Message Thread:
Servo creep on CNT Motion Router
8/6/16
Good day folks,
We've been experiencing a troublesome servo creep, which results in out-of-square cabinet parts somewhat intermittently. About half the sheets we cut are just fine, and about half have parts that are anywhere from "almost square" to 2-3mm out of square.
This happened to our X axis last winter, but were able to isolate the problem to a faulty motor drive card, so we replace it and the problem went away. A few months later, it started happening on the V side.
I have talked at considerable length with the tech support at CNT Motion but seem to have exhausted almost (almost) everything they can suggest.
Here is what we've found through observation and the use of servo-tune software. We will home the machine and calibrate the servo-tune software to reflect "motor position 0" and then we'd cut a sheet, and then command the machine to go to position Y0. The resulting motor position tells us if it's going back to 0 or not.
About half the time, it goes back to 0. Other times it's out as much as 2-3mm (which explains why some parts are 2-3mm out of square).
X and Z axis has been checked, as well as the Y-side servo. NONE of them give us any creep. They always return to 0.
1) The problem always, always happens on the V-servo side of the machine (Y axis has two servos, the V and Y servo).
2) We have tried three different motor drive cards on the V-servo side, no change. Thus I think we can rule out the card. One of those cards is hardly 8 months old.
3) We have put every servo motor on the machine in the V-location. No change. V side still creeps, none of the others do.
4) Replaced the pigtail wire that goes from the V drive card to the V motor. No change.
5) Replaced the main pigtail wire that goes from the CNC computer's main PC Cabinet to the actual CNC itself. No change.
6) Ran test cuts (empty spoilboard) with dust collector physically disconnected from the router, vacuum pump off, and every electrical device (including overhead lights, edgebander, etc) turned off. Trying to see if there's perhaps interference from electrical devices causing this. No change.
Any suggestions? I'm happy to answer any other questions related to this.
8/7/16 #2: Servo creep on CNT Motion Router ...
Have you tried taking the servos off and moving it by hand to see if there is and drag on one side.
8/8/16 #3: Servo creep on CNT Motion Router ...
Can you swap the servo cards or servo amps between the motors. Maybe its the card not the servo.
Maybe bad settings or faulty card?
8/9/16 #4: Servo creep on CNT Motion Router ...
Jerry, we did have a bearing that sounded worn (making "slightly" more noise than it should), and the shaft attached to the servo (on that same bearing) was running slightly untrue. I cannot fathom how that would give servo creep, because it's not slipping and the belt that attaches to this is toothed.
We did replace the bearing yesterday and have ordered a new shaft.
Regarding the cards, yes, I have actually tried every card on this machine except the Z card to run the V servo, with no change. One of the cards is hardly 8 months old (X axis) and the X-axis runs perfectly fine.
We have also moved all the servos around, as they are all 4 universal. Every servo on this machine has spent time on the V-servo side, with no change.
We are going to monitor today's sheet goods to see if yesterday's bearing swap made any difference, but I'm skeptical that it did. The bearing we pulled off was not all that worn, just made a bit more clicking noise than a new one when it's under load.
8/9/16 #5: Servo creep on CNT Motion Router ...
Different machine, different symptom.
I had a older CNC that would occasionally throw an e stop on the rapid back to 0,0,0. Spent about a year trying to track it down. In the middle of a bundle of cables was a woven ground strap that had worn through. The bundle was holding it together for the most part, but occasionally the ground would break and throw the fault.
Just an idea.
8/9/16 #6: Servo creep on CNT Motion Router ...
Dropout, that's a good thought that has crossed my mind before. I've checked the obvious ground wires that are easily seen, but haven't done more than a cursory examination.
I'll take a closer look on them. Thanks,
Matt
8/11/16 #7: Servo creep on CNT Motion Router ...
Also check the shielded wire and grounds on the VFD for the spindle.
8/17/16 #8: Servo creep on CNT Motion Router ...
