G76 -Somtimes- faults and does not synchronize

Once in a blue moon I am having issues with the G76. I did a batch of 96 bars with threads on both ends and two of the bars the synchronization came completely out of whack. And what is odd, I don't know what happened different during those two passes compared to the others. What also is noticeable is it seems to have a smooth shaft. No half threads/almost threads/other pieces showing where the thread was.

I passed that off as odd, but didn't really matter. Today I did a new piece with an internal thread and the same thing happened. It seems the thread is off by a few degrees each time or something.

I do have feed and speed control, but I had those locked in at 100% for both cases. I make sure they are locked in place but I don't know if that could do it.

What can I look at to try to figure it out? My current bet is maybe the encoder, but the encoder has a very solid signal. Contrary to one of my prior post, I do have a A,B,Z on the encoder.

A smooth cut could possibly be caused by a bad index and multiple passes
it definiitely does seem encoder related.

Here is the encoder signal when it was spinning at 1000rpm. As is I have ferrite rings on both ends of all the pairs for A, B and Z to reduce noise signals. The ground for the measurement is pre-ferrite, and the Z has a ferrite core right at the end, so there may be excess noise there.

I may be tempted to get an optocoupler to clamp down on some of that noise, but I am unsure. It seems clean, but my guess is the Z is not as clean of a signal for the Mesa

I take back what I had before about the Z signal. Looking just between Z and ZO, here is the signal. Clean, just under 5V, not much noise. I may have the Z and Z0 signal reversed, but no noise problem there.

I am stumped. The signals are clean and clear. This is the spindle rpm signal when spinning at 2000rpm. When I turn off the spindle it does a clean signal slowdown. The steady-state rpm isn't that clean off of the motor. The old spindle motor is a shunting motor that is in bad need of replacement, but I don't think the spindle is the issue as I have used the machine before to produce a clean thread. It was only twice on the last batch of 200 some-odd thread cuts and now every piece I try produces a bad cut at 100, 300, and 600 rpm.

What can be causing loss of synchronization and what can I do to see if it is lost?

This seems more like a total failure some times rather than a random noise
issue.

Is it possible its a mechanical/internal issue with the encoder?

(I once debugged an encoder issue where the encoder would only count
accurately in one direction, it turned out the encoder had a little hairball
that would get dragged into the optics in one direction and pushed out in
the other )

I can look at the encoder again. The encoder is on its own separate timing belt on the main spindle. I made sure it was tight, however it is a (slightly) worn sprocket. I wonder if that wheel is not 1:1 with the spindle. Maybe? The belt is tight and I cant make it slip. I am not sure how to fix the sprocket or belt or how to find the same size sprocket or how damaged the spindle's sprocket is. It is wrapped around the spindle near where the collet closer is.

I am only spinning the spindle one way, so it isn't that, but I will take it apart and see if it may need some cleaning. I did spin it and measure the pulses on the oscilloscope. Everything looks like a clean signal.

My current guess is that the diameter of the encoder wheel is worn smaller than the spindle wheel. There are the right number of notches to make a 1:1 ratio. So if the belt is loose but held on with the notches, it worked as both had the same number of teeth. By tightening, I think the belt is jumping and snapping into the grooves, sometimes jumping forward and back, as the diameter of the wheel on the encoder and the spindle are slightly diffetent, maybe by half a tooth or so. This made noise on the rpm signal and slipped the engagement for locking the spindle.

Pictures of status
Belt looks good. I measured (roughly) 1.61"/8 divisions. I believe it is an XL belt

Main spindle pulley looks good

The driven pulley. Oddly enough, I believe the pulley did not come with flanges. The small divots on the side seems to suggest it was originally flat, no flange.


I got a price from SDP for $145 for the pulley. Toronto Gear for about $34 but will need boring and machining. Hardinge says $1450 for the gear + $1024 for a new shaft as the shaft design has changed. I am going with Toronto Gear

Other pictures if people are interested:
Pulley linked to encoder

Encoder linked to spindle

Encoder with cover

Aside, and not the encoder on the spindle, but this is whats inside the Dr. Heidenhain encoders