roundness affected but not diameter

[[An...]]

Using Calypso 2023.  We measure various bores and outside surfaces for diameter as well as roundness.  Recently we had a part where the roundness didn't agree with a manual measurement(difference of about 0.004").  The manual measurement was verified by myself as well as a supervisor.  When measured again on the CMM it was almost exactly what the manual measurement was.  The diameter measurement was exactly the same both times.  I ran a roundness check on the probes using a ring gauge and the results were fine.  I checked the probes for damage and nothing I could find.  I checked to make sure that none of the probes were loose and they weren't.  I'm using LSQ for an evaluation method on the circles, but when the graph for roundness is pulled up the evaluation method is Minimum Feature.  Two questions;

1)What could be the cause of the difference in roundness measurements that doesn't affect the diameter measurements

2)Is there a way to change the evaluation method on roundness measurement

[[Du...]]

LSQ is just an average between the high and low, if you were to switch your diameter evaluation to outer tangential and then inner tangential, you'd be able to see the difference. you could also create a 2 point diameter characteristic that would show you where the high and low points were on the part.

[[No...]]

Thought experiment:

Imagine you have a perfectly round circle. You measure its diameter with LSQ and get exactly the value you expected.

Now imagine you squeeze the circle into a symmetric ellipse, where the width increases by the same amount the height decreases. If you measure this ellipse as an LSQ circle again, what diameter would you get? Answer: Exactly the same as before! If you squeeze the ellipse even further, maintaining the width/height ratio, the result stays the same. This is the effect of the LSQ algorithm. If your circle is evenly deformed, you might not notice any difference in diameter.