Page 1 of 2

Quadrature Error

Posted: Tue Sep 06, 2022 7:13 am
by Talyrond
Hi, I’ve been chasing a quadrature error for weeks now. My setup is:

Oak with DMM DYN4 drives and motor. Servo cables from DMM also. I’m UK based so 240v (single phase)

I’ve been working with DMM on this for some time now and we are struggling to resolve the issue. I’m not sure where to start. I have tried so many things, I’ll try and summarize.

The initial bench testing went OK, with the motors on the bench, the problems all started when I bolted the motors to the machine. At that point I had no earth on the machine, adding that helped, but the Oak was powered separately at that time

Where I am at now is that if I power the Oak from a different socket in the building, rather than from the panel, it will at least run for a while. However it always eventually errors. Interestingly, if I power the Oak from the panel, but move the earth to a different socket it runs better, but eventually errors.
So to recap running the OAK from the panel it will not run at all.

All this points to noise, ground loops issues etc. I have tried everything I can think of. In the attached images right now most components are lifted off the aluminium panel to try and prevent ground loops. Every earth is taken back to a single star point. The machine earth is temporary as seen in the attached (directly to the X saddle at the moment)

Right now I have disconnected all other contactors/power supplies, to help isolate the fault. The E-Stop circuit is also removed. The wiring diagram is stripped down for clarity. The wiring is as per DMM recommendations.

Also I have an isolating transformer that drops to 115v. I have tried that before the Oak PS, the drives logic power and finally on the main supply before the reactor. It helps but still eventually errors.

Report attached, I’m only setup for one axis at the moment.

Appreciate any advice.

Julian

Re: Quadrature Error

Posted: Tue Sep 06, 2022 7:49 am
by ashesman
So, an important thing to understand is that earth from the mains is primarily for electrical safety and noise noise. It can help as a return path for noise introduced by the mains supply itself.

You have done the right thing star pointing the earths. But, remember eliminating noise is about reducing the loop area of circuits. Twist all pairs where you know one wire is the return path for the other. Make sure all cable shields are only connected at one end.

Personally I would start by powering only the oak and one DMM drive from the.same supply. And absolutely nothing else. Then start introducing parts.

Did you have filters on the control mains into the DYN4? Is that the silver things?

If things get desperate you may need ferrites on some motor cables.

Have you done the tests for encoder feedback working? One test would be to disable the motor and move it manually heaps by pushing the table back and forward, then eventually back to the home spot. Encoder counts should be identical with no errors anywhere.

What else do you have hanging off the 24v supply under the oak. The 24v supply powers the logic signals from the oak and encoder inputs. Noise on that supply would be less than ideal.

I would personally take advantage of your metal backplate and use it. Earth everything that is connected to earth onto the back plane.

Most importantly, as you have been doing, keep removing parts until it's the simplest possible system.

Re: Quadrature Error

Posted: Wed Sep 07, 2022 7:46 am
by Talyrond
@ashesman, thanks for the info.

>Personally I would start by powering only the oak and one DMM drive from the same supply. And >absolutely nothing else. Then start introducing parts.
Yes that's what I did. I took everything off the panel and hard wired a minimum setup. Drives and Oak on separate sockets in the wall, but next to each other. Seemed OK. As soon as I introduce the panel and motors bolted to the machine the problems start. I’ve tried so many things, but the closer the Oak power feed gets to the drives power feed the worse it gets.

>Make sure all cable shields are only connected at one end.
The drive cable supplied by DMM are shielded correctly, I can’t see that any other cables would require that? Encoder cables are supplied by DMM, I know nothing about them.

> Twist all pairs where you know one wire is the return path for the other
Should I twist AC pairs?

>If things get desperate you may need ferrites on some motor cables.
Yes tried that.

>Did you have filters on the control mains into the DYN4? Is that the silver things?
Yes those are the EMI filters, one large one on the main power input and then a smaller one for each drive logic power feed

>Have you done the tests for encoder feedback working?
Just tried that and I can’t even turn the ballscrew without it giving the quadrature error. If I take the OAK PS earth off then I can test and the position is spot on.

>What else do you have hanging off the 24v supply under the oak. The 24v supply powers the logic >signals from the oak and encoder inputs. Noise on that supply would be less than ideal.
I have noting else on the Oak PS.

I’ve made a quick video as well: https://e1.pcloud.link/publink/show?cod ... xMFYWSMNaX

The system seems so sensitive, should it really be that susceptible to noise? I’m running out of ideas. I will do more research on ground loops. Appreciate your help!

Julian

Re: Quadrature Error

Posted: Wed Sep 07, 2022 1:41 pm
by cncsnw
You should find an oscilloscope, and if possible someone who is experienced using it.

Then you could find out what electromagnetic noise can be seen on your system before you enable the DMM drive, and after you enable the DMM drive.

Alternately, if better servos are not in the budget, then you could throw up your hands; accept that a few encoder counts' worth of position loss is tolerable; and set Parameter 334 to a value of 7 (disable quadrature faults on encoder inputs #1, #2 and #3). If you don't want to see the messages on the screen either, you could set Parameter 335 to a value of 7 as well. See Chapter 15 in the operator's manual.

Re: Quadrature Error

Posted: Wed Sep 07, 2022 3:31 pm
by ashesman
>Have you done the tests for encoder feedback working?
Just tried that and I can’t even turn the ballscrew without it giving the quadrature error. If I take the OAK PS earth off then I can test and the position is spot on.

I think the answer lies here. If with the motor drives disabled you have encoder issues, then that is a good place to start as you have already shown that the issue is not generated by the motor drives.

The fact that removing that earth seems to solve the issue is a good start. .you have done well to find that. Is that earth connected to the negative output of the power supply? Or is that the earth for mains input to the power supply?

