Clearpath setup losing steps

[kbmyers]

Hello - I've been having trouble with my Acorn controlled router with Clearpath servos. I thought I had everything working correctly, but on a recent engraving job with a lot of fine Z axis movements I noticed that I was losing steps. On programs with broad motions this wasn't noticeable, but it is readily apparent with a lot of small motion. I'm betting all of my axis have this problem, but it is only easily repeatable on Z. I've made a few short videos to show the problem.

I'm using Clearpath SDSK-2321-EQN drives, which have the enhanced 6400 steps/rev. When I first saw the problem, I pulled the drive out and ran a modified "bench test" with the drive spinning freely. I plugged in an older SDSD-2311-RLN and it worked fine. Here's the video:
Test 1 SDSK-2311-RLN success

So I assumed the other drive was bad and sent the RLN drive back to Teknic. They tested it and returned it as no trouble found.
The next video shows the 2321-EQN drive as returned to me with the newer firmware update. This fails almost immediately with a step input timing error. Test 2 SDSK-2321-EQN fail new firmware

The last test shows the drive after downgrading the firmware. The older firmware basically masks the step input timing error. It runs the test, but it isn't accurate and gets out of sync with the control software. This is the original problem. Test 3 SDSK-2321-EQN fail new firmware.

So I started trying pretty much every combination of step rate, max speed and acceleration I could think of trying to find some commonality:

A few random observations:
* Both drives are at the default tune
* The test program doesn't turn on the spindle or the vacuum system. I've also disconnected the 220v power to the control box to try to eliminate any electrical noise possibility
* The test program g-code was output from Vectric Vcarve Pro using their Centroid post processor, but for fun I also generated the same toolpath with Autodesk Fusion 360. No difference
* The issue only occurs with the small engraving motions and not with larger jog commands. The test program is trying to engrave text .006" deep

Attached a current set of the wiring diagrams and a fresh report.

Teknic is saying that the drive is fine, and that I'll need an oscilloscope to further debug this issue. They don't have an answer for why the two drives behave so differently.

I'm working on getting an oscilloscope, but in the meantime does anyone have any suggestions?

Thanks,
Kent

[Dave_C]

Kent,

I had this same issue with my Clearpath Z axis servo. It was dropping deeper as I machined a part.

X and Y seemed to be OK.

So what I did, and I no longer have the problem, was to re-run the tuning software from Clearpath. The reason I think the Z axis has the issue is because the load is different in each direction where as the X and Y loads are more consistent.

It has been rock solid after I re-ran the tuning. If that does not fix it, it might be a noise issue!

Dave C.

[kbmyers]

Thanks Dave - This seems to behave the same way with or without tuning. I've probably tuned this a half dozen times. In the video the servos are running in air with no load.

When mine was connected, it was lifting the head as I machined the part. At the end of an engraving routine it would be about .020" higher than when it started.

[cnckeith]

i think saw something similar about a year ago, solution was to get a new clearpath motor/drive

[Gary Campbell]

Both a previous thread and the ClearPath manual show using a 2k resistor from the step terminal to 24+ to eliminate this issue. Have you done that?

[kbmyers]

Teknic says this motor passes all of their tests and isn't bad. Tomorrow I'll swap the X axis servo for Z and re-run the test.

I do have the resistors installed per the manual. That was one of the first steps Teknic had me do. It didn't make a difference at the time. The original solution was to downgrade the firmware so it ignored the step input timing errors. Now it appears that those original errors were legitimate so suppressing them with an older firmware just pushed the problem down the line.

Also, at one point I had Teknic manually tune all of the axis through a remote session. I was really happy with the smooth motion, but that doesn't count if you're missing steps.

[Richards]

I'm wondering what would happen if you set the counts per revolution to 6,400 instead of 800? Right now, you're only counting 1 of eight pulses. When the motor is making extremely small moves, what happens to those non-counted pulses? Are they cumulative or do they start a new count for each move?

[kbmyers]

I've tried both 6400 and 800 counts per revolution. The spreadsheet attached shows some of the combos I've tried. The older 2311 drive I was testing with only supports 800 counts so I changed the setup to accommodate it.

[Dave_C]

Keith,

I took the time to watch all of the above videos to see if I could see anything unusual. I have two observations:

1. I've never used the CNC computer to run a CNC program and the Clearpath software at the same time. Maybe not an issue but I use my laptop to run the tuning software and then just run the CNC as normal. I'm not one to make the CNC computer do two task at once!

2. I noticed that the servo "jitters" when it returns to a stopped position, or at least it looks that way in the video, so maybe it is stable.

Either way, it is losing steps for sure!

I'd definitely go back to the native 6400 counts and then make sure you don't leave the gearing set wrong.

Since Clearpath has tested the servo and they say there is nothing wrong with it, then we need to look for something else. Obviously the other motor tested worked so that would tell me all the wiring is good and that there are no noise issues. (sounds reasonable)

So all I can see is the setup of the servo itself! This does not seem to be an intermittent issue as it repeats on every run.

With that said, I have nothing else to help....

Dave C.

[Black Forest]

It will be interesting to hear the results of swapping out the X axis motor for the Z axis and see if the problem still exists.