1.0000 - 0.9998 CMM Calibrated to .0002

Hello All,

I'm working at a smaller aerospace company on contract, and they are qualifying parts on a CMM. To me it seems to be within the tolerance zone of the machine. Is this a grey area or is it like I have been trained at other company's not a qualified measurement.

1.0000 - 0.9998 CMM Calibrated to .0002

From my previous experience it measure within two tenths 1.0000-0.9998 I would need a CMM calibrated to .00001 not .0002. Are many CCM's calibrated to this level or is this just a grey area where many measurements to .0000 are sometimes done within this range. How expensive is it to obtain a CMM that measures in .00001 millionths?

Thanks for your thoughts.

Just stumbled on this board today. Very good content keep up the good postings everyone.

 

Many aerospace companies use a simple minimum 4:1 rule regarding measurement device accuracy relative to the specification tolerance. For example, if the spec tolerance was +/- .0001 the measurement device accuracy required would be +/- .000025

 

I work in the automotive field and we have to have our CMM calibrated through an accredited source per IATF standards. The calibration company that we go through uses tolerances for Linear Accuracy in X, Y, and Z-Axis to +/- .000100", Squareness X-Y, Y-Z, and Z-X to +/- .000050" and the Repeatability for all Axis is .000000". I assume aerospace would be more meticulous. But remember, it also depends on the customer's requirements. Oh and make sure the granite surface plate is calibrated.

 

Hi David,

As usual in metrology, the answer to your question is not straightforward.

I am assuming that you are working in inches, so you are validating a feature of size between 1.0000" and 0.9998" therefore the tolerance zone is 0.0002"....correct?

There are many approaches to assessing suitability of the measurement process, the easiest I find first of all requires an understanding of accuracy, repeatability and the role of calibration in the first instance.

Calibration is a process in which an instruments accuracy is compared against a recognised standard. It doesn't necessarily infer that once the comparison has been made that its acceptable, this is particularly true of CMM's because any adjustments towards a spec are time consuming, can be complex and therefore expensive.

Accuracy is the closeness in the agreement between the result of a measurement and the true value of the thing being measured.

Repeatability is largely self explanatory, the result of repeated measurements on the same thing.

For your application, the most pragmatic thing to do would be to run a study of an object of approximately the same size and form with a traceable measured size....slip, plug or ring gauge for example. Compute the standard deviation, multiply this by 0.675 to arrive at the Probable Error. The specified tolerance of 0.0002" requires a Probable Error (PE) of 0.00004" or 0.00003" to be able to classify parts (5 to 6 times the probable error- select your confidence)

The upshot of all this is that before you even think about accuracy (if you know this you can ignore it or compensate your results) you have to have a machine that repeats at the level of PE I have outlined. In my experience, this level of performance is very tight, even for machines which are housed in standards laboratories, you need a robust cleanliness regime for part and probe, dirt and grease from your skin would contribute more than this tolerance and you need a good degree of temperature control.

It may be possible to achieve the tolerance using a comparative method against a high quality gauge of the same form and size but its still tight.

For more information on the PE please see the attached paper.

Hope this is helpful