Is it common to do a repeatability check during calibration of an instrument?

[greif]

Our business has a section which does calibrations of customer ruled scales and other 2D images, and some % reflectance and % transmission photometric work.


We have a newly hired guy who has a standard cal shop background (mics, calipers, voltmeters, etc.) who says that when we calibrate our measuring equipment (we do almost all of this in house) it is not worth the time it takes to do a repeatability check (of at least 10 repeats (or 30 if they are not time consuming)). His stance is that if it meets the pass/fail specs, then all is ok.

My question for the forum members that are actively doing metrology or in a calibration lab is;

1) Is it normal practice for your lab to do routine repeatability checks on your measuring equipment? At normal calibration intervals?

2) If so, how many repeats? How often?

3) What kind of measurements do you do for the equipment that has repeat checks done (linear measurement, % reflectance, voltage, etc.)?

4) are the repeat test values recorded?

I would like as many answers as possible, as I suspect some do and some don't and I would like to get an idea of the % of each.

Thanks!

 

[Marc]

I have never seen repeatability done as part of calibration. If you have been doing this, what is your reason?

 

[greif]

Keep in mind that this is for calibration of our in house measuring equipment. I feel that looking at the repeatability of the measuring instrument and resulting sigma will give me a more sensitive way to know in advance if something is going wrong (as opposed to waiting for calibration measurements to exceed spec)

 

[dwperron]

The only time I ever saw repeatability required was for FDA calibrations. CFR 820.72 (b) says:
"Calibration. Calibration procedures shall include specific directions and limits for accuracy and precision."
The fun part with this is that most manufacturers do not specify "precision", so it can be a hassle to come up with and justify a precision tolerance.
Auditors will look to see that you do repeatability on the instrument, and that you check the entire range of each range of each function.

Also, in the 17025 accredited world you will need to produce repeatability values for your uncertainty budgets for lab standards.

I can see where photometric devices might be good candidates for repeatability results, not being in that business I don't know if there are industry standards that might address this.

 

[Marc]

Thanks for correcting me.

 

[greif]

The more I think about it, it seems that you really don't know how well a system is functioning unless you know that std. dev., range, and average are all reasonably near expected values.