Definition Calibration vs. Verification - Definition of

[Joe Cruse]

Shannon,

I'd fight that. For your in-house verification/calibration status check activities, you can use whatever you deem necessary for your system. For CALIBRATION purposes, I'd agree that traceable standards are needed, but the standard does not paint you into a corner as far as verification/calibration status check activities are concerned (assuming YOU didn't paint yourself into that corner by stating you would always use traceable standards for verification in your own QMS).

In our lab, we do happen to use traceable weight standards to verify calibration status on our scale/balance sets. Out in parts of the plant that either smelt or package finished product, we don't though. One thing we have taken to doing is this: when our 3rd party calibration vendor comes in to calibrate scales, when they've finished the job, we put our in-house verification weights on the scale and note the weights for our system. We always do this when introducing a new test weight into the system, and use that recorded weight as the "true" value to verify against. Our registrar has even suggested we do this with our lab test weights, and save ourselves some $$ on having a calibration lab certify our test weights every year.

 

[Graeme]

SPC using a check standard with regular comparison to a traceable mass standard

Shannon,

Are you recording the values from your verification checks? If you are, and if your scale has reasonably good sensitivity and resolution, you might be able to turn the verification into an actual SPC program for the scales and without having your weight calibrated. I am making two assumptions here -- that you do one verification per day, and that your scale calibration vendor has 50 lb. as one of the test points.

• Record the readings from your daily checks.
• Decide on a rational subgroup - if you work Monday-Friday then five days should be about right.
• Plot the data on an X-bar/R control chart in subgroups of 5 (or whatever you decide, but not less than 3).
• When your scale is calibrated, compare the 50 lb calibration data to your chart. (You may want to actually plot it separately on the control chart.)
• As an additional check, plot the 50 lb calibration data over time on a trend chart.

There you have a possible solution -- an in-house SPC program for the scale with regular calibration and direct comparison to a traceable weight!

This should work even with some variations in the plan, and it can be extended to additional scales. It is important, thought, that each scale has its own chart even if the same weight is used for all of them. (Each scale is a different process.) Another refinement would be to measure your weight a number of times (3 to 5?) immediately before the scale is calibrated and again immediately after the calibration. The averages will give you a way to compare the mass of your weight to the one used for calibration, and the standard deviations can give you an indication of the repeatability.

With this type of plan, the actual value of the SPC check standard (a new name for your verification weight?) should be at a calibrated point on the scale, but that does not matter as much as its stability and repeatability. The traceability of measurements made with the scale is taken care of by the calibrations. Regular use of the check standard with recording and charting the data measures the behaviour of the process (the scale) in between calibrations. Just be sure your check standard weight is "controlled" -- kept clean and protected from damage, and only used for the scale verifications.