remove dead band at this stage Range = 0 Gain = 1

I don't know how to try to get the red - commanded position to be less wide?

Measured velocity tends to be noisy because of limited encoder resolution and being sampled at a high rate. At 100000 counts/sec the encoder changes either 9, 10, or 11 counts per sample. So velocity resolution is poor. This is normal and don't be concerned about it. It does show that the desired Velocity is being achieved

What is the resolution of your system in counts/inch?

I also entered the suggested 500 freq on the third item of trajectory planner.

I assume you meant to say IIR Filter Screen not Trajectory Planner. You might post the screen so we can see what you did.

[Scott Pancheau]

Thanks again Tom. I went back through the wiki and tried to follow it better. Thanks for telling me about the 2047 max limit and it causing wrap around. I was have strange results at times. I did mean filter screen in regard to the 500 frequency entry. I am including a shot of that screen. I am including a shot of trajectory planner with the derived counts per inch (68,277) for X - the axis I am working on. I did not do the save function this time but I will from now forward. I have included shots of two different size moves in the bode function you told me to use. The drive did not seem to want anything but P added, and very little of that. The settings shown are giving what I think is the max smooth action I could attain. At one more P fraction it starts to show osciliation tendancy and adding D does not help, nor does feed forward. I am getting near full voltage output and I think good following. The axis is returning to the exact count position it starts at when returning from a move. Thanks again for you patience.
Scott

[TomKerekes]

Hi Scott,

I did mean filter screen in regard to the 500 frequency entry. I am including a shot of that screen.

Yes, the IIR Filter Screen looks fine except you entered Q=1.401 instead of 1.4. This wouldn't make any significant difference but just seems silly.

I have included shots of two different size moves in the bode function you told me to use.

This is the "Step Response Screen" which has nothing to do with Bode Plot Screen. You probably won't need to use the Bode Plot Screen.

The drive did not seem to want anything but P added, and very little of that. The settings shown are giving what I think is the max smooth action I could attain.

Those look good for only P Gain. But there are still errors of ~800 counts or 0.012 inches.

At one more P fraction it starts to show osciliation tendancy

Not sure what you mean by "fraction"? Maybe you mean 0.1? But good it seems you have maximized P Gain properly at this stage. At what value did it go unstable? You should then back off something like 10% to be sure you are not right on the edge of being unstable.

adding D does not help

What values did you try? What happened? D gain usually needs to be 10~100X the P value to have an effect. At what level of D did the system go unstable and oscillate? It may be that D Gain doesn't help as the Velocity Mode Drives should be providing Velocity feedback/control which is essentially the same as D Gain. But it isn't clear whether you have really established this.

nor does feed forward

Do not add Feed Forward at this stage

I am getting near full voltage output and I think good following.

Yes you are moving at nearly full speed - 10V or 2000 DAC Counts of Output. But following is not that good yet.

The axis is returning to the exact count position it starts at when returning from a move.

Good. Does it also seem to move physically back to the same physical position?

Note you still have 1 count of dead band range. I prefer to leave dead band off (Range=0 Gain=1) at this stage.

Thanks again for you patience.

And yours!

[Scott Pancheau]

Thanks Tom,
I turned off the dead band and added D and I. Corrected the clerical error on filter screen ( Q to 1.4) Tried to raise the P after adding D, but caused oscilation on longer moves. With the shown settings the axis is returning to the same encoder count and a travel indicator shows the actual position is also returning to the same position. Could you explain which plot you are looking at to see the following error and how to read it? If I did the save right, I saved two different size moves for your evaluation. I added the I to try to correct the following error, but since I do not see how to read the result I don't know if it is improved. Is the error you are talking about the traveling error before it reaches the final destination? If not then why does it show returning to the same start point? Sorry to be so slow to follow.
Thanks,
Scott

[Scott Pancheau]

I don't see the attachments I thought I included from the save data function on plot screen. I saved the plots to documents, copied to a USB, and drug from there into the text area of my post. They showed transferring, but apparently did not stick. Is there a specific procedure to use. The computer at the lathe does not have internet service so I have to do all via a USB. Thanks again Tom,
Scott

[TomKerekes]

Hi Scott,

Use the "Attachments" Tab and "Add files" button to attach files.

I turned off the dead band and added D and I. Corrected the clerical error on filter screen ( Q to 1.4) Tried to raise the P after adding D, but caused oscilation on longer moves. With the shown settings the axis is returning to the same encoder count and a travel indicator shows the actual position is also returning to the same position.

good

Could you explain which plot you are looking at to see the following error and how to read it?

Once the following error becomes small its hard to see the difference between the Command and Position (without zooming in). So we normally plot the error instead. If you would have actually attached the data we could do either of these.

Note Kanalog has no way to observe the supply voltage for your drives (like our SnapAmp can) so no sense in plotting Supply Voltage.

I added the I to try to correct the following error, but since I do not see how to read the result I don't know if it is improved.

Because the max limit Integrator is zero you are not allowing it to do anything even though you have a huge I Gain specified. A normal I Gain is something like 0.01. Max Integrator should be around 4000.

Is the error you are talking about the traveling error before it reaches the final destination? If not then why does it show returning to the same start point? Sorry to be so slow to follow.

The idea to to have the Position (red) follow the Command (blue) throughout the move not just at the ends. Plotting Position Error will show the error throughout the move.

Sorry to be so slow to follow.

No worries. For not being familiar with this you are doing well.