Cal sphere test for verifying CMM measuring correctly

[baseballfan]

Hi. Hope all is well.
I searched for this many ways and could not find anything.

we have 7107, PCDMIS 2014.1 Leitz probe, working in MM

We have over 15 different tips that we use, so we run a calibration of tips used for each routine when running each routine.
After calibration, we run a "cal sphere test" and measure the cal sphere with all the angles of each styli used on the routine. We start by measuring the cal sphere test manually with the master tip with 5 points ( 4 on the equator and 1 on top ) and set that as the origin. Then the master tip in DCC mode to measure the cal sphere again and set the DCC measured sphere as the final origin.
For example, using a 1x20 tip with A90B-900, A90B0, A90B90, A90B180, our cal sphere test ( DCC mode ) would measure the cal sphere with the master tip 15 points at 4 level ( each stylus and stylus angle ) and make this the first DCC measurement with master tip as origin, then put up the master tip, pick up 1x20, put up 1x20, pick up master tip and measure cal sphere again. This allows for checking the wrist and weight balancing by putting up the styli and picking it up again, and making sure the cmm is repeating the measurement.
Puts up master tips and picks up the 1x20 and measure the cal sphere with all 4 tip angles, then puts up the 1x20, picks up the master tip, puts up the master tip and picks up the 1x20 and again measure the cal sphere with all 4 tip angles.

we are looking for results of each stylus angle measurement such that the XYZ nominal = 0 ± 0.005 and the diameter is the cal sphere diameter ± 0.005


We do this cal sphere test after calibration and before running parts, and then again after finishing the part/lot inspection to make sure that the CMM is measuring correctly before and after part inspection.

We change parts on each cmm from one shift to another. Rarely does a lot not get finished on a shift and the next shift has to continue the lot inspection.

Some of these cal sphere test can eat up some time on the clock.


I am just curious

1. how many of you do this, or do not?
2. recommend this or not? ( Hexagon recommend this when we first got our 7107.)
3. If a lot passed, consider to forgo the cal sphere test after the lot inspection was complete?
4. Only run the cal sphere test after a lot inspection if any of the parts fail?

We have had a few probes go bad ( operators crashing them, faulty, undocumented enhancement in PCDMIS that make a probe take off in a direction it is not supposed to go and crash )
and finding these bad probes would not have happened if not for the cal sphere test, and not failing good parts on rare occasions.
I am for doing the testing before and after.
Just wanted to get a feel for what others may or may not do and their thoughts.. for those willing to respond.


Thanks

all the best

[Kp61dude!]

Questions:
1. how many of you do this, or do not?
2. recommend this or not? ( Hexagon recommend this when we first got our 7107.)
3. If a lot passed, consider to forgo the cal sphere test after the lot inspection was complete?
4. Only run the cal sphere test after a lot inspection if any of the parts fail?

My humble Answers:
1. Nope I don't
2. I can't say I'd recommend to do this everyday all day
3. Nope
4. Nope. Inspect the part with a different instrument/method and compare results and if they differ greatly investigate CMM and/or part routine


This is a deep rabbit hole.. with that being said, if no crash happens, I only calibrate my tips once every 1-2 months (half the forum cringed I bet). That is 9 probes and 84 tip angles total.

In my earlier years I spent a bunch of 'free' time proving to myself that this machine is accurate/precise/ yadda-yadda and what hurts it or in other words I got to know my machine.

My current setup is clean, environmentally controlled and in a cage where I am the only one who has access to the CMM other than a Robot loading/unloading.

If you worked with me... I would say that is extremely overkill (again, in my environment). Can't just assume it's always 'good enough' just because you calibrate. Spend some time and run some studies... what's that called? GR&R? I would like to think that the goal is to be faster than a manual inspector on the granite.

[Douglas]

half cringed I bet is right... I fully agree with double checking by manual or alternative methods when possible, this has serious value if for nothing but a sanity check at times

[Kp61dude!]

I mean after sooooo many sanity checks you eventually have to be convinced... no? Even after 2 months with no calibration and 24/7 production (though it's not constantly running) running evals confirmed my results are **** (y)!

What I would like to see is fact that says "do this, don't do this", it's been mostly speculation. No 2 Hex will give you the same answer. There are papers floating around, someone might be nice enough to post links to them.

I don't disagree with sanity checks, they'll never disappear, but what the OP is talking about is extremely time consuming. Many inspection rooms fail to have any sense of urgency (clearly the OP does have as he/she is seeking advice).

[Douglas]

oh yes it's waaaay beyond anything I do with probes too, one machine I have I can let calibrations go a month or more, unless operator bumps it maybe. My other machine is in a bit of a temperature fluctuation zone so I do more to keep up to that, maybe probes used in a program before running it if I have tight tolerances. Last time I did all probes all at once was after replacing the wrist in November or something

[JEFMAN]

It's true that it's time consuming, but it's a very nice test !
I only do something like this when using a very long probe, measuring the sphere at the start and at the end of the program (using ASSIGN/V1=QUALTOOLDATA("xyz","mysphere") in the startup alignment.
I also check once a year or so calibrating tip angles, then measure the sphere with 1 hit/position, and look at the deviations (test performed during the night, because the cmm isn't used during this time!).

For all other situations, I use autocal at the start or during the prog (but we don't work on series of parts).

And I don't forget that any collision, even if the speed is low, with a LSPX1 needs a new calibration... !

[William Johnson]

Sounds like the CMM is spending more time calibrating probes than checking parts. Was it not purchased to inspect parts or calibrate it's self. I would say it is overkill.

What I do here is we calibrate the entire rack 1 time a week on both CMM's. If there is a crash then we run a program on the qualification sphere that will verify that we are still good with the results. If not we do a full calibration (run time 2 1/2 hours) And all probes cleaned and visually checked under microscope for damage or build up 2x per shift. 2 shifts.
We have a TP200 on both machines with more tip rotations than I care to count.
We are a medical device company and have very tight tolerances and for the most part have good success with our results. When we do have issues it is usually part loaded badly or dirty probe. That is why we clean so often.
We do a 2nd cleaning after the operator brings in their part before we run them.

[NinjaBadger]

IMHO it's overkill.


BUT it depends somewhat on what you're measuring (tolerances) and what the product goes into (aerospace/medical).



This is how we roll...

12 probes, god knows how many probe-tip combinations

Every Monday we run a probe check - similar to your program but we just pick up the ref sphere with master, then check two tips (A0B0 and another tip at A90Bxx) from ever probe build.

Any probe with more than ±0.005mm in XY or Z get re-qualified. Typically if something's 'out' it's by 0.001 or less.

This means we don't spend ages calibrating probes which are fine.

Monday morning check takes about an hour, maybe two if we need to re-calibrate a probe with a lot of angles.

By virtue of the fact things are never more than 0.002 over the 0.005 (0.007 total) I think this is a fair limit for probes to disagree with one another for the tolerances we work to.

[vlmitkov]

Hmm... thats actually a fine idea. I've been pushed lately to find a way to shorten the time our machines spend calibrating.
Only problem is that we move the sphere between 4 machines. So probably in my case i would need a program where i find the sphere manually first.
Do you think that moving around the sphere will negatively affect the measuring results?
Do you think i can trust such a calibration check program if i move the sphere between the machines?
Is there a chance you could share the program file with me so i can take a look at it and see if it will work in my case?

[NinjaBadger]

No that would be fine. The idea is to check that different probes used on the same part relate correctly to each other. Therefore you can perform the test anywhere you want.

Obviously if the sphere is in a different location you may need to pick it up manually first.

DM me your email address and I'll email you the .prg.

[vlmitkov]

DM sent Thx in advance.

[baseballfan]

we have change rack only holds 9 adapter plates with styli, we have 16 styli, so we cannot do a rack calibration once a week...it would take most of a day to calibrate all tips and angles used. Best for us to calibrate tips needs for each program when running a program. Plus, we have 3 shifts now, so we no longer have time between shifts to run a rack calibration even if we wanted to.
Our RI dept. is in one room and it is controlled. We used to use a cal sphere test of only the A0B0 and A90B-90 or A90B90, but we started finding issues such that CMM results and manual inspection did not match, so we started checking all angles used and we uncovered that just because the A0B0 and A90B90 were passing, did not mean all angles were passing.
without this type of testing, we also would have never uncovered some bad probes.

In the end, it appears what we are doing is over kill and leave it to management here to decide

thanks

[Kp61dude!]

we have change rack only holds 9 adapter plates with styli, we have 16 styli, so we cannot do a rack calibration once a week...it would take most of a day to calibrate all tips and angles used. Best for us to calibrate tips needs for each program when running a program. Plus, we have 3 shifts now, so we no longer have time between shifts to run a rack calibration even if we wanted to.
Our RI dept. is in one room and it is controlled. We used to use a cal sphere test of only the A0B0 and A90B-90 or A90B90, but we started finding issues such that CMM results and manual inspection did not match, so we started checking all angles used and we uncovered that just because the A0B0 and A90B90 were passing, did not mean all angles were passing.
without this type of testing, we also would have never uncovered some bad probes.

In the end, it appears what we are doing is over kill and leave it to management here to decide

thanks

This is a valid reason to start running tests. Tests should uncover the problem, finding the problem means you can correct it and eventually stop running tests.

[baseballfan]

when the actual problem is found, you should be able to turn it on and off like a switch. However, I think what is happening is CMM operators are crashing probes / heads and not saying anything and we can't corner that problem.

Thanks again for everyone's input.

[kmcmm]

This is a deep, deep rabbit hole. The 5 basic angles will not represent all angles, and then the question becomes how much of the measuring envelope should be tested and a bunch of other questions.

A couple of our plants use what they call master parts, they're parts that the lab has kept and runs at the beginning of each shift to verify the probe calibration status. What they did to validate the master part was to calibrate all of the probes used in the program, then run the master part 10 times removing the part from the fixture before each run. Then they calculated the average for each measurement and used that for the nominal for each dimension. There's been a debate on what tolerance to apply, one suggestion is the range from the 10 trial run and another was the measurement uncertainty of the CMM itself.

Another idea is to use an "artifact" of known values that would be calibrated on an annual basis just like any other gage. The intent of this is to have a known value to compare the measurement results, in this scenario the tolerance applied would be the measurement uncertainty of the CMM. This would accomplish two things;

1. It provides data to complete a proficiency test, this would tell you if your measurements match an ISO/IEC 17025 accredited lab as well as establishing correlation between your CMMs.
2. It would provide a way to evaluate the calibration status of the probes.

Additionally if it was designed to cover the majority of the measurement envelop it may allow you to perform intermittent evaluations of the calibration status of the CMM.