Optimization of Calibration Periodicity for Temperature Instruments

[helatalbi45]

Hello everyone,

I am currently working on a project to optimize the calibration periodicity of instruments in an electronics industry. My instruments involve temperature measurements (oven, incubator, probe, etc.).

I have identified two possible methods to determine this periodicity: the drift method and the Opperet method. However, being an intern during the summer, the Opperet method seems difficult to implement without a team of experts.

I am wondering if the drift method, although often used for dimensional instruments, can be adapted for my temperature instruments. During the calibration of the oven, the laboratory installed 9 temperature sensors and recorded the minimum and maximum values. Which value should I take to apply this drift method optimally?

I would appreciate any advice on the most appropriate method and any detailed documents on calculating calibration periodicity.

Thank you for your help.

 

[AllTheThings]

I'm sure someone else here can weigh in with more knowledge on the application of methods to determine cal intervals.

But: The 9 probe method for temperature validation of an oven is NOT temporal in nature. It is an evaluation of uniformity of the conditioned environment at a given point in time. The difference/range of the 9 probes should be recorded after the system has reached equilibrium, or if it is a profile, during the critical portion(s) of the profile. For drift, you would look at a single sensor (or maybe an average of all 9) with respect to change over a much longer period of time or over multiple calibration events.

With respect to cal intervals generally, you also need to account for risk: Some entities will calibrate (or verify) things like scales on a very short interval, just because the cost of cal is nothing compared to having to recall huge amounts of shipped goods. The folks here who work in pharma probably have some related war stories...

Glad you posted on Reddit too...I will be curious to compare responses between platforms

 

[Hendi]

i think time is your main enemy and your best friend here.
You typically don't calibrate for the calibration, you do it for process control. Is there historical data to look at (literally, plot and stare at)?
Is there logging during operation? If the temperatures at all 9 sensors are roughly the same all times, process and monitoring are less questionable than when wild things are going on - and logging will allow finding out before calibration. Can you cool down the oven, open the door and put a fan in to get roughly equal temperature everywhere? Compare to another thermometer for ambient temp measurement?
Setting up a periodicity scheme with lots of "just guessing" now to me sounds not as good than creating the ability to find out which method works.