[mmurray70]

Hey guys Ive been using a dmm motor/servo on a 4th axis for few years now and it has been working great. I added a second machine and wanted to be able to use the same 4th axis so bought a second dmm drive for this machine, but im having issues with the encoder output. The encoder signal randomly disappears for a few tenths of a second and then comes back at mid-high speed. See attached pics. Note horizonal sections which indicate lost movement.

I tried swapping drives with original machine and new machine works great with the original drive. So i thought i must just have a bad drive, but i ordered a replacement and installed it, and its doing the exact same thing. Im thinking it must be some kind of issue with the encoder output on these newer dmm drives? Is there anything else i can be overlooking to cause this or is it just a problem with the drive? Thanks.

[TomKerekes]

All I can think of is possibly a bad cable where one of the signals is lost momentarily. Does this cable move/flex? But then I would expect the original Motor to also show the problem.

Also check at Kanalog if all 4 of the quadrature signals (A+, A-, B+, B-) are toggling to proper high and low levels. A missing signal can still work but intermittently.

You might zoom way in on the horizontal sections.to see if it is really not changing or doing something strange like +1, -1, +1 ...

[mmurray70]

Connections are all good. It runs fine at lower speeds, only a problem at higher speeds. Everything seems to point to the drives considering the following:

Test 1 - New machine (Fadal #2), new drive (DMM #2), old 4th axis and DMM servo. All sorts of errors and faults at high speed. Ran pretty good at about 40% speed and under, with following error increased from 1000 to 4000. We ran like this for several weeks at reduced speed while i tried various things to improve it without any luck. Still faulted maybe once or twice a day. Sometimes position would be off on the 4th axis after a fault or following error.

Test 2 - New machine (Fadal #2), Old drive (DMM #1), old 4th axis and DMM servo. I pulled the dmm drive from my old Fadal #1 that has been running for years just to eliminate any possibility of bad wiring or setup on the new machine. Put following error back to 1000 and max speed back up and everything ran perfect. Ran them machine like this for a full week, made about 500 parts, tons of movements in all axes and everything was absolutely perfect. I assumed at this point that DMM drive #2 was just defective and ordered a new one.

Test 3 - New machine (Fadal #2), Brand new drive (DMM #3), old 4th axis and DMM servo. Installed a brand new DMM drive and i get the exact same results as Test 1. I did some more testing at this point in Kmotions tuning software and this is when i realized that position is being lost temporarily causing all the problems.

Seems really strange. All my tests point to the drives, but i feel like the chances of getting 2 bad drives are really low? I will check on zooming way in, i guess this would suggest its only an issue with one channel right?. What are the proper levels you expect to see at Kanalog?

[TomKerekes]

I think your logic is correct.

Maybe a bug in the Drive Firmware? Is there a way to compare Firmware Versions?

What are the proper levels you expect to see at Kanalog?

A Line Driver should guarantee a voltage more than 2V (normally significantly more) for a high and less than 0.8V (normally significantly less) for a low voltage. If one signal is floating it will follow the complementary signal rather than being its complement because of the terminating resistor between the two signals.

[mmurray70]

You might zoom way in on the horizontal sections.to see if it is really not changing or doing something strange like +1, -1, +1 ...

Hi Tom,

I did the same test again today. I zoomed in until the scale showed single digits and it is not changing +1/-1, its completely straight. Zooming in on other parts of the curve does show individual counts but the flat section is completely flat. So if there was a bad connection on one channel i should still see the pulses from the other channel on the flat line right?

So considering both channels read fine most time, and then suddenly both are lost at exact same time... this pretty much 100% confirms its an issue with the drives right? Any other possibility you can think of? Thanks again for the help.

Mark

[TomKerekes]

Hi Mark,

I did the same test again today. I zoomed in until the scale showed single digits and it is not changing +1/-1, its completely straight. Zooming in on other parts of the curve does show individual counts but the flat section is completely flat. So if there was a bad connection on one channel i should still see the pulses from the other channel on the flat line right?

Yes

So considering both channels read fine most time, and then suddenly both are lost at exact same time... this pretty much 100% confirms its an issue with the drives right? Any other possibility you can think of?

Pretty much. I suppose anything is possible, but that would be most likely. I suppose a rouge Program in KFLOP could do such a thing. If you could catch the sudden loss of pulses on an oscilloscope that would be most definitive.

[mmurray70]

If you could catch the sudden loss of pulses on an oscilloscope that would be most definitive.

Hi Tom,

I just picked up an oscilloscope few days ago, haven't used it (or any other oscilloscope) at all yet. I got a Rigol DS1102Z-E, hopefully its suitable for this.

I've read that there is a problem with the fact that the probe ground is directly connected to earth ground, and this is an issue with differential encoders? Is there a way to read a differential encoder signal without purchasing differential probes? I think i read a post you made in the past about removing the scopes connection to ground? Would this be as easy as an extension cord with ground pin cut off? Or is that wishful thinking lol? Can you provide more detail on how to do this without damaging anything? Thanks.

