I am looking for advice from people who calibrate gage blocks and feel they do a good job (in terms of being realistic with uncertainty, capabilities and what you tell your customers).

I recently sent out a grade 2 set of gage blocks for calibration (as I have each year for the past 5 or 6 years).

The report came back showing measuriment results and the associated uncertainty (which is good since I sent it to a lab accreidted by A2LA to ISO 17025).

For a 1 inch gage block, my block had a reading of +1 uin. The labs uncertainty was (3.4 + 2L)uin which is 5.4uin at 1 in.

No matter which spec the vendor used (his cert was not very specific which is another issue), the uncertainty makes it possible that the block would be out of tolerance. This made me go look at a bunch of other uncertanties for gage blocks on other accreditation scopes and I found the same issue. Most uncertainties for commercial calibration labs are larger that the gage block tolerances.

How do these labs determine in tolerance or out of tolerance conditions (or do they)?

Should they all just tell their customers up front to ignore the grade and look at their uncertainty because the report has no real possibility of saying that they meet an actual grade? The customer can then decide what they need.

Should they then tell their customers that they will only reject a block if the reading is so far off that even the uncertainty does not make it possible for it to be in tolerance?

Any thoughts would be appreciated.

Thank you!

I would question the uncertainty you got.....1 microinch is achievable if everything is perfect.....but not usually for a grade 2 block.....

Tolerance is a bit different than uncertainty as a rule, though for gage blocks there is very little practical difference.

I would send them to somewhere like Mitutoyo or Starrett to have them recalibrated.

Hope that helps.

Hershal

How do these labs determine in tolerance or out of tolerance conditions (or do they)?

Should they all just tell their customers up front to ignore the grade and look at their uncertainty because the report has no real possibility of saying that they meet an actual grade? The customer can then decide what they need.

Should they then tell their customers that they will only reject a block if the reading is so far off that even the uncertainty does not make it possible for it to be in tolerance?

Any thoughts would be appreciated.

Thank you!

Joe,

For labs accredited by A2LA, at least, there is a requirement set in their policies. Their Traceability Policy states: "Measurement uncertainty shall be taken into account when statements of compliance with specifications are made". Also, their Calibration Program Requirements state:"... when parameters are certified to be within specified tolerance, the associated uncertainty of the measurement result is properly taken into account with respect to the tolerance by a documented procedure or policy established and implemented by the laboratory that defines the decision rules used by the laboratory for declaring in or out of tolerance conditions."

Therefore, the lab must have this 'decision rule' to note how the lab deals with this exact situation. Some labs may choose to only "fail" the block if it exceeds the limit of the uncertainty band (or tolerance band, whichever is greater). Or, a lab may choose to simply report the uncertainty and not consider it when reporting pass or fail on a block. I don't know how many people look at the uncertainty on calibration certificates, but it's well worth doing. You should make sure you know if your contracted lab is considering the uncertainty or not when stating the verdict. You also have the option of seeking a new vendor for gage blocks and pay close attention to the best uncertainty stated on their scope of accreditation. (Don't make that your only criterion, though.) A2LA has got a very user friendly search system on their site to look for labs. You'd have to pull up each scope individually and search for the lowest gage block uncertainty.


I have to say, it is difficult with gage blocks when the tolerances are so tight and the various uncertainty contributors have got to add up to a few millionths even with a stable environment, etc.

Good Luck

Hi there!

I am a quality manager at a calibration lab, but I started out measuring gage blocks in the back. I have since addressed the same issue you have with some of our customers.

Our gage block Measurement Uncertainty (MU) is 3.5 microinches less than 1 inch. I will spare you the details of where this number comes from.

Our policy is that we report the value the LabMaster reads. If this number is within the tolerance limit, the unit is in, and vice versa. We cannot report a number to a customer other than that which our machine reads. And we cannot say something is in tolerance if it reads out!

Therefore, we leave it to the customer's discretion to intrepret our stated MU and decide if they want to replace OOT blocks or downgrade the set. We do not issue a calibration certificate if the measurement is outside the given limits. And, personally, I do not believe any reputable lab should.

I think of MU as a "measure" of our uncertainty, not the uncertainty of our measurements. There is a slight but crucial difference. The numerical reading, tolerances, grades etc. are all tangible things that can be thought of as qualities of the unit. But, MU is a semi-intangible idea that DESCRIBES our specific measurement process. It should not be added on top of a tolerance for a certain grade to give a larger tolerance!

The thing to keep in mind is that we are dealing with microinches. The unit is so small I can hardle comprehend it! Therefore, we try not to have customer's get all hung up on grades and if a unit is OOT by a few divisions. Instead, we encourage customers to downgrade to a "workshop" grade with tolerance +/-50 millionths. Or to simply take a Report of Value and throw the grades out the window!

The important thing to take away is that a cal lab should NEVER report a value other than that which they read. And it is untruthful to say a unit is intolerance when it reads OOT. MU is a "description" of the process, not a property of the unit.

Hope this helps!

Michelle