Sounds to me more like an EMP situation. Your controller, or as some call it, amplifier, may have picked up a dirty AC line. I have battled this over the years and you have to be part detective. Think back to when you first noticed this problem. Have you added any new machinery in your shop? Not knowing how your AC is fed to your building I would also add even if you have a new printer in the office count it too as capable of emanating AC feedback into the same line your amplifier uses. In my case a new computer was added next to the one controlling the CNC and ran the monitor wire next to the cable that ran to the amplifier on the side of the router. It drove me nuts for months because failure was on Z axis and also was intermittent. It improved after I put some carbon chokes on all the AC cords but did not go away. Looking back every time I moved a monitor or jostled something or repositioned the monitor the behavior would change. Sometimes it would be reduced, other times it was a nightmare. It got so bad that when the spindle revers came on (when turning the spindle off) the Z axis would ratchet up 2 inches. I could not even plane my spoilboard I would be cutting air by the time it finished. I then added an APC backup and plugged the amplifier into that, all by itself. There may be some filters in the APC. I don’t know but that did it. Problems evaporated. For me it was so bad and unpredictable I could not even do a pocket operation. It affected X axis but not as often. X would just shut down and the arc it was cutting turned suddenly to +Y and I thought I was gonna go nuts. Here is a brief todo checklist
1 Next time you are running a job that you think may cause the anomaly to show up, unplug every machine in the shop but the router and run the job. If that doesn’t work….
2 Make sure you have not disturbed the run of wires that pertain to your router. Make sure they do not cross each other and also just as important make sure they do not run next to each other.
3 For good measure throw some chokes on any AC wire around. Put em on everything. They are cheap and useful
.
8/18/16 #9: Servo creep on CNT Motion Router ...
We actually did just put an isolator on the actual PC for the router, if that's what you mean by "choke" (looks like a black voltage transformer).
We did this yesterday, and the first half dozen sheets we cut still had SOME drift on the Y axis, but the parts were only about half as bad as they had been (meaning, about 1mm out of square instead of the 2mm).
That's not enough of a sample size to say that we've found the smoking gun, but we will know more as we route more over the next few days.
But overall, Rich, I think you've probably hit the nail on the head. What you said is the same thing our supplier tech support said could be the issue, so we bought that isolator and just got it yesterday. Thanks for confirming this as a potential solution.
8/18/16 #10: Servo creep on CNT Motion Router ...
Well, we put the additional isolator onto the PC and monitor yesterday, initial tests indicated that "maybe" we had reduced the problem, but definitely not eliminated it.
Today, we've run several other tests, including one that I've been using repeatedly as dry-passes (no sheet goods), and we're still having the same, consistent problem.
Another indicator is that, after each sheet, I can type in G53Y0 to send Y axis back to where it thinks 0 is, and it's consistently 1mm off from where it should be. Verified both visually (using etchings on the router) and using servo-tune software. Anything from 0.75 to 2.25 mm off each time.
8/18/16 #11: Servo creep on CNT Motion Router ...
For me, Z was affected most for some reason. X and Y also but the movements are more pronounced in Z. I could run a job and when it finished (I insert a G0 X0 Y0 at the end of each of my jobs) Z would be .1" higher. Sometimes during the job Z would not go all the way down and the end of the job we are left cleaning up parts that weren't fully cut out. Planing my spoil board was futile as you can see in the pic. I consulted all the experts to no avail and stumbled on the link by accident when moving some wires around my desk. EMI is unpredictable in frequency of attacks. There is something you did recently that has caused this change, and it can be a brainbuster tracing all causes. My buddy had the same problem with a huge Trumpf Press brake they traced to a neighboring shop causing interference. Had to run in new AC to solve it. Mine is so sensitive it acted up once again and I traced and verified it was caused by my phone charger. You read correctly, I could cause Z to tick up by plugging my phone in.
