[vebers]

Has anyone had trouble with the rutex 2020 drive system. I can take one motor on the bench and supply the motor and controller with 24v and make it run well. However when I take that same set up to 100v on the motor all hell breaks loose. For instance with the 100v off the encoder signal increments perfectly in the tuning software window, but if I flip the switch the encoder starts jumping around and I begin to get SPI reset faults. I attributed this to noise so I set about buying sheilded cable for the encoder and motor cables. I also made sure to route things away from each other. I have contacted Tom and he seems baffled, I have tried other computers, and I even bought a US digital encoder with no better results. I would appreciate any help that could be offered I am getting slightly upset about the whole thing.

Vince

[dpaulson]

Are you using the sheilded cables now? Also recheck your encoder cable that you have it wired and soldered to the correct pins on the db9 end.

[vebers]

I am still using sheilded cable on both the motor and encoder wires. I have rechecked the encoder wiring and it is correct. I even had someone else look at it just in case I was overlooking something. I have actually had two axis up and running however you never know when they are just going to take off on you. I am presuming that it is doing this in an effort to catch the bouncing encouder readings. At one point I had put a magic marker in the spindle and had the X & Y axis drawing circles on a piece of aluminum, and after drawing a couple it started a third and just took off stopped moved back and completed the circle. I am doing this in Mach using step and direction

Using the tuning software in SPI mode when I run test profiles I don't get clean graphs sometimes the motor runs smooth and I get nice graphs however if I run the same profile again my results do not repeat. If I do manage to get a nice graph with no erronous lines hanging off of it I will not get it twice.

If anyone has any ideas I am all ears. I have put better than two months into this problem and I am looking forward to moving on to the next phase of the project.

Vince

[dpaulson]

'At one point I had put a magic marker in the spindle and had the X & Y axis drawing circles on a piece of aluminum, and after drawing a couple it started a third and just took off stopped moved back and completed the circle'

This statement tells me that the encoders are indeed wired correctly, I would probably put the blame on a poor connection that you have made on the wires attached to the encoder, probably on the encoder side. Likely it is on one axis only and on the axis that ran away. I've had that happen to me. When the wires were moved at a certain position, I lost connection to one of the encoder signals, therefore the run away. As far as the rest, I think you need more work on the PID tuning. I have only used the r510 boards and PID tuning is more trial and error.

[vebers]

The connections are painstakenly soldered onto amphinol type connectors on the motor end and are factory on the RJ45 end. I don't have a scope. I wish I did, but I have run a resistance check between all the pins and the factory end of the cable and all are low and similiar on all three axis. I should point out that I have the motor take off unexplained on all three axis at some point or another. Most of the guys I have been in contact with have used the R series boards and are very satisfied.

I can disconnect the motor power from the board and watch the encoder increment up and down smooth as glass in the tuning software by turning the motor by hand, but as soon as I connect motor power to the boards the encoder readings become unstable and I begin to get communication errors. The motor power does not have to be actually connected to the motors, just the boards. I am begining to think that the R series boards are a stable system but the 2000 series have some bugs. Incedintly I am running 100 VDC and I am running through a 1 ohm 100 watt resistor to the motor. If I take out the resistor it gets a little better however I don't meet the Ohm requirement under that scenerio. Also I have found the the step multiplier isn't working as it should. I purchased these boards because I could use the 100 vdc power supply I already had and I could use the step multiplier to keep the machines original speed with mach pulsing at 35k. I planned on using a step multiplier of 2 which should have given me a resolution of about .0001. However In actuality I am having to use 4 because the something in the systems not right. In other words mechanical, pulse and multiplier math indicates a setting of two, but I am having to use 4. I know this is a bug. I believe the problems are noise but am open to other ideas. I believe the encoders are fine due to the fact that I have no problems with the readings until I hook up the 100 v power to the boards I am at a loss as to how to solve it. Tom is out of ideas, and I run a business and haven't had time to hunt dowm Vlad

[dpaulson]

Just a FYI, if you terminate the shield on both sides, you'll still let noise thru. Only terminate the shield on one side.

[miljnor]

I am in the set up and tune phase with the same boards as you and had a similar problem.