[TomKerekes]

Hi Mark,

That's a great scope for the money. I'm not aware of any problem with using a scope to observe differential signals. Unless the scope GND is connected to one of the signals. That could short the signal to Earth GND and cause damage if the scope is earth Grounded. Even if the scope is floating (Earth GND Plug cut off) it wouldn't be a good idea because the GND connection to the scope chassis would add a large capacitance and disturb the signal. So don't do that. Instead connect the scope GND to KFLOP GND. Still probably a good idea to have the scope floating to not cause new GND loops. Then probe the A+ A- B+ B- signals separately with the probe tips. Any one of them should be sufficient to see if pulses stop separately. If you want to do a differential measurement you can use two probes one on the + and the other on the - and use the math function in the scope to subtract them.

It's not clear how often you can reproduce the problem. I can't think of an easy way to trigger the scope when the pulses stop. The nice thing about those Rigol scopes is the large memory depth, I think 24M samples. So sampling at 1us, 24 seconds can be captured. Then you can scan through and zoom in on problems with great detail.

You could write a KFLOP C Program to monitor the encoder position. If it has been changing, then instantly stops, pulse an output to trigger the scope.

Curious to know what you find.

[mmurray70]

Ok i did some quick tests with the scope. Really dont have much confidence in what im doing but i did get to see some distorted signals.

Scope power cord is not grounded. Lead is grounded to kanalog ground. and probe is touching the screw where A and A- are connected for the 4th axis.

Seems to be a ton of noise. Is this as bad as it looks? Could it be something im doing wrong? This is with 4th axis rotating very slowly (75 servo rpm). Going much faster the noise tends to drown out the signal. The noise in the first picture is showing up on all the encoder inputs. Even without anything moving. Its much better with the dmm drive powered off, but still some noise there.

[mmurray70]

I checked few things on my original fadal that works perfectly and seems like im getting lots of noise there too. Also showing some noise on all encoder inputs even when not moving. Heres a shot of the spindle encoder input to kanalog with spindle running.

Looks terrible, but this machine works prefect. Am i doing something wrong?

[TomKerekes]

That's crazy noise. Are you measuring at the Kanalog or DMM end? If Kanalog, do you have the scope GND (Black alligator lead on probe) connected to Kanalog GND as close as possible to the Kanalog Encoder inputs?

[mmurray70]

That's crazy noise. Are you measuring at the Kanalog or DMM end? If Kanalog, do you have the scope GND (Black alligator lead on probe) connected to Kanalog GND as close as possible to the Kanalog Encoder inputs?

Yes measuring at Kanalog. Alligator lead on kanalog ground just below encoder inputs (one section of screw terminals removed and clip is on the pins). Tip of probes is on a 1" long piece of wire added to encoder inputs A and A-. So im not doing anything wrong, its actually that bad hey? What would be the voltage of these spikes? The first picture is a 2v scale, is this 2v for full screen or is it 2v for each grid line?

The yellow signal in first picture is showing up on all encoder inputs even without anything moving. I think this is the source of most of the noise in the second picture too, just different horizontal scale. I will investigate more on monday, off for a few days now. Any other suggestions for my next tests?

[TomKerekes]

You might turn on the scope's BW Limit to get rid of ns spikes.

You might connect the probe to the Scope GND with and without a connection to Kanalog GND. These should theoretically read 0V with no noise. This would give some idea of how much noise is actually on the Encoder signals vs just being picked up by the probe.

[mmurray70]

Hi Tom,

I spent fair bit more time with the scope on the mill today. Most of the noise is definitely from the DMM drive. As soon as it powers up everything gets real noisy. I put the probe alligator clip on dmm drive ground and probe on the encoder output connector housing and got over 100v and a strange wave form, see below. Probing some of the pins on this connector gave a similar result, and others (assuming the ones used for A/B outputs) were lower in voltage but way to noisy to even see the pulses. I disconnected all input/output and results were the same. Probed the high voltage power coming into drive and it seemed reasonable.

After messing with that I put the DMM #1 drive back in to compare and its pretty much just as bad. Everything really noisy with it on. But everything works perfectly! we ran it for 3-4 hours today and everything 100% perfect. I dont get it. DMM is not offering much help so far. Told me they are looking into it almost a week ago without any update. What a mess this is turning out to be.

Mark

[TomKerekes]

Hi Mark,

I would expect something like that if the scope was not grounded. Obviously the measurement is incorrect as 100V would have already damaged Kanalog.

It looks to be around 20KHz which would be the switching frequency of the power output. The DMM power GND and Logic GND are probably isolated. Are you sure the scope GND is on Logic GND and not Power GND?

The scope probe has a 1X 10X switch is it on 10X?

The scope probes should be tuned. On the lower right of the scope there are two points to connect the scope probe and scope GND that has a test square wave. Adjust the potentiometer on the scope probe until the signal looks square.

You might also use single mode to take one measurement rather than many on top of each other which might be less confusing.