View higher quality, full size image (2560 X 1920)
8/18/16 #12: Servo creep on CNT Motion Router ...
The measurements you give indicate the frequency of each tick on the servo. Run a job without the spindle on and see if that helps If your servo's are exposed, or visible and you can put a hand on them as they run, see if there is a tick, you can feel it, as it moves in X or Y. IE, there should be no X move on a line drawn in Y, right? So if you feel the X servo 'tick' during the move in Y, you know it is an EMI.
8/18/16 #13: Servo creep on CNT Motion Router ...
Well I meant to say run a job with lots of straight lines but with spindle OFF, not on, so you can better sense bad moves.
8/18/16 #14: Servo creep on CNT Motion Router ...
This will scare you if it happens to you
If you are seeing this message, your version of flash needs to be upgraded to view this video. Click here to get the latest version of the Flash player. |
8/18/16 #15: Servo creep on CNT Motion Router ...
Good idea Rich. I'm not sure on my machine how to run a program with the spindle off, but most likely the shop owner does.
Since it's my V servo (second servo on the Y axis) I think I'd need to run a program of only X axis cuts.
Feeling the V servo would be quite tricky on my machine for most cuts though. You'd have to be standing on the spoilboard to do it.
I wonder if I were to use a V point bit, like the miterfold bit, to make a bare-visible etch on the surface of a sheet of 1/4" white melamine if perhaps I could see, specifically, where the V servo failed. Either I'd see one line make a small jog left/right, or I would have a larger gap between some of the lines than the other.
Running that sheet a few times could tell me perhaps if I'm fighting an EMI or similar electronic bug (random V movement), or if my problem is mechanical (movements happen in the same place consistently.
Thoughts?
I am grateful for your help Rich. For what it's worth, I haven't yet tried unplugging all the 110V stuff (shop is rather large so I'd need to do it one day when everyone is gone).
I have, however, done it with all machinery off, dust collector physically disconnected, all lights off, no vaccum pump. No change there.
8/18/16 #16: Servo creep on CNT Motion Router ...
Sounds like a plan. The movements are jerky, one 'tick' at a time and should be detectable. Sorry to overload the forum with my 'experiences' battling this but if it helps anybody tracking down these elusive EMI's then it is worth it.
8/22/16 #17: Servo creep on CNT Motion Router ...
What controller does this machine use?
So far I have never experienced this, but I have only used Fanuc controllers. Interesting, it's as though the servo made the initial move correctly and the encoder reported back to the driver correctly and then then did not.
8/22/16 #18: Servo creep on CNT Motion Router ...
The controller on this is a Windows PC running WinCNC.
Rich, glad that you shared your experience here. We are going to start doing "unplugs" to see if anything changes. Something, somehow, is interfering with this machine and it's driving us nuts.
8/22/16 #19: Servo creep on CNT Motion Router ...
Happy to Matt. I to have a PC based unit running Wincnc. I would check the 37 pin connector from the back of the PC daughter board to the amplifier to make sure it is not falling out.
9/21/16 #20: Servo creep on CNT Motion Router ...
Website: http://plasticworks.ca
Matt, I had a very similar problem on a CNT 1000. Do you have the servo drives on the gantry?
The cables from the PC control to the servo drives is a weak link. Any emi can add steps that the servos will accept as proper movement commands. No encoder fault if they think they are where the steps said they should be. There is no feedback loop to the PC with systems like this (nearly any step and direction servo drive is the same, but most are in the same cabinet as the controller, well wired and no movement)
In my case, the @$%! tiny 220v spindle cooling fan power wire had developed a fault. We had noticed it was intermittent, but mostly on and we don't tax the spindle so a fix wasn't high priority. Un-seen was that this fault added steps along the cable chain
On my CNT, there is a switch that gives power to the spindle. how I found the problem and how I'd suggest you test is to have the servo test software open, and turn on and off power to the spindle, and maybe other component if you have, and see if the encoder counts move.
