[NICKKINSMAN]

I am retrofitting an old industrial lathe Cadillac NC100 with new controls. It has Fanuc model 0 servo motors 4 brush, rated 12 amps 'Continuous Rated RMS Current at Stall TENV' per the datasheet. It has quadrature encoders with 2000 counts/rev = 8,000 counts/rev total (with index pulse, but not used). The system architecture is Mach3 --> smoothstepper ESS --> PMDX 126 --> Viper 200

I wanted to get opinion on what to try next on tuning. I’ve got some pictures of settings and error jog traces here to look at. I am a little afraid to push the numbers, as I’m not sure if where I am at is good enough or not.

I first set L, I kept it pretty high – it seems to have less frequency on the lower L settings on the chopping of the error, but I was afraid to go lower than 6 (I tried 7 & 8 too).

Then I tried bump method, pushing up P&D in 50&100 increments respectfully, (to keep D twice P). I bumped the pulley, but really couldn’t see much difference in performance. So I backed back down and tried doing jog movements and watching the trace. While increasing P&D, the trace error reduced significantly, it was in the 240 counts range and went down to the 100 count range (I was jogging in the 25ipm range for setting the P&D).

The thing I am especially not sure of is I have very choppy error on the move plateau during any jog movement, and it gets much choppier as speed is increased. I tried increasing the FF slowly and finally to match D, but it didn’t seem to have much effect. I took some pictures of the settings & traces at 5, 20, 50 & 75 ipm jog rates. This is only the Z axis, don’t have the X hooked up yet, with short jogs using Mach3 & smoothstepper. I am running in bank 0 as well for this initial testing.

As you would expect, error amplitude is much lower at lower ipm than at higher ipm.
Picture 37 is 5 ipm
38 is 20 ipm
39 is 50 ipm
40 is 75 ipm

Finally, I found that the error also has large spikes (200+cts) right at 10or11 ipm for some reason. I didn’t take a picture of it yet, but it doesn’t look that bad below and even above 10, but for some reason right at 10 it gets bad. Is that odd or what?

Thanks
Nick

[NICKKINSMAN]

Next I'm going to try another tact, backing down L start from low P & D slowly increasing (but no lower than 4 on L as it get's very unstable). It seems like lower L was starting to smooth out the spiking under feed moves. The settling looks pretty good at the end of moves, it is really only the spikes in the middle that are the problem.

[NICKKINSMAN]

OK, I ran thru some scenarios and here is what I found:

I ended up staying with Loop of 6, anything else seemed to get unstable. P (and 2x relative D) over 450 gets bad, it faults the drive right away on any move.

On Larry's suggestion, I also increased hysteresis to improve the starting to overcome friction, and that worked really well, the starting spike lowered down right away, even with a low number (2 in this case). Anything over 2 seems to cause the error to start jumping, so I stettled on 2.

Also, I ended up increasing integral gain as well, when the P & D got higher, it tended to start not settling out all the way and jumping around. With I at 5, it seems to settle, sometimes it will jump around 0/-1 on the error counts but does stop eventually.

I also increased the feed forward - that helped move the error scale much closer to 0 and worked well. I stopped adding when the peaks/valleys are right around 0 error. It didn't do anything with the spikes though.

Also, I get very high error spikes right at 10 ipm. Above and below that look pretty good. It's not just a little, the amplitutde of the spikes is closer to what I see at 100 ipm!! Very odd.

Finally, for some reason it does not like the exerciser, as soon as I try it the drive faults. Not sure why this would be.

I tried/plotted feed moves all the way up to max planned speed of 100 ipm. Attached are the error plots of the moves. Considering the speed, the error doesn't look to bad as far as total, however I am still getting the spiking amplitude and nothing I do seems to change that.

Please review and see if there is another suggested path forward, or if I should retune with another base starting point.

(Pictures are titled with IPM, hover over to see title or refer to title block.)

[NICKKINSMAN]

Here are the Fanuc model 0 servo specs which has the datasheet for the Model 0 showing the loads these servo's take. These have the outline drawings and the general specs for Fanuc 0 and Fanuc 5 sized motors, as well as general info for Fanuc 00, 10, 10H, 20, 20H, 30, 30H, 40, 50, 60, 60H, 0L, 5L, 7L, 10L

[Larken]

Nick, it seems your motors are drawing a lot of current , even for 90 volt high current motors. My short circuit trip comes on at 45 amps

For a test, powere the motor directly with a 12 volt battery and an Ammeter and see what it draws just movine the axis. It should only be a few amps moving an unloaded machine. You may have something causeing a lot of load.

Larry K

[NICKKINSMAN]

OK, I will give that a try - I don't have an ammeter readily available, but I will try to find one. The board seems to be driving the motors well, the only issue I see is the oscillating error readings. do you think varying amperage is what is causing the spikes?

I've only run 1 servo motor on the machine so far, I had the X axis apart for some work, I'll also try tuning the X servo and see what the results are.

[Larken]

The error display will always show a spike when a motor suddenly changes direction. For a split second the will be an error until the PID corrects it. This is normal, but how much error there is is reduced by tuning.

You should be looking at the step and encoder wave and see how closely they follow each other.

[NICKKINSMAN]

Yes, I get the spike when motor is changing direction, but why would there be recurring & regular error spikes during the actual move? I would have thought once the move gets to full speed, the error would zero out as it would be running constant velocity? If you look at my picture posts, I have large spikes in erorr during a constant 1-direction feed move. The plots I show are all for 1 constat movement cycle in 1 direction start to stop, no back and forth at all.

How can I look at the step and encoder wave on the tuning, or are you meaning only do it with the oscillator? I've only been looking at the error counts, as with any amount of feed move using the 'step and encoder' wave readouts, the counts immediately jump way off the chart, as the scale is only settable to several hundred counts on the oscilliscope?

Thanks.
Nick

[Larken]

Yes, i agree it shouldn't have those spikes while moving steady, thats why you need to put a Ammeter on it to see if current is fluctuating, indicating too much friction,or a sticky ball screw , etc. Preferably an analog ammeter so you can see the needle move. run at a slow speed (low voltage 6-12 volts) and check all 3 axis and compare too.

Btw, your not running with a series resistor in the motor circuit, eh ?

Larry

[NICKKINSMAN]

I did get the X-Axis hooked up and working - seeing the same results in the graphs with spikey errors on moves. It's not quite as bad as the Z axis. It does seem to run pretty well other than that. The tuning values seem to work at much lower values than the Z axis. I tried some jogging around with both axis and it seems to run pretty well. Haven't tried to run a program yet.

I still haven't had a chance to work on it with an ammeter & battery, I'm in process of moving to a new location for my job, so it will take me a couple of weeks before I can get back to it, as the machine is being moved.

And no, I do not have any resistors in series with the motors. They are directly connected.

PS - when I hook up the lower voltage 12V battery, will I have to retune everything just to get it to work, or should I try plotting moves right away with the current settings?

Thanks for looking at it.
Nick

[Larken]

PS - when I hook up the lower voltage 12V battery, will I have to retune everything just to get it to work, or should I try plotting moves right away with the current settings?

You don't put the viper in the circuit... Just the battery and ammeter !

Larry

[NICKKINSMAN]

OK, I understand better - I'll hook them up directly and give it a try.

PS - what is the spec on the pin jumpers, as I prepped for this move I managed to loose the jumpers I need for the X drive settings! Had to borrow some off the Z drive! :tired:

Thanks
Nick