We currently acquired a second cmm. They are both identical models and are used with controlled programs and fixturing. Basically, all the operator has to do is load the part and push the button.
I ran a small sample G RR on the two machines using the same part. Because my specification range is so small (.0002"), I keep getting numbers greater than 30%. Although I understand that this may be ok in some cases, I still have the following questions:

1) What type of variation would be expected between two active scanning cmm's? Assuming a controlled environment and fixturing.
2) Because you are talking of measurements in the millionths, would a correlation excercise be better?

Our measurements are very critical and they are our specifications are very tight (.0002" is fairly normal). Any suggetions on how show equivalency between the two machines at these measurment levels would be greatly appreciated.
Thanks.

We currently acquired a second cmm. They are both identical models and are used with controlled programs and fixturing. Basically, all the operator has to do is load the part and push the button.
I ran a small sample G RR on the two machines using the same part. Because my specification range is so small (.0002"), I keep getting numbers greater than 30%. Although I understand that this may be ok in some cases, I still have the following questions:

1) What type of variation would be expected between two active scanning cmm's? Assuming a controlled environment and fixturing.
2) Because you are talking of measurements in the millionths, would a correlation excercise be better?

Our measurements are very critical and they are our specifications are very tight (.0002" is fairly normal). Any suggetions on how show equivalency between the two machines at these measurment levels would be greatly appreciated.
Thanks.

You might have a calibration issue. If the new CMM wasn't calibrated after installation, it should it have been. It would probably be a good idea to have both calibrated and then run the GR&R again.

You might have a calibration issue. If the new CMM wasn't calibrated after installation, it should it have been. It would probably be a good idea to have both calibrated and then run the GR&R again.


I agree. A 30% GRR variation cannot work with yout high precision requirements.

I would study the numbers and see how much of the variation is assigned to the measurement of the same part. There should be virtually no variation in the part, or it would be calibration.

There should be little variation between operators, because the machine is doing the indexing.

That leaves fixturing as a variable. Does your part have irregularities that could cause users to fixture the part differently?

Some thoghts
One thing I would look at is the pressure setting on the touch probe. As you know speed has an influance and your programed prehit set is the same so I am wondering if the pressure settings on the reneshaw head could be the influance at the time the mesurement is taken. You also might want to increase the number of points taken for your tip qual to see if this helps reduce variation.
What is the repeatability with in a single machine for mutiple setup ands runs on the same part as opposed to that seen between the two machines for the same part?

Are you using one fixture or two? This could affect the outcome.

Are the machines in the same room with the same environmental conditions?

Does each machine pass a GRR separately with this part/fixture/program?

CMM's of all types will have a maximum permissible error value stated to different standards depending on the age of the machines, these are however guarenteed values supplied by the maker for all machines of that type. It is quite likely that any two machines could be of different accuracies within that spec. Unless your machines are of a very high specification, Mitutoyo Legex, Lietz, Ziess etc, then at 10% - 30% of your tolerance 0.00004" - 0.00012" required for a capable system; it is possible that the variation you see is within spec. Remember that even the M.P.E quoted under ISO 10360 is achieved under a very specific and idealised set of criterior, i.e. short styli. Any user defined styli configurations will cause greater errors.
Stylus condition could be another factor, even figure marks, let alone dirt can have a major impact when tolerances get tight.
I hope I am not teaching Granny to suck eggs.

I once discussed calibration practices with a CMM calibration technician, and was horrified how inadequate his attitude, understanding and appreciation was of the calibration process. I jokingly suggested to a collegue that he was at the level of an average copier repair guy.

So, calibration may play a role in the differences, even when that calibration is performed by a supposed certified technician!:nope: