Measuring arc segments on manual CMM.

I have part that we are working now that looks like a 40 degree segment of a wheel. kind of like a banana shape. We are measuring the rads with a mitutoy Manual CMM with geopack and mcosmos. We cannot accurately measure the radius'. The CMM is giving us numbers that are .005-.010 big or small. And not repeating. The Radius' have +-.003 tolerance. I am not to familiar with the CMM and its software as it is new to us. What is the proper way to check these? The parts are good because they have been inspected by the customer on their CMM. Talking with someone there, they said there is a specific way to check arc segments? Any one care to chime in?

I had to deal with a similar situation at my company and we never found an accurate solution with our manual CMM. Our measurements were as you described. I would suggest trying the following:
1. Make sure your measurement is centered, normal, parallel, etc. Figure out the ideal probe angles and approaches. Minimize all the potential for measurement error.
2. Use your fine adjustment knobs to manually take as many evenly spaced points as you reasonable can. To simplify things we moved each axis .010 and took a normal measurement as best we could.
3. Do you have a calibrated ring gage handy? Near the diameter you are trying to measure? Experiment with that. Take measurements around all 360 degrees, then 180, 90, 45, 40, etc. If you can find a technique that accurately measures a known diameter, you can be more confident in what you're doing.

At any rate, we found we needed 90 degrees in order to get a reasonable amount of error for our tolerance, but as to your original question, I never was able to get it to work perfectly. You could try different measurements like Arc or Point Cloud. If you want to avoid the CMM try making Go and No-Go gaging, or an optical comparitor.

Thanks for the response. I've been reading about the limitations of manual cmm's, and this seems to be an example.

when measuring an arc where you don't have enough of the arc to measure, you need to measure from the theoretical center point to the arc surface.

If you beep along the arc and then ask the cmm to figure it out, the data points are too close together and the results are bullshit, as you have found. You need to measure it the other way around. Create a point at the theoretical center point of the arc, then measure from that point to points along the arc. In theory, all the measurements should be the same, so you may need to jog the center point around a little to get it to work.

If your arc is in the X-Y plane, are you locking the Z? Theoretically it shouldn't matter, but you never know.

we have gotten better numbers after some digging on the web. The part is a 40 deg ring. We measured the to ends and projected them to a point and aligned te coordinate system to the theoretical center line. And yes we lock the z for XY plane.