It did in my case and that quickly led to the problem. I have stuck a 110v fan overtop the problem fan (or fan with the problem power line) and just fed a 110v extension through the chain to fix. Hokey, but better than 2 years of phantom movement.
Since you haven't tried without the spindle, the scenario sounds likely to me symptom wise, and certainly you'd want to rule it out. I think it had occasionally effected the Z and W vertical axis too, but the dual drive is the worst.
Call me at the 1800 on the web link if you want to discuss.
9/21/16 #21: Servo creep on CNT Motion Router ...
Greg, that's very helpful, thank you for responding.
Yes, our servo drives are in the gray box on the back of the gantry.
Following your information, we tried to disconnect the fan and run a dry-pass (cutting only air), but WinCNC considers the disconnected pigtail to be "Fan over temp" and aborts anything we try to run.
I'm working with CNT to find a way to disconnect the fan without tripping the safety. Our fan runs the entire time the machine is powered on (not just when the spindle is running), so I can't just toggle the spindle off to shut off the fan.
The fan is indeed a 220v like yours is.
9/21/16 #22: Servo creep on CNT Motion Router ...
Website: http://plasticworks.ca
On my machine, one of the two black switches on the cabinet (similar to the photo here) turns off the power to the spindle....presumably for safety on manual tool changes....that's what I used. Honestly, if you think it's only happening during movement, then it's a harder problem to trace.
Also, when talking to CNT, they suggesting taking out fuses on the cabinet side to cut power, which is probably the way to do it and not on the spindle side. You might be able to trip the in-cabinet relay/breaker on purpose? But yours sounds like the spindle is a little more wired up if it's warning about the fan.....my setup did not. I'm sure you and CNT /winCNC guys can figure that out.
Hope they are taking these problems as a good reason to take those drivers off the gantry.....even a box on the table, like Cam-master does is a better design....and the chinese routers that compete with CNT and Cam-master have them all in with the control, which is better again.
How old is your machine, out of curiousity?
9/22/16 #23: Servo creep on CNT Motion Router ...
Ok, I do have a toggle for the spindle on WinCNC, I just never need to use it, so I had forgotten it was there.
Tested with servo tune software, and toggled spindle on and off quite a few times, and nothing happened.
We also pulled the fuse on the 220v fan, which allowed us to run without tripping the over-temp alarm....... and we ran a few test panels, and we had no improvement.
I'm not sure in my case the fan can be considered a culprit anymore.
I do greatly appreciate your help. Our machine is 12 years old, so shes no spring chicken.
9/22/16 #24: Servo creep on CNT Motion Router ...
Website: http://plasticworks.ca
Ok, assuming it's not the spindle, and that you don't have 2 spindles or extra drills, etc, then you are left with the servo motor power feeds as the likely source. Tough to test anything with those off...but might be possible as long as it's not either side of the dual axis.
The only way to fix that I can think of is to a) get a new cable made up or b) make your own cable/cables for the data connections and leave the existing cable in place.
I was planning B for my machine, but as I am selling, may just try to run a new wire for the fan and call it a day. Here's how I would test this approach. Take an shielded ethernet cable, two "keystone" type female ethernet jacks, and trace which wires are the problem motors step and direction signals (and probably enable too) and bypass the existing cable from the wincnc daugtherboard to the servo drive. I would not bother putting it in the chain right away, just leave it loose to test, and that also keeps away from the unknown emi source, assuming I'm correct on any of this for your issues.
If that works, you could repeat for all the axis and run a separate chain on the other side of the machine to the box on the gantry, and problem solve in a robust and future proof way, with added bonus of easy to change wires for each axis.
9/22/16 #25: Servo creep on CNT Motion Router ...
We actually do have new Set A, Set B cables on the way, so you predicted that exactly lol.
That's an interesting idea about making our own cables, I will keep that in mind if the need should arise.
|