Calibration is the process of establishing a KNOWN RELATIONSHIP - Traceability
Subject: Re: Calibration definitions/.../Gazley/Pfrang/Gazley/Pfrang
Date: Mon, 11 Jan 1999 10:01:37 -0600
From: ISO Standards Discussion
From: (Doug Pfrang)
Subject: Re: Calibration definitions/../Gazley/Pfrang/Gazley/Pfrang
>From: "John Gazley"
> My comments were not directed to a piece of equipment in which no
> known standard exists
> In the case of a gage where no known standard exists to calibrate it
> yes, you are right.
> But in the case of something like calipers you must be traceable to a
> known standard. There is a known standard which exists for this gage-
> gage blocks traceable to international or national standards.
> If for example you choose to calibrate these with something like a
> piece of bar stock that you control in-house, you must at some time
> calibrate this bar stock to an international or national standard. If
> after this you wish to use this for your calibrations you can. You
> must in some way show traceability to a recognized standard (unless of
> course no known standard exists, then you are free to improvise).
> ISO is specific in the case where there is a known standard, you must
> calibrate against certifified equipment having a known valid
> relationship to internationally or nationally recognized standard.
> Unlike many aspects of ISO9000 it doesn't get any clearer or more
> specific than this.
Just because you measure a parameter for which an international reference standard exists, does not mean you must have traceability to that reference standard to calibrate a device. It depends on how you use the device. Let's look at a few examples:
1. Let's say I have a prototype of a product I wish to produce. I take my calipers, measure each component of the prototype, and write down all the dimensions of all the components. To make my example easy, let's say I measure a particular part and my calipers tells me it is 5.2 inches long. Then I take that same calipers and I use it to set up my machines to make production parts. I make a batch of production parts, and measure them with my calipers. Each time I measure my production parts, my calipers tells me they are 5.2 inches long. Do I know if these parts meet specification? Of course I do, because they gave the same reading on my calipers, and therefore I know that my production parts are the same length as my prototype part. Do I care if my parts really are 5.2 inches long? No, I don't. All I care is that my parts match my prototype. If I use my calipers in this way, it does not matter if the calipers are calibrated using an international reference standard. In fact, my calipers could be way out of tolerance and they would still work fine in this example. As long as I use the same calipers to measure my production parts as I used to establish my specification, then I don't care if 5.2 inches on my calipers really is 5.2 inches according to some international reference standard; I have "calibrated" my calipers using my prototype as my controlled reference standard.
2. Now, let's extend the example. Let's say that instead of making the parts myself, I call my supplier and tell him that I want a particular part to be 5.2 inches long. He sends me a box of parts, which I measure with my calipers. My calipers tell me the parts are 5.2 inches long. No harm, no foul. We have basically the same situation as above: the parts meet specifications. I don't know if my calipers match some international reference standard, and I don't know if my supplier's calipers match some international reference standard. I do know that my calipers match my supplier's calipers, and that's all I care about.
3. Now, let's say that I get my parts back from my supplier, and my calipers tell me they are 4.2 inches long. Now I have a problem. The calipers I am using really is measuring differently from the calipers my supplier is using. Do I need to use an international reference standard now? Not necessarily.
a. Let's say my supplier is located across the street from my facility. I take my calipers over to my supplier and I explain my problem. He brings out his calipers and we compare them. Sure enough, his always reads one inch longer than mine. (I.e., if my calipers says something is 5.2 inches long, then his says it is 1.0 inch longer -- 6.2 inches long. Thus, when I ask for a 5.2 inch part according to my calipers, he makes the part one inch shorter, because that is what corresponds to 5.2 inches on his calipers. Unfortunately, my calipers measures this part to be 4.2 inches, which is out of specification.) Is our solution to calibrate both calipers using an international reference standard? Not necessarily. Since I know that my calipers reads one inch shorter than his calipers, we agree to use his calipers as our reference standard. From now on, whenever I order parts from him using my calipers, I will add one inch to my specification. That way, we have compensated for the error between his calipers and mine and the readings on our respective calipers will match, even though neither of our calipers is calibrated to an international reference standard.
b. Now, let's say my supplier is located in Taiwan. It is too much of a hassle to ship my calipers to him to compare the two calipers directly, so I decide to calibrate my calipers instead using a reference standard that enables us to compare our calipers indirectly. This is when I use the international reference standard to perform my calibration. By using an international reference standard to perform my calibration, I am using a reference standard that will be the same for my supplier. Assuming he also calibrates his calipers using the same reference standard, then both our calipers will read the same.
If you analyze each of the above situations, you will observe that the true definition of calibration is simply this: calibration is the process of establishing a KNOWN RELATIONSHIP between (a) the readings on the device which was used to create your specification, and (b) the readings on the device which is used to test conformance to the specification. As long as there is a _known relationship_ between these two readings, the device you use in (b) will be calibrated -- whether or not an international reference standard was used (as an intermediary) to help establish the known relationship. In example 1, the _known relationship_ was obtained because I used the same calipers to create the specification as I used to test conformance to the specification. In example 2, the _known relationship_ was obtained because I again used the same calipers to create the specification as I used to test conformance to the specification, even though I did not make the parts myself. In example 3.a, the _known relationship_ was obtained because I compared my calipers directly to my supplier's calipers. Finally, in example 3.b., the _known relationship_ was obtained using an international reference standard, because I had no other way to establish a known relationship through more direct methods. In each of the examples, my calipers were calibrated; only the situation in example 3.b. required me to use an international reference standard to get there.
And this analysis works for ALL measurement devices, no matter what they are and what the measurement accuracy needs to be.
-- Doug Pfrang