Howdy, experts:

Here's another question for you: Is it recommended to perform a calibration on a gage as part of the preparation for a Gage R&R study? In general, what sorts of pre-requisites, if any, should the gages fulfill before undergoing Gage R&R?

Talk to you soon,
Craig

Howdy, experts:

Here's another question for you: Is it recommended to perform a calibration on a gage as part of the preparation for a Gage R&R study? In general, what sorts of pre-requisites, if any, should the gages fulfill before undergoing Gage R&R?

Talk to you soon,
Craig

Two schools of thought: the first says that you should always calibrate first, because it provides a solid baseline for the R&R study. On the other hand, if the gage is in the calibration system and calibration status is current, there are those who would say that it's not necessary. I think it depends a lot on what it takes to calibrate the thing, and how much time is available, and the criticality of the measurements...
As a rule of thumb, though, there should be at least a cursory check of the gage against the appropriate standard(s) before proceeding, if not a "formal" calibration.

Howdy, experts:

Here's another question for you: Is it recommended to perform a calibration on a gage as part of the preparation for a Gage R&R study? In general, what sorts of pre-requisites, if any, should the gages fulfill before undergoing Gage R&R?

Talk to you soon,
Craig

Craig:

I agree with previous post about the gauge -- if it is already calibrated and serviceable, and the history of calibration results are good, then move forward.

As for pre-requisites: the MSA from AIAG provides a nice set of guidelines for those first performing gage studies. It becomes dependent on the type of gauge and the product that will be measured. If you need in-depth information, I will pull my MSA copy and write a short novel.

--QG

I agree with JSW05. It is not always necessary to calibrate the instruments prior to conducting a GRR study, but it would be nice to know the calibration status. No point in conducting a GRR study today if you are going to send it for calibration tomorrow.

As QG says, MSA manual is a good guide, another one is " Concepts for R&R Studies" by Larry Barrentine (ASQ Press).

Generally:
- Ensure that you select parts that truly represent the process variation (if not the tolerance).
- Ensure that all operators are equally proficient.
- Ensure that a procedure is developed for taking the measurement, and that all operators understand and follow it correctly. (The procedure may require that the gage be calibrated / reset etc prior to each trial. In that case calibration "effect" will be combined with repeatability.)
- Carefully control the location of the measurement so as to reduce within-part variation.
- Randomize the measurement sequence.
- Observe the study and take notes where necessary.

Generally:
- Ensure that you select parts that truly represent the process variation (if not the tolerance).
- Ensure that all operators are equally proficient.
- Ensure that a procedure is developed for taking the measurement, and that all operators understand and follow it correctly. (The procedure may require that the gage be calibrated / reset etc prior to each trial. In that case calibration "effect" will be combined with repeatability.)
- Carefully control the location of the measurement so as to reduce within-part variation.
- Randomize the measurement sequence.
- Observe the study and take notes where necessary.

I agree with all except the second and fourth points, and those might be correct, depending on the circumstances. In general, part of the GR&R exercise is finding out whether operators are equally proficient--you can't know this unless you've proven it experimentally already, in other words, and GR&R is a controlled experiment. Also, part of the process involves determining how measurement location affects results. Again, GR&R is an experiment that will help to establish production measurement requirements, and if, during the experiment, the measurement location is controlled, it must be assumed afterwards that measurement location will have to be controlled in production, which might not be necessary. I think it's best in most cases to use nominally competent operators and just let them measure the parts.

JSW05, you may have a point there. What I'm saying is operators should have equal competecy level. Else, it will introduce a source of unwanted variation in the study. In any case, if you 'Know' that one is less competent than the other (or new or untrained) , you probably do not need a GRR study- unless you specifically want to compare operators. (This can be seen on the control charts.) Also if you have a large number of operators, you must select 2 or 3 randomly.

Similarly, if you know or suspect that there will be significant within-part variation, you can use the method prescribed for that.

So I will add an item 0 to the list:
Decide upon the objectives of the study.

So I will add an item 0 to the list:
Decide upon the objectives of the study.
Yes. Because A GR&R analysis is, as I said earlier, a controlled experiment, and experiements must begin from a hypothesis, and the hypothesis will be defined by the objectives of the study.