IMO, using the same 24v supply for the oak and external wiring is bad practice, even though centroid recommend it. You are running long wires out around noisy stuff and conducting that noise onto the oaks power supply. I use a separate power supply for the IO. At least if something shorts on that supply, it wont take the oak down.

So, to take your diagnostics a bit further... You need to get to the point where you can move the table without motors powered and not get errors.

The oak power supply needs to be earthed on its input for safety and to meet emc.

Have you tried disconnecting everything but the oak from that power supply? Have you tried a different power supply?

Are you confident your mains neutral is good and you dont have return current through earth?

Have you tried swapping out the oak to DMM cables? If you have one of the encoder feedback signals only connected by one wire rather than the two I would expect the sort of thing you are seeing. Remove the covers off the DMM end of the cable and ohm meter check all the a,b,z signals.

Disable all but one axis so that the problem only needs to be solved for one encoder.

I would be interested to see what the encoder counts number is doing on the PID screen with and without earth connected.

A scope would be quite handy if you have access to one. But needs to be carefully used.

Re: Quadrature Error

Posted: Thu Sep 08, 2022 11:20 am
by Talyrond
The fact that removing that earth seems to solve the issue is a good start. .you have done well to find that. Is that earth connected to the negative output of the power supply? Or is that the earth for mains input to the power supply?
Yes the earth is linked to the 0v common, that is in fact done by Centroid. The PS comes supplied with the Oak. I did remove the link but no different.

IMO, using the same 24v supply for the oak and external wiring is bad practice, even though centroid recommend it. You are running long wires out around noisy stuff and conducting that noise onto the oaks power supply. I use a separate power supply for the IO. At least if something shorts on that supply, it wont take the oak down.
Have you tried disconnecting everything but the oak from that power supply? Have you tried a different power supply?
The Oak PS only supplies the Oak.

Are you confident your mains neutral is good and you dont have return current through earth?
No at all! Needs further investigation.

Have you tried swapping out the oak to DMM cables? If you have one of the encoder feedback signals only connected by one wire rather than the two I would expect the sort of thing you are seeing. Remove the covers off the DMM end of the cable and ohm meter check all the a,b,z signals.
Yes tried all 3 cable sets I have.

Disable all but one axis so that the problem only needs to be solved for one encoder.
Yes I’m only working on one axis right now.

I would be interested to see what the encoder counts number is doing on the PID screen with and without earth connected.
I make a video, may be relevant: https://e1.pcloud.link/publink/show?cod ... Q3oJ0kKfdy

A scope would be quite handy if you have access to one. But needs to be carefully used.
I may resort to this, as per @cncsnw, I do know someone who may be able to help. I think this could be be useful.

Thanks again

Julian

Re: Quadrature Error

Posted: Thu Sep 08, 2022 11:28 am
by Talyrond
@cncsnw. I do know someone with a scope who may be able to help. I will contact them. Failing that, yes the budget will stretch to better servos. I take it you think they are the cause for sure of my problem?

I'd rather avoid the expense, but I would upgrade if I was confident it would fix the problem. (assuming I fail to find the problem)

Thanks

Julian

Re: Quadrature Error

Posted: Thu Sep 08, 2022 3:41 pm
by ashesman
The * in the PID screen indicates the Z feedback signal to the Oak is active. It should only be active once per revolution for a tiny distance. It is the index pulse from the DMM drive.

Most likely if that signal is noisy, the a and b signals will be too.

So just to clarify, its earthling the 0v output of the oak power supply that causes the issue?

I dont know that it really needs to be earthed. It is a fully isolated system. My machine does not have e that point earthed. From memory.i did not earth that point on my machine.

All the signals to DMM drives are isolated and differential so dont require an earth return path.

I dont think the DMM drives are the issue. While they are not without other issues, I doubt they are causing this problem.

If it is working well without that earth wire attached, dont earth that point. It is not an electrical safety risk. That said, that point should be able to be earthed without creating issues so I still dont quite understand what is going on there.

Another test would be to use a separate 24v supply for the oak. They are cheap enough. You can use the 24v from the oak supply for other stuff. I use a separate one for the 24v to oak just so I know that signal system is al separate and standalone.

If you dont have a spare supply, your VFD probably has a 24v output that would be more than capable. Its actually not a bad option to use it as it also means the VFD analog signals form part of the closed system. I have tried this on three VFDs and it always works.

I honestly think the answer here is in the wiring and you will get there.

Re: Quadrature Error

Posted: Wed Sep 14, 2022 9:58 am
by Talyrond
>So just to clarify, its earthling the 0v output of the oak power supply that causes the issue?
Yes it's the earth that feeds the Oak PS, that in turn is linked to the common 0v. However have had it error out even when the earth is disconnected now, so it improves things but not completely.

>Another test would be to use a separate 24v supply for the oak.
Yes just tried that, no luck

My next move before getting a scope involved, is to test the panel at home.

Keep you posted.

Thanks!

Julian

Re: Quadrature Error

Posted: Wed Sep 14, 2022 3:17 pm
by ashesman
There is the possibility that there is an issue with the hardware on the Oak. Maybe a dicky power supply or something not happy in the drive interface part of the board. I am not sure how common or likely this is but dont rule it out.

Another option would be do do an absolute bare bones bench test setup without your panel. Just write oak, power supply and one drive. Without anything else. No filters, no stop relay etc. Just a single mains supply to the drive and oak psu (with a fuse). Nothing else wired to the oak. Jumper estop input on the oak or disable it if need be.

I know you are close to this setup now, but there might still be a step further you can go. The filters are not necessary if the mains is reasonably clean and you are not drawing significant current.