I ordered and got the little differential boards to hook to my servos (I was using TTL encoders). The problem has disapeared. I would look to the encoder because your problem sound exactly like the one I was having.

[vebers]

I have tried connecting the sheild drain on one end and on both ends together it doesn't seem to make a diffrence either way. I currently have it connected on one end.

I am using a diffrential encoders. I have tried the ones that came on the motors originally and I bought a US digital encoder to try. I am having the same problems regardless of the encoders I use.

Thanks for the responces guys.

Vince

[dpaulson]

Can it be possible that your 100vdc power source is at fault? Can you change the rectifier. Maybe you can check to see if some AC current is getting thru on the DC side. I'm just thinking out loud here, I don't know if it would cause any problem.

[miljnor]

how many leads do the encoders have? I though mine were differential but I was confusing Quadrature with differential. Differential endcoders will have 7-8 outputs depending on indexing. TTL only have 4-5 outputs.

that definitely sounds like a noise problem though. It is the same symptoms I had.

Have you tried spiral wraping the wire from the power supply to the servo drives?

I only have very short wires from my caps to my servo boards but this was the first thing I did on my box and the noise problem got better but only the encoders solved the problem. After I started having problems I spiraled every pair of wires I had in the box. Unless they were loners!

edit: I also switched all the motor wiring to sheilded wiring.

[vebers]

I am using the original power supply out of this mill. I don't know about changing the recifier. How would I check to see if AC power is getting though on the DC side?

The encoder outputs
A
A_
B
B_
Z
Z-
+5
0

I have not tried spiral wrapping and will do so. I probably can shorten my power wires some as well.

Thanks for the ideas please keep them coming

Vince

[triberman]

Vebers,
I'm no expert but I did get errors alot until I got my servos under control in spi. I ocassionally will still get errors in spi, but I ran my mill all week on the same job in mach3 and repeating position was not a problem. I too got good graphs only some of the time, especially with the visual basic program. The later at night it was, I found myself forgetting to jump the pins when switching from mach to spi..

I wasn't prepared for the violent ossilating that can happen when tuning. Does your setup still take off with all the Kx settings at zero except a little KP?

I think I ended up with kp 1100, ki 75, kd 2250, ki index 2 for x and y, but z for some reason preformed better with all the k values higher. I know every situation will be drastically different.

[ViperTX]

Oscilloscope......now for a question.....are these 100 volt servos...like in DC Servos.....??

[HillBilly]

One thing to check will be your armature to groung resistance. Servo motors that have been in use for a while can built up brush dust and conduct (see attached pic.).

Darek

[vebers]

Thanks for all of your help and advice. I think (very hopefull) that we have the problem solved. What I ended up having to do was leave the sheild on both ends of the motor cable all the way from the motor to the board. I stripped less than a half an inch back to get a couple of ring terminal on it. This seems to have made a world of diffrence in this machines behavior. I am still occasionally getting some funky stuff when I tune, but the encoder readings have quit jumping and I am no longer getting communication errors. The servo's are 100 VDC. I am going to try hooking the other two axis up and hopefully my luck holds. I will let everyong know. I also noticed that it works best with the sheild drain not connected on either end and the motor case ground not connected. I hope this thread can save someone some time putting together a system like mine in the future. Thanks for the help.

Vince

[HillBilly]

I also noticed that it works best with the sheild drain not connected on either end and the motor case ground not connected.

Vince

That sounds like a sure sign of armature current leaking to ground and returning through the sheild. If this is the problem a regular ohm meter should see it, a megar would be better.

Darek

[miljnor]

Not only what hillbilly said, but you might as well not use sheilded wire at all if your not grounding one end.

Without a return path for the voltage created in the shield it doesn't really do anything!

[dpaulson]

You are using the sheild as a common ground. I would suggest running a single wire as a ground and disconnecting one side of the shield drain and using it to sheild noise.

[vebers]

I would Like to thank all of you for your contributions, your advice and encouragment kept me from throwing the whole thing against the wall. To update you the machine is making money. A combination of many of the ideas you all provided helped get it that way.

Thanks

Vince Ebers
MVM Precision Machine

[miljnor]

good job man and glad to hear it!

:cheers: