CMM Equipment Validation and Qualification Approach

[PeterLe]

Hello everyone!

I just got a new Quality Engineering position and I have been asked to do an Equipment Qualification on a CMM (http://www.ogpnet.com/ogpVidQuest300.jsp). I'm pretty familiar with process validation, however I don't have any experience doing an Equipment Qualification on a measurement system. I believe equipment qualification is an IQ/OQ.

Now, I'm a bit unsure on how to conduct the OQ portion of the equipment qualification. I've read you can do a gauge R&R, DOE, etc. But I'm wondering what is the general approach for conducting an equipment qualification for a CMM? Thank you so much in advance!

 

[Stijloor]

A Quick Bump!

Can you help a fellow Cover??

Thank you very much!!

 

[PeterLe]

Thanks Stijloor!

After doing some research, what do you guys think of this general approach?

IQ: We can reference the calibration and preventive maintenance documentation.

OQ: Take a lot (traceable) of known parts with known dimensions (using current inspection plan) and have one operator measure key dimensions. Compare the CMM to the inspection plan results by doing a t-test. Any idea what the correct sample size would be?

PQ: I'm not sure if this is necessary as we are qualifying the equipment and not validating. Please let me know your thoughts. But if we were to do a PQ, would this be a gauge R&R study? 3x10x10?

Any help would be much appreciated! Thank you in advance!

 

[yodon]

To me, that would not put you in a very defensible position.

IQ is typically (much) more than just PM and calibration. You should consider overall installation (is it set up properly according to the manufacturer's instruction, etc.). Consider the software running on the machine - is it up-to-date, are there any known issues that might affect your operations? I presume there are custom 'programs' for individual parts - installation should show that these are under control, etc.

For OQ, you can either test specific pieces or do a more general test (range of capability). If you go with individual pieces, you would need to repeat OQ and PQ each time you added a new part. If you went with a general OQ approach, confirming the entire range of expected pieces, then you don't have to do additional testing when adding a new piece (presuming it's within the range tested). OQ should also show that bad pieces are identified. I don't know if there are any safety measures, but if so, those would also be verified.

Sample size - whether for OQ testing or PQ testing needs to be statistically defensible (outside my area).

For PQ, since there's not a "production run" you may want to just demonstrate some capability for changing between parts to measure (has no impact). The decisions for what you would should be documented in a Validation Plan.

 

[PeterLe]

To me, that would not put you in a very defensible position.

IQ is typically (much) more than just PM and calibration. You should consider overall installation (is it set up properly according to the manufacturer's instruction, etc.). Consider the software running on the machine - is it up-to-date, are there any known issues that might affect your operations? I presume there are custom 'programs' for individual parts - installation should show that these are under control, etc.

For OQ, you can either test specific pieces or do a more general test (range of capability). If you go with individual pieces, you would need to repeat OQ and PQ each time you added a new part. If you went with a general OQ approach, confirming the entire range of expected pieces, then you don't have to do additional testing when adding a new piece (presuming it's within the range tested). OQ should also show that bad pieces are identified. I don't know if there are any safety measures, but if so, those would also be verified.

Sample size - whether for OQ testing or PQ testing needs to be statistically defensible (outside my area).

For PQ, since there's not a "production run" you may want to just demonstrate some capability for changing between parts to measure (has no impact). The decisions for what you would should be documented in a Validation Plan.

Hello yodon! Thank you so much for your reply!

After doing research, the past few days, I agree with you the IQ needs to be much more stringent.

As far as the OQ goes, I still need to find out what management wants: Whether they just want the equipment qualified (confirming the specification of the manufacturer) or validate a specific routine with a specific part. The latter would require a gauge r&r. Then of course if management were to go with the gauge r&r route, then yes we would definitely have to conduct a validation if the company wants to use other routines/parts.

 

[PeterLe]

Hello everyone!

So after some research, I'm still not sure how to approach the OQ. What I would like to do is write some test scripts for the CMM. For example when we go to the software and want to measure the diameter, I want to verify that when I push the diameter button on the software, that it will indeed measure the diameter. So in that case, I will need to get a piece with a known diameter and then measure it using the diameter feature on the software to see if it does measure the diameter. Does this make sense? Is this the right approach for the OQ?

Then I'd also have to write test scripts for the other features such as line, radius, length, etc?

Again I want to just qualify the equipment and not conduct a PQ. The PQ will be saved later for specific routines that we will use.

 

[yodon]

OQ should be based on your requirements on how you use the system. You would need to define, for example, the types of measurements done (as you have already alluded to), but more importantly, the accuracy.

Suggest you try to establish a range for the measurements or items to be measured and then the OQ can confirm operations throughout the range. This way, when you add a new part to be measured, you don't need to re-run any validation, it will already be covered by the existing validation (as long as it's within the range of the validation conducted).

You may have other requirements like speed, reporting, security, etc. These should be considered.

Hope that helps.

 

[danile1]

Hello PeterL, can i ask you how did you performed the validation of the CMM after all?
I am in the same position rightnow, i need to validate the company's CMM and have no idea where to start. Any information will be appreciated.

 

[Eredhel]

We have two CMMs that get calibrated once per year. They are done by the manufacturer, a tech comes down for a couple days per machine. There are also independent places out there that can do calibrations. Is your company willing to pay to farm that out? I can't imagine trying to do it in house.

 

[Chrisx]

I believe the calibration adequately validates that the device will provide accurate/precise measurements within allowable tolerances. I haven't validated a CMM, but I have validated vision systems. My take on it was that a software validation was probably more suitable. I ran test parts and went thru each measurement function in the software. I ran test cases to demonstrate it would generate a failure for out of specification results and correctly pass parts that were in specification. A training part can be very valuable for this exercise. I believe our CMM came with such a part to use for learning each measurement function.