Mach/Granite losing position

[Kevin77]

Hello all,

I'm having an issue with the lathe I'm building. I'm using Mach3 and a smoothstepper to control a VSD-E in dualDC mode to drive two electrocraft brush DC servos. For the most part the setup runs well, but I'm losing position slightly. The X axis isn't bad, but the Z axis can lose or gain .010-.015" while running a part. Occasionally even more. I'm using US Digital encoders at 4000 counts per rev, with a 5 tpi ball screw giving me 20,000 counts per inch. I've played with the PID settings in DCtool, and gotten the following error down to 6-8 counts max, with the drive set to trip at 12 counts error. Settling time is around 50ms on both axes. The curves don't look great, but those numbers don't sound bad to me.

A bigger issue is when machining an arc or angle (chamfer), I can really lose position (.020" or more.)

Where should I begin looking to sort this issue? Mach? SS? VSD?

Mechanically, the machine is tight. Ways, ballscrews, belts, etc. are in very good condition so I don't suspect that to be the problem.

Also, the step multiplier in the drive is set to 1 for both axes.

Any ideas? I much appreciate the help!

Kevin

[osphoto]

What CAM software are you using?

I use MecSoft CAM and have had a similar issue on my cnc machine. I checked every on my machine. Turns out that the CAM software was causing the drifting in the cutting process. I re-computed the gCode and it took care of my problem. I want to say it was some sort of numerical rounding issue in the math. Not sure, but it was just weird in that my drifting only took place on two letters of A and S in the wording of allspark.

All the other letters had no drifting taking place. Which pointed the problem away from the actual machine hardware and pushed the issue some place in the software computational issues.

My fix to my drift was re computing the gCode. And I was using ARC's in my cutting process.

[Kevin77]

I'm writing the code by hand. I did change from constant velocity to exact stop mode, which helped immensely, but still slowly loses position.

Kevin

[osphoto]

I did the constant velocity to exact stop mode but saw no differences.

Are you able to track the issue to any particular piece in the code?

[Kevin77]

The issue is most noticeable during arc and angle movements, i.e. when interpolating movements between two axes. Straight line moves in a single axis rarely showed problems.

[Xerxes]

It is probably either caused by noise in step/dir signals or encoder signals. Can you post photos of your installation & wiring? Maybe there's some problem that we can spot.

[Kevin77]

I don't have any photos, but both the encoder wiring and CMD wiring is done using double shielded (foil covered by braid) cable. The encoder shielding grounds to the drive, and the CMD shielding grounds to the breakout board. The encoders have caps on them across the power and ground connections to suppress noise there, as well as resistors or caps across the differential pairs at the drive (whatever was recommended in the drive manual, I don't remember.)

For what its worth, I have a VFD running my spindle, and having that on doesn't seem to make the problem worse. VFDs are generally quite noisy.

Thanks for the replies,

Kevin

[Xerxes]

Hmm, sounds like shielding is well done.

If the problem occurs only with acs, it means that two drives/motors interfere with each other. Solution to it may be found from ferrite beads:

EMI suppression cores - Granite Devices Knowledge Wiki

Also, try separating wires and electronics little bit further from each other.

Let me know if this helps!