FDA 483 Warning Letter for use of Calipers

I'm writing some procedures for measurement(general). In particular I am looking into the use of calipers. I was looking through the FDA website and warning letters and I noticed the FDA issued a 483, for the use of calipers in the following manner. A part is listed as .25+-.005. Most calipers have a stated accuracy of .001. The FDA wrote this observation because the company didn't direct it's inspectors to reject a part that had a result of .250. Is this standard practice, in the use of calipers, so that a tolerance of .245-.255, is really .246-.254 when measured with a caliper?

Re: FDA 483 for use of Calipers

jdm2008 said:

I'm writing some procedures for measurement(general). In particular I am looking into the use of calipers. I was looking through the FDA website and warning letters and I noticed the FDA issued a 483, for the use of calipers in the following manner. A part is listed as .25+-.005. Most calipers have a stated accuracy of .001. The FDA wrote this observation because the company didn't direct it's inspectors to reject a part that had a result of .250. Is this standard practice, in the use of calipers, so that a tolerance of .245-.255, is really .246-.254 when measured with a caliper?

Click to expand...

I think there might be an error in your post. You say that the size/tolerance is .25 ?.005, but the FDA says that a measurement of .250 (the nominal) should be rejected?

Re: FDA 483 for use of Calipers

jdm2008 said:

I'm writing some procedures for measurement(general). In particular I am looking into the use of calipers. I was looking through the FDA website and warning letters and I noticed the FDA issued a 483, for the use of calipers in the following manner. A part is listed as .25+-.005. Most calipers have a stated accuracy of .001. The FDA wrote this observation because the company didn't direct it's inspectors to reject a part that had a result of .250. Is this standard practice, in the use of calipers, so that a tolerance of .245-.255, is really .246-.254 when measured with a caliper?

Click to expand...

You lost me here....

First, 0.250 is in spec, why would I reject it?
Second, if the MSA on the caliper is acceptable, you do not change the specification, but rather decide to incorporate measurement error in setting your specifications.

Re: FDA 483 for use of Calipers

Yes my error. The FDA says that .255 is out of spec(or should be).

"According to your firm's management, your calipers read out to 1/10000 of an inch(ie four decimal places). Suppose the measurement reads .2550 inches and so it is deemed to be with your design specification. However due to the precision limit of your firm's caliper, the actual flange height may be in fact be .2560 inches. Your firm's most recent revision of Standard Operating Procedure Titled Nonconformance/Deviation Report defines a non conformance as "the failure of a product, process, or Device History documentation to meet specified requirements. In this situation your firm would incorrectly allow the (redacted) to pass this particular check, as opposed to deeming the part to be a nonconformance per SOP(redacted)."

Re: FDA 483 for use of Calipers

Well, the measurement reliability is getting out on the hairy edge there. I don't know the answer to the question but I'm going to pose one that's related.

Back in the sticks, stones and bones days of precision measurement when I was helping to build the pyramids, the philosophy went that since the common man's dial caliper only discriminated down to .001 inch that your estimate of reliability had to take that into account. And if you take user skill into effect, then your measurement is even less likely to be correct when you have unskilled users making measurements.

Then the digital caliper came along, discriminating down to .0005 inch. Did our measurements suddenly become that much more precise when we were still using the same frame with a different readout device?

It's always been my approach that the caliper is the pair of pliers of precision measurement - good for lots of approximations under ideal conditions, but if you get out on the edge and the result is important then it's time to go get a tool better suited to the task, use the appropriate technique and make certain that the person performing the measurement is familiar with the area(s) of uncertainty and understands the reasons why.

Apologies if this neither answered the question or shed additional confusion on the issue.

Re: FDA 483 for use of Calipers

jdm2008 said:

Yes my error. The FDA says that .255 is out of spec(or should be).

"According to your firm's management, your calipers read out to 1/10000 of an inch(ie four decimal places). Suppose the measurement reads .2550 inches and so it is deemed to be with your design specification. However due to the precision limit of your firm's caliper, the actual flange height may be in fact be .2560 inches. Your firm's most recent revision of Standard Operating Procedure Titled Nonconformance/Deviation Report defines a non conformance as "the failure of a product, process, or Device History documentation to meet specified requirements. In this situation your firm would incorrectly allow the (redacted) to pass this particular check, as opposed to deeming the part to be a nonconformance per SOP(redacted)."

Click to expand...

I have no experience in dealing with the FDA, but I'm having trouble following the logic in that quote. The letter as quoted says that the calipers "read out to 1/10000 of an inch..." Using an example of a thing being measured at .2550" (the upper spec limit), it goes on to say, "...due to the precision limit of your firm's caliper, the actual flange height may in fact be .2560 inches."

I don't see where that conclusion has any rational antecedent. While it's true that a measurement result of .2550 could be, say, .2551 and technically out of tolerance, I don't know how the .2560 figure can be justified without actual evidence. Everyone who has responsibility for measuring things should be aware that when results get close to specification limits the result should be confirmed either with a different device or the same type of device. This is, of course, where measurment uncertainty comes into play, and if the measurement results are likely to be affected by uncertainty, more precise means of measurement are called for.

The caliper reads out 4 decimal places but is only accurate to .001. If you try and buy a caliper accuracy is only guaranteed to .001, and calibration is deemed pass if it is accurate to .001.

https://www.mcmaster.com/#electronic-calipers/=sqiroj
If you look at the listings, you notice these calipers resolve to 4 decimal places. However the accuracy is listed as .001. so if the caliper is only accurate to .001, that is how they got .256(worst case if you get .255, and the caliper could off by .001, .255 is out of spec).
I am certain this is not an error(by error, i mean miscalculation, typo etc) but my question is, is this standard practice in fda regulated industry, that the tolerance window must be reduced by the accuracy of the device in order to be certain you are getting a in spec result for passing units.
So .250+-.005, is .246-.254 if you are using a caliper, .2451-.2499 is you are using a micrometer etc, etc.

jdm2008 said:

The caliper reads out 4 decimal places but is only accurate to .001. If you try and buy a caliper accuracy is only guaranteed to .001, and calibration is deemed pass if it is accurate to .001.

Click to expand...

I don't think it's that cut-and-dry, is it? Many instruments are only guarenteed to a nominal value, yet individually can be much better in practice. I don't think that's why the FDA said what they did.

One certainly could guardband the spec to insure the caliper is not used to establish a part value at the spec limit.