Weird problem

[[SH...]]

Machine spectrum, XDT sensor.
Today I was checking a regular job (5 nos), four numbers ran successfully, all dimensions ok. Fifth job i got weird results few dimensions out by big margin. Nothing wrong in loading , no problem in fixture. I run it again I got same result, then I check the repeatability , there is no repeatability in the measurements l got 0.2 mm variations. I removed the stylus system and loaded another one, that was working fine good repeatability.Again i loaded the old one and observed very closely, l could see probe was slipping ( unsual) while taking points .
It did not loose and picked it up two times still no change , then I did geometry requalification still there is no change, finally i did passive qualification and it starts working well as normal conditions, good repeatability.
Anyone faced this like problem,why does it happens?? 🙄 🙄

[[Cl...]]

Please sign in to view this quote.

It may have something to do with the fact that the Geometry Re-qualification doesn't scan to calculate the bending parameters.

[[SH...]]

Please sign in to view this quote.

But why suddenly it gives useless result.. ❓

[[Er...]]

The slipping indicated the probe was not where the CMM expected it to be. But a re-qualification should have fixed that problem.

Are there carbon fiber stems involved? Could be a previous "hit" damaged it and it just now "cracked" open. That would explain passive qualification fixing it. It is flexing differently.

That other plates worked ok says the sensor is ok.

Were the diameter and sigma as expected?

[[Da...]]

I once had similar problems. On of of my brand new probes I couldn't get good results. It turned out that the glued on ruby ball was slightly moving (ever so slightly that there was no way to observe it optically). During calibration, Calypso measures the bending parameters by using a small and a high force to measure the differences and then save those bending parameters permenantly. During the high force run, some probes (either the stems or the ball or the screwed joints or the glued joints) will deform (in a non-repeatable way), giving bad results, but when you do a point-only recal, only a standard force will be used, and all of a sudden, the stubborn probes will pass.

If you wish to be on the safe side, exchange your probe, and throw the old one in the trash (if a new one of the same type gives better results).

There can be situations though, when point-only calibration is the best way to go from the start. I've had a probe once with a length of 400 mm to one side and 500 mm to the other. With point-only calibration, it had a repeatability of only 0,0005 mm (yes, under 1 micron) from one ball to the other. You just have to stay on the slow side when scanning, because the bending parameters can't be trusted to compensate the errors when you scan too fast in those cases.

[[Lo...]]

Look at the Ruby under magnification. Same issue, years ago, found a fracture in the ruby. Replaced it and it cleared up.