Stylus radius correction settings in curve measurements

[[Th...]]

This is a follow-on to what I was asking about the last time I made a thread. I've got a length of non-standard geometry that I'm measuring as a 2D curve. Only certain portions of it are critical, so I've got the CMM measuring the whole thing once to provide visual reference data, and measuring a copy of the same curve with all the non-critical bits masked out so as to evaluate the profile only to those critical segments.

The problem I'm encountering now is that a portion of this scan is measuring out of tolerance by something like 1.5mm despite being close to nominal on the reference scan. A small arc about 1.25mm in length and 1.27mm in diameter precedes this in my scanned features, which somehow seems to be causing this to happen. Removing this arc from the scan brings my other segment back into tolerance, but the arc is also a critical feature, so I need to keep it.

Ultimately, I discovered changing the stylus radius correction from nominal spline to measured points fixed whatever was happening to my scan, but I don't actually understand why. Can someone shed some light on the effects and pros/cons of altering this setting? I checked the user manuals for the base software and Curve function, but I didn't find anything covering this setting.

[[De...]]

I will preface this by stating that this is how I think it works, I am fairly sure i'm correct but i'm not 100% on this.

Essentially if using measured points, the stylus radius is corrected in the direction of the actual measured points, if it is using nominal points, it is corrected based on the 'nominal' location of the points.

[[Th...]]

Are there any particular caveats to using measured points that I would need to be aware of? I don't have a choice for this particular measurement, but knowing what its use cases are can only be beneficial.

[[De...]]

Well, I think it probably needs to be considered on a case by case basis with the answer dependent on what you are trying to get from the measurement and what you are measuring.

As the image shows, depending on the variation between the nominal values and the measured values and the geometry being measured, the difference in stylus radius correction direction could be extreme.

I think in your case, the variation in the location of the radius you were talking about probably means the difference between the stylus correction vector being based off of the nominal curve versus being based off the actual measured points which did not lie in the radius.

Keep in mind my sketch is showing lines, but I am fairly sure internally Calypso is considering these things as individual points with nominal vectors assigned. In order for Calypso to get a nominal vector for the 'measured points' I am making the assumption they are using a spline function of some type to determine the nominal vector based on the points that are measured.

On a 2D shape this is probably fine however keep in mind that if it is a 3D shape, Calypso would probably have an impossible time calculating the correct nominal vector since based on the part geometry the point of contact of the probe on the part may not lie on the curve at all. I am guessing a bit here but my thinking would be in most cases actual points could be safely used on 2D curves but on 3D curves the best option would be nominal points.

[[Th...]]

Ok, so generally as long as I'm sticking to 2D curves, I shouldn't need to worry too much. Most of the curve measurements I do ultimately exist on a 2D plane anyway.

[[Da...]]

I would say ditto on the case-by-case basis. I'm running into a situation where the actual curve deviation is distorted if I use the nominal vector, whereas the measured vector more closely aligns with what the surface actually looks like when viewed with optical system. This is not the same situation with a very similar measurement in the same plan.

Nominal Spline Method
Measured Points Method