Servos pausing and starting

[InsaneEPP]

If anyone can help me find out what is causing this I would be extremely grateful. I added a differential driver to my open ended encoders to reduce noise, etc. That worked well. While I was rewiring the encoders I rerouted everything so the power and signal lines were in separate energy chains. There are a couple of spots where they get within 6" or so, but are usually at least a foot apart.

The servos are rock solid when not commanded to move. If I run a part program, they do fine for a minute or so, but then the Y and X axes will pause and start/accellerate violently. The machine start rocking all over the place. I don't think it is noise because I wouldn't think the servos would just sit still for .25-.5 seconds. It also looks like they go back to the correct position. I read that if the drives are not cooled well this could happen. I felt the drives (as best as I could) while mounted and they did not feel hot. I added several heat sinks and a large fan to the drives, but that has not corrected the problem.

Possible issues:

I do not have shielded cable running to the servos. I called around last week and no one carries 10 AWG 2 conductor shielded cable. Anyone have a source? Not sure if this is the reason for my problem.

Loose power connection to the servo/drive? I wouldn't think this is a problem since it seems to be fine initially.

Driver issue? I have 990Hs. My servos are 10A continuous and I am running them at 67V. I also read in the driver docs to "Make sure that no motor wires are routed above or close to the servo drive." I don't have power to the servos routed near the drivers, but like I said before, it is unshielded so perhaps this is limiting the current to the servos?

Am I missing something? Is this really noise or what?

[HuFlungDung]

"Running a part program" could mean a lot of things
Suppose you had a bunch of short segment moves, forgot to change modes from G00 to G01, or reduce the feedrate, or inadvertantly switch from G90 absolute mode, to G91 incremental mode.

Run your tests on simple, linear movements, until you get a handle on what is happening. I'll assume you have properly tuned the servos before all this?

[InsaneEPP]

Well, I have OneCNC XR Professional and that is where the part program come from. I made a 2" hemisphere and a did a few operations on it. Anyways, I was cutting air above the tabletop.

The machine will do simple moves already. I haven't had a problem with that since I rewired the machine with the differential drivers for the encoders. I have tuned the servos although (to me) it seems like there is a wide range of values that will work adequately.

I don't think the problems I'm experiencing are coming from the controller (Mach 2) on up. I set the accelleration low (30"/s/s) in Mach 2. I then ran the program and you can tell when it is going bonkers. The pauses and the subsequent accelleration is more in the 150"/s/s range (it is really rough). It's the same type of accelleration you get when doing the tuning software. So I am assuming there is error in the servo's current position versus the intended position.

[HuFlungDung]

Ok, then this is most likely a very short segment program. Maybe someone else can chime in here with the required settings for you to make in Mach2.

I am guessing that the rough motion has to do with the number of lines of "lookahead" which gives your controller a chance to issue a blended path, instead of stopping at the end of every segment. Believe me, if your cnc does stop at the end of every movement, it gets really jittery, and is very hard on the mechanical components especially if the feedrate is fairly decent.

[InsaneEPP]

I have it set for CV (continuous velocity) motion in Mach 2 which should take care of that, I think. It has a setting that will keep the axes moving as long as the next segment is within so many degrees in direction as the previous segment. I have it set to 10 degrees (the default). I actually can't remember if the setting is 80 or 10, but it is the default from when I had the software in front of me. The servos don't come to a complete stop between segments in any case. When it is doing the Z level work it does a nice smooth full circle until it has to go down in Z. The movement down in Z is a little notchy, but not bad. It is *nothing* like when it goes haywire. Also, it doesn't happen in the same place every time in the program. I once had it go through the whole program without problems.

I really think I have something else going on more on the driver or electronics level. I just need a hint in the right direction. Please somebody help!

[HuFlungDung]

What kind of machine is this that you are running on? Is there any chance of "slip-stick" happening on the guide rails? Since you mention that it works well in Zlevel arcs, that makes the guides a little suspect, IMO. With the motors off, can you move all the slides freely?

As far as shielding the power conductors, you should be able to use armoured cable, or tec cable, such as a local industrial electrician may have in stock.

When you ground the shields on your encoder cables, make sure that you only ground them at one end, because they are supposed to act like a grounded receiving antennae, not forming another circuit.

[InsaneEPP]

The problem is that it works ok in the beginning of the part program (the first couple of minutes) and then starts having problems. If I rewind the program and start from the beginning again it happens sooner. I think something is heating up in the driver and shutting off current intermittently. The PC is still sending steps, but the drive cannot apply power to the servos. When the driver finally cools down, it applies max power to the drive and causes the rough motion. The only problem is that I have the drivers mounted to a 3/8" slab of aluminum with several PC processor heatsink plus a large fan directed at the whole thing. I have heatsink compound between everything. Also, I can't feel it getting hot at all, but I can't really touch the drivers where I want to because I'm using the 990 motherboard, which puts the three axis drivers very close to each other.

This is a home built machine. It is heavy duty -- all steel frame with large rails (1.5" diameter open Thompsons, 35mm THKs/NSKs on the X and Y). I'm not sure what slip-stick is exactly. All of the slides run freely, although there is obviously some drag, but the servos are 300oz/in continuous (1500oz/in max) so they don't have a problem moving them at all. I can accellerate at 125"/s/s with no problem other than the machine needs to be bolted to the floor.

It looks like armoured cable is solid wire. I was looking for stranded since it will be bending in the energy chains. Also, I have limited room in the energy chains. Armored cable is pretty large in diameter from the ones I found on the internet. I think I can get the shielded if I special order it. No one here carries shielded over 14 AWG and I need 10 AWG.

As far as grounding....maybe that's part of my problem. All of the encoder shielding is connected together at the driver end only, but am I supposed to connect the driver end to anything else? Am I supposed to attach the ground from the 110 supply (outlet) to the driver ground? I was always confused about that.

[ESjaavik]

And do you use ballscrews, driven directly or through belt or gear (not worm drive).
Most servo controller likes to "see the load", so mechanics with high damping of the backdrive is not good.
Or to put it another way: If you uncouple your motor, you should be able to push the axis and thus make the spindle turn.

[InsaneEPP]

I use ballscrews driven directly and I can push the axes and make the screws turn.

I am now thinking I may have a problem with either the drive power supply or the servo power supply. I talked to RutexUS and it doesn't make sense that two of the drives are acting up at the same time. At this point I don't know why either power supply would be intermittent. At least I have something to check. If anyone else has ideas, let me know. I'd much rather be checking several things than have no idea what's wrong.

I figure I need to run the machine and put a voltmeter on the both power supplies and see what happens as I run it.

[InsaneEPP]

I checked the power supply for the servos and the drives and they were both good.

I also checked for lost steps (ie - ran the program until it jumped all over the place then went to almost to zero on every axis, then to zero). There were no lost steps.

I then went to the diagnostics screen on Mach 2 and noticed that the machine started jumping when the "buffer load" was at 100% for more than a second or so. Hmmm....very interesting. It was repeatable and so I now am quite certain this is the problem. So, after all this Hu was pretty darn close to figuring out the problem!

I am a VERY happy camper now. I just need to upgrade that PC and I should be good to go. WOO HOO!!

It would have been nice if the buffer load was mentioned somewhere in the docs for Mach 2. There's everything else in there.

Thank you Hu very much for helping!