Encoder Issues, Help!

I sorted out the issue with the EXE boxes and ran headlong into another perplexing problem. I now get repeatable movement, but the scaling is not quite right. I carefully calculated the encoder scaling and came up with 24500. Unfortunately, this results in movement that is 0.007" short per inch (0.993" movement when 1" was commanded). There is a 2 to 1 ratio between the motor/encoder and the ballscrew but unless that ratio is somehow slightly off everything else, ballscrew pitch, counts/rev, interpolation should be fixed and would not need fine tuning. Confusing!

Without anything better to do I scaled the encoder count, 25400*1.007=25578. Closer but still off by about 0.002". I tried a number of iterations and got the really confusing results as below.

Scale Error
25550 -0.001
25546 -0.0004
25544 -0.002
25542 -0.0012
25540 -0.007

This makes zero sense. A count of 2 should be only 0.000078" and the direction of the error also changes. This is repeatable and I rechecked everything multiple times. Any ideas?

Did you try the commanded movement in the same direction each time? To ensure you aren't measuring backlash?

That's the only idea I can think of. You are way ahead of me in your build. I'm currently trying to install linuxcnc to prep for my cards next week.

EDIT: are you getting the same results on each axis?

This issue is frustratingly consistent. I get the same error with both directions. I also have measured and compensated for backlash and it's below 0.0004". But I have only tried the X axis, I'll try Y and Z to see if there is a difference.

The ballscrews are 5mm pitch. The thing that is particularly confusing is how the 2 count changes ended up having completely unpredictable effects.

I just did the same test with the Y and Z axis. I see scaling errors with all 3 axis but the amount is different

X = 0.007
Y = 0.012
Z = 0.002

I don't think I could have come up with a more confusing and unhelpful result.

With these Heidenhain encoders [yours, I have the same style], are they 11uApp, or 1uApp? (I would assume 11uApp since EXE is the choice I see all the time)

I've attached the Series manual that explains what each model is capable of.

csbrady,

Did you get yours sorted out?
I'm trying to hook my signals up on mine now, I haven't figured out the signal names that correspond to A, B, and index channels. Would you happen to have a pin out from your exe box to the 7i77 handy?

I have the 602 pin-outs and my pin-outs but trying to figure out what goes where has me perplexed at the moment.

Which one is index, A phase TTL, B phase TTL and which one goes to where on the 7i77 is what I'm needing. I have read a TON on the encoders and have convinced myself that they are 11uApp signals. So the EXE boxes were a correct choice.

I at last figured out the problem. It's a bit embarrassing, but the problem was my measurement setup. I was using a dial indicator on a test stand with a magnetic base. I thought that the results were too consistent to be the test setup so I dismissed that as a possibility. With nothing better to do I made a couple shallow cuts 5" apart and measured them. 5.000" exactly, hooray!!

Thanks everyone for the help and sorry for the wasted time on a dumb mistake.