I don't understand, how can I validate a calibration method by a calibration.
Actually, you pose a very good question. The statement in the standard does appear a bit circular in logic.
This is just kind of my take on this. I would look at why this section is here.
Say you have a request to calibrate a new instrument. You write up a procedure, and start calibrating them. Strangely, you begin to have a high # of failures in the as-found category. Too, the customers begin complaining their instrument does not seem to operate correctly. Where is the problem?
You question the techs, you at the standard, etc. to try to find the issue. Ironically, after spending a lot of time and effort, you realize the procedure was wrong. There were a couple of steps left out that really made a difference.
The intent of the validation requirement quoted is to validate the process.
- Does it yield the results it should?
- Have you identified all the variables involved?
- Does workstation A yield similar results as workstation B?
- Do days of the week affect the results?
Perform the calibration (with all the data, of course) then send it to another lab (maybe a competitor). Are the results fairly similar? How do their uncertainty compare to yours?
I feel I am rambling
, so I will stop.
Just remember you need to verify the entire process (and document the verification) prior to considering it acceptable for performing calibrations. Once it is verified, all involved will have more confidence in it.