Qualification Vs Verification

[[Do...]]

Hi all,
I have been reviewing various posts and polls related to the topic and am interested in obtaining additional information. Currently, our nightly routine involves running a probe re-qualification program, followed by a morning verification. The verification process generates reports highlighting X and Z deviations, along with sphere diameter and form.

We are stuck in an ongoing cycle of re-runs, with operators mandated to verify that all probes adhere to the designated tolerances for these parameters. Consequently, this situation has resulted in extended periods of machine downtime and incurred costs for replacing probes that might actually be in proper working condition.

I've come across comments suggesting that the XYZ values should never be "0" during this test. While I agree with this perspective, I am curious if there should be a consistent tolerance that we can anticipate hitting, or if these values are arbitrary and do not necessarily correlate with the probe's quality. Should we really only be looking at sigma values, diameters, etc.?

Ideally, I feel we should be running re-qualification on Mondays and only running verifications during the week, with exceptions for incidents like collisions. What specific targets should we aim for on a daily basis, and what standards do you typically adhere to in your operations?

Also, should there be additional consideration given to CAA calibrated probes as they typically have a higher sigma value during calibration?

We are primarily operating XXT probes running on 2014 Conturas. Any insight would be greatly appreciated.

Cheers,
Don

[[Mi...]]

XYZ mean very little and honestly won't mean anything meaningful to your standard operator. You can use them to compare new values to old, or current vs expected. You can use them to judge how well you build a stylus system etc...

This sounds like a training and cleanliness issue more than anything.

We have 33 CMMs, 2 programmers and half a dozen people that will calibrate. 9/10 times somebody is struggling with calibration myself or the other programmer will redo it with no issues. We take the probes they say are no good and run them somewhere else with no issues.

We set limits of 0.001 for Sigma on Direkt XXT, 0.0003 for XT and 0.003 for our RDS (With Renishaw probing - yuck). But I hold myself to half those and extremely rarely have any issues meeting it. We don't use CAA calibration on our RDS Machines... we calibrate in all the angles used... but we only have about 30 angles.

They think they are cleaning well enough - they aren't. At the very least don't let anybody but your most experienced decide a probe is no good. I bet you've got a person who has no issues, find out their cleaning method and adopt it as a standard practice.

[[Ow...]]

Michael Shultze is right on track regarding cleanliness. Just a hair or spec of anything will throw the standard deviation/sigma out and temperature fluctuations will alter sphere size recorded. Isopropyl alcohol works best for cleaning to ensure there is no sticky residue left on the probe or sphere as most common household cleaners will leave a residue and then everything will stick to it.

Regarding xyz numbers, I believe they are irrelevant most of the time, maybe excluding cracked carbon shafts or flat spots but , if you're concerned about them, create a program that will scan the reference sphere with the master-probe to set location and size and then scan it with each probe used and report the size and location.

If they’re just wanting to ensure the CMM is repeating as it should, you might just need a known ‘Master’ part that covers the majority of the axis’s, probes, size and form of the features/characteristics commonly checked on the CMM. For instance, maybe keep the largest or average size part you make or check routinely in the CMM room and check it before calibration, after calibration (to see changes, if any) and ESPECAIALLY after a significant crash because even a good probe calibration WILL NOT detect if the CMM has been knocked out of square.

[[Mi...]]

You can accomplish a good chunk of the verification with a ring gauge or two as well. We have a program to check the same circle with multiple probes/probe systems as an indicator of needing calibration. Depending on what parts you run, this may be enough.

For us, we have so many CMMs with the same programs it's easier to run the same part 3 places and then calibrate the one that doesn't agree.

[[Cl...]]

We do a verification with several Master Parts. Ensuring all styli are used.
Then Applying a tolerance to the actuals average gathered from historical
data. Then a tolerance is applied based on the (min/max) range.

[[Do...]]

I appreciate all of the feedback so far. I had also failed to mention in the original post that we are using the Zeiss reference sphere for both qualification and verification. The verification program checks probes in the cardinal directions as well as A0B0 and A180B0.

When verifying a stylus system via ring gage, are you measuring just one orientation or at multiple angles? I can look into a master part option as well.

Cleanliness has not been our strong suite and I've requested that operators wipe down guideways, probes, adapter plates, etc. on a daily basis. In the past, operators had a general housekeeping checklist, which was filled out half-heartedly and/or when someone remembered do it. I am looking into creative ways to ensure compliance with these PMs and if anyone has a process that works for them and their team, please chime in.

To Owen's point, we are scanning the sphere with the masterprobe and then reporting the XYZ values, size and form for all of our high running styli. We had at some point settled on ±.0004" tolerances for the positional values, which is where we are having the most trouble. We maintain tighter tolerances for the form and diameter to .0002" and rarely have issues there.

With that in mind, should values be expected to meet that tolerance given the method described or is it common to see results that exceed .0004" in a single direction? I will look into pulling historical data to see what our range is and determine if an increase to the tolerances is required.

Thanks everyone!