Perpendicularity Strategies

[[Pa...]]

I'm having tough time getting perpendicularity measurements to read in tolerance, specifically cylinder to plane.
Modus within the machining cell shows good, but Calypso regularly reads out. Bench confirms Modus.
I'm thinking we giving calypso too much data (typically 1000s of points) and inevitably a few points fall out of the envelope.
I have spread out the cylinder and plane as far as realistically possible to get a true representation of the feature and avoid projection error.
Any ideas?
Maybe just use 4 point cylinder and planes?
Outliers and Filters are on, Evaluation is LSQ.
PaulSchuyler

[[Da...]]

So the plane is the datum? How are you doing this on the bench? If you flip datums then you will probably get different readings.

[[Ri...]]

Ditto to what Dave said.

In addition, 1000s of points is going to be much more accurate and repeatable when constructing a cylinder. It is using those to construct an axis line. A few "bad" points isn't going to throw off an LSQ axis line. This is also why we use things like Outlier Elimination.

Thanks.

[[De...]]

The feature being checked should be using a minimum feature evaluation because perpendicular is a zone tolerance.

Not sure if it will help or not but it can't hurt.

[[Ro...]]

Ive had a problem when using Helix Scan in the past for perpendicular check.
I found calypso likes 2 circles with 400º sweeps high and low, and you can put a helix 2 or 3 revolutions in between if you want.