VSD-A Servo drive documentation

[Xerxes]

Jerry,

Responses look fairly good. However, you should use much smaller test travel so you can see small errors easier. Separated PID curves may help spotting a problem such as source of oscillation or overshoot, but there is no much more use for them.

You're right that legends are hard to read. This is a feature of Qwt library which is being used in plot window. We need to modify the library to make thicker lines.

Acceleration limit applies only in tuning/SPI mode and velocity limit works in all modes. Accel limit is disabled in step/dir to avoid double limiting because CNC software does limiting too.

[JerryFlyGuy]

Thanks Tero! I might try pushing that servo a bit more on the PID but... at this point I'm pretty happy in regards to how it runs. I'm running 750ipm max speed w/ a 13in/s/s accel in Mach3. 0-750ipm in well under 1 second.. Max following error is ~40-50 counts which equates to 0.0005" or so.. on shorter/slower [ie; normal cut speeds up to 400ipm] runs its down at ~10 counts..

Thanks for all of your excellent and patient help w/ getting me up to speed on tuning servos!!

Jerry

[H.O]

Hi,
(Continuing discussion from this thread)

I've played some more with my setup and made plots of the commanded/achieved torque - they are attached as pictures to this post. It seems like it's outputting 12.5A although I asked for 15A but IIRC there's a bug in the presenation in GDTool, and I am getting 15A - correct?

I wonder where that nasty looking negative spike in plot number two comes from....

I have 14400 counts/rev, the multiplier setup for a total ratio of 14.4 (Multiplier 81, Divider 90, Filter 16). With the following error limit set to 7200 I can get to 2000rpm in around 0.5 seconds without faulting, lowering the limit to 3600 or loading the motor slightly makes the drive fault during acceleration.

I haven't plotted the HV voltage bus but the supply is a sturdy 4kVA 3-phase transformer, 3-phase bridge and two 15000uF caps in parallel, giving me around 130V DC.

Any hints, clues, pointers or ideas on what I could try to increase the performance is much appreciated.

/Henrik.

Oh, almost forgot.... If I read the settings from the drive or a file the step multiplier Filter setting resets to 1 even though it has been set to 16 (or whatever) when saving the settings (to drive and/or file).

[Xerxes]

Henrik,

How were these plots produced? I guess the first is torque step test and second a position mode step response. Am I correct?

When acceleration limit is set correctly, the torque command should not saturate badly in step test. If it saturates (like in the second picture) then it tries to accelerate faster than possible and position error grows as motor lags behind commanded position.

Drive seems to be following commanded torque fine, except in the second picture when output goes negative. You could plot also output voltage amplitude to see what torque controller is doing to get this torque. The error is not so bad, though. Spiky appearance of torque target is typical in position mode.

Current reaches 15A since it is actually 819 clicks per amp in the figure (scope signals are not properly scaled yet in current GDtool version). I think you have found a bug in filter setting loading. That should be fixed in the next version.

[H.O]

Hi Tero,
Yes, first plot is torque step response (15000mA) and second one is position step response (1000 steps).

After reading your response I went out and played some more. I changed the settings in the Trajectory Planner so the effective velocity limit was ~2300rpm and effective acceleration limit was ~0.25s (roughly what you calculated). Now, I thought that the acceleration limit didn't have any effect when running S/D-mode but it now performs much much better.

Running Mach3, I'm reaching 2000rpm's in ~0.5 seconds and have been able to reduce the following error limit from 7200 to 1800 and I can hold a towel around the motor pulley, applying 'reasonable' pressure while doing back-and-forth moves without faulting on following error

However, sometimes it still faults out with the red and green blinking simultaneously indicating possible over current or other fault. I don't know how to simulate those conditions in GDTool in order to find out what the problem might be.... I'm using an unneccesary long encoder cable at the moment, can encoder signal problems cause this fault condition?

I'm attaching a Step-response plot of 2500 steps with the output voltage plotted as well. At first I thought it looked REALLY strange untill I realized the 16bit register must have overflowed....except from that it looks pretty good, I think.

Still a bit to go but feeling a lot better, thank you!

/Henrik.

[Xerxes]

Looks like you're on right track :-)

Acceleration limit is effective only in SPI/tuning mode in latest firmware. However, it is important to set realistic limit during tunig to get proper behavior and possibilize decent PID tuning. In step/dir the limiter deactivates since pulse source usually has a limiter.

The error could be caused by overvoltage or overcurrent. Ovecurrent would be caused by overshooting response of torque controller (some overshooting/instability is visible in your previous plots). Try adjusting torque PI a bit lower (especially I).