Considering a Full Rewire on a Working Schaublin 125 CNC

I don’t think it can be the Z-axis clutch, as I have a glass scale on Z.
So a slipping clutch should either be corrected or, more likely, end in a stop due to following error.

Sorry, I forgot that. Can you get a scope onto the spindle encoder index pulse?
 

You don’t have to apologize at all. I had a look at the index signal before, but not with the rest of the machine running. So I will test this with the axis drives and VFD running to make sure.

Shot in the dark here...

When I switched my spindle encoder to a fairly high resolution version I ran in to intermittent problems with the signal.  Turns out Mesa cards have a filter on the dedicated encoder inputs that can be adjusted:

Setting it to False sorted things out in my case.

I've no idea if this is causing you a problem, but it's a easy & free thing to test.

After that, next suggestion would be to remove all the PID loops and other stuff from the spindle motor control loop.  Run the VFD open loop, variator set to one position.  Take everything that changes spindle speed out of the equation and do another series of G33 tests.

I don't know that the spindle speed variations are affecting the threading, but it cant be helping things.  Listening to that WAAAHH-uuhhhh-WAAAHH-uuhhhh in the video was driving me nutsimage missing /media/kunena/emoticons/wink.png

Next thought about the collet closer...

I'd measure the spindle shaft and see if it's still round.  If it is, and you can get the bearing rings out, it could be possible to use the air fitting threaded holes to inject Moglice in to the annulus between the rotary union bore and the part it rides on. 

Once its cured and you've removed it - not a trivial task to be sure - you could use a grooving tool to clear out the existing air grooves.  Maybe even hone the cured Moglice to provide a much smaller gap - less air leakage, but still some to counter dust ingress and provide a little cooling.

If you're not keen on Moglice, maybe shrink-fit or glue a couple pieces of thin-wall brass or bronze tube inside the union - leave a gap between them for the air grooves - and bore it out to a closer fit.

Last option would be to groove the union ID to fit a couple of teflon seals.  Shouldn't be too hard to find some low-tension rings that just fit around the rotating part and don't drag too much.

Thanks for the ideas. I will try running the spindle in open loop and see what that does. I noticed the spindle variations in the video as well, while using the machine, I did not even notice them. More spindle tuning is needed, it seems.

The filter is set to “True” right now. I will try turning it off to make sure that does not change anything.

Regarding the collet closer, I thought about cutting grooves, but as the part is hardened, I don’t like my chances of getting that right. Moglice seems like the best way. I remember Robin Renzetti using it on his D-bit grinder.

The only issue there is that the shaft has holes to port the air to the cylinder. I would have to close these in some reversible way and then machine them back down to the shaft diameter to be able to remove the collar after injecting the goo. Epoxy putty, maybe?