I have built a nice DIY CNC router with much help from ebay.
However, I have run into trouble with the servo tuning stage, to the
point at which I have just ripped a metal flexicoupling in half
While tuning the X axis, I have seen the commanded move of approx 50mm
take place ok, then the servo threw itself into reverse at max speed
for a seemingly random distance. This ultimately saw it charge through
the limit switches and rip the coupling in half.
The Y axis seems unable to move at all, each time flagging 'Position
The only axis that is behaving is the Z axis.
Sorry if this is a bit long, but any suggestions would be gratefully
email me at novacustard@here
where h e r e can be replaced by yahooDOTcoD O Tuk
Servos, drives, controller and software are all Parker/Compumotor,
Controller is a Parker 6K4, all connecttions are by ribbon cable or
power cable where necesary. (Not shielded).
X axis is an SE230AE with 10:1 gearhead and OEM770T drive with 75vdc
p/s and THK44 Linear Actuator with 1 rev to 20mm leadscrew and 500mm
Y axis is an SE230AE with 10:1 gearhead and OEM770T drive with 75vdc
p/s and THK33 Linear Actuator with 1 rev to 10mm leadscrew and 200mm
Z axis is an SE232BE direct coupled to unknown Linear Actuator with 10
rev to 25.4mm leadscrew and 200mm travel. OEM770T drive with 24vdc supply.
I dont't know anything about the Parker stuff, but it seems like you have an error in the limit switch or home switch circuits.
I had a similar (non-distructive) occurance with a Gecko control system and EMC. Stuff wouldn't happen until I got the wiring to the switches right.
Actually, I think those part numbers are brushless, but it may not matter. If they are brushed, you probably have the power windings backwards. If they are brushless, you may have the encoder backwards, don't think they will run if you have the windings swapped.
The OEM770T is an analog control version of the amplifier. Do you actually have the OEM770SD? If you are supplying an analog control voltage, it's possible you have that control voltage backwards, that will cause a runaway. What software are you using to control this system? On Edit: now I see that you have a compumotor control and software.
It's always possible that the encoders are reading noise. You might want to consider new wiring.
You need to uncouple the x and y motors from the leadscrews until you get this straightened out.
Did you do your own wiring, or did it come complete with the system? Was the system working before you bought it?
I have had the X & Y axis running, at which point the X axis was powered by the 24vdc supply and the Y axis had a SE232BE motor. The Z axis was not built so I rigged up a simple gantry with a pen and produced a proof of principle drawing.
At this point I was feeling pretty pleased - all the motor connections are wired the same, so are interchangeable. I then moved the router from one room to another and decided to chnage the Y axis to the same motor/gearbox as the X axis and upgrading the X & Y axes to 75vdc. This meant changing some of the resistors inside the drive, but should be a no brainer.
One thing I did do is swap the X axis from one drive to another (identical) drive.
I guess what I have is a wiring issue, although i have carefully checked the wiring and it looks ok - I'll have to get a multimeter onto it.
Another possibility is that I have a faulty connection somewhere.
It certainly sounds like there is an encoder issue.
All wiring has been done by me, and I have not used shielded cables, but even separating the power and signal cables by several feet made no difference.
The drives are definately OEM770T's.
Best disconnect the Y drive (seeing as the X drive disconnected itself!).
Do you have confidence that it was running before you moved it? If so, it makes me wonder if something was damaged. However, encoders cause a lot of people problems, particularly if they are not differential. I was looking at the tuning procedures. It seems possible that you could drive the system unstable by changing the adjustments. You probably want to follow the procedure to return the adjustments to their default settings and start from there.
Yes, I'm pretty confident that a. the X & Y axes were running ok before I moved and rejigged the control circuits. I will just have to go back to basics and carefully check the encoder wiring (and others). I have disconnected and removed the motors for this next stage as I don't want any more mishaps. Certainly the power and speed from these relatevely small motors coupled to 10:1 gearboxes is most impressive, just need to get them under control.
Thanks for the replies.
Please continue to add any suggestions that you feel may be helpful.
Using Motion Planner's Servo Tuner, with a trapezoidal move profile, and an Integral value of between 7 and 10, I had been getting smooth, but possibly overdamped moves. (I have tried to attach a jpg showing on the left the ideal command and response, and on the right the cammand and response I am getting).
One strange thing I noticed - you can set the move profile, distance moved, velocity and accelleration/decelleration. Whereas before these had made a difference, now, the X axis would move at what ever speed it wanted - changing the velocity made no difference to the speed. However, if I changed only the Integral value from the default 0.5 to 5 or 10, then the speed of the move increased, even though I had not changed the commanded velocity value.
I am more and more convinced that the problem lies in the encoders or their wiring.
Andy, I've tried to email you a few times but the messages keep bouncing back.
If you could ask your parker expert some questions I'd be really grateful.
Basically, the instructions for the tuning process are a bit vague - can your guy suggest some data values for the Proportional, Derivative/Velocity, Integral, Integral Limit, Accel Feedforward, Vel Feedforward boxes on the Servo Tuner? I guess that I don't need to complete all the boxes...
My previous post showed the ideal command/response (ont the left) and the command/response I'm getting - which seems so different from the ideal that I wonder if its a problem, or what I could do to get closer to the ideal?
Any help gratefully received.
All the best,
PS, I'll take a few photos and post them later this week.
Why doesn't your command ever flatten out? Basically, it looks like your machine was doing what it was supposed to, including smashing into the end blocks -- the poisition is following the command. I think there is a problem with the way the tuning software is set up.
From what you've said, it appears that there are two sets of gains, one on the OEM770's and one in the controller. So you were changing the gains in the controller.
Integral gains will tend to destablize a system. Integral gain is there to eliminate steady state error, you command a position and it ends up there eventually if there is some integral gain. You should be able to set this very low. The integral limit is there to keep the integrator from driving the position away from where it should be just to get rid of the integrated gain. Should be set low. Derivitive gain is wonderful for stability and performance in theory, but in practice it adds noise. Also should be set low at first. The feedforward gains are for tracking accuracy, and at first you should be able to set them very low. So at first, the proportional gain should be much larger than the other gains.
I think what i actually did was use a value of 7 to 10 in the Proportional box, not the Integral, but thanks for the tips - this is just what I need.
Yes, I wonder why the command never flattens out - when you compare it to what it should look like it just seems so wrong.
As for there being two sets of gains, one for the drive and one for the controller, I don't think that is the case. The Servo Tuner software operates through the controller, but the gains are set for each drive on each axis, and once set up are applied everytime the Setup program is run (ie on startup of the controller).