Question about FIXED_RAD best fit and the RN value

[Douglas]

I'm checking a radius on a part, and only 47° of it can be probed, the radius has a .002 total tolerance on it. I understand that FIXED_RAD is often used to locate the center point, then that center point is used to do a LEAST_SQR best fit to determine the actual radius, can anybody confirm I have this right?

My question is about the legacy reporting of the FIXED_RAD best fit math, I'm attaching a clip of it. My take on this is that it does a best fit of that R1.712, tells me where the x,z location of that fixed rad is... this much I am pretty sure is correct, but unsure of the RN value... is this showing me the maximum deviation of the measured surface from the theoretical rad at the reported x,z value? Maybe oversimplifying it to say how much the fixed rad varies from the probed data using the measured center point? I'm basically wondering if I may be able to skip the second analysis of the feature if I tolerance the x and z as well as the RN of FIXED_RAD so that it still represents a functional fit of the mating part.

I know most of you cannot stray at all from drawing specs but I have pretty much total freedom in how I use the CMMs to approve a part, if I think engineering has messy datums(more often none at all) I use what works for me to measure it then make them follow that to disposition any deviations


filedata/fetch?filedataid=16587

This is my auto feature window, in case it answers any questions about my inquiry. Also how many probe points would you take to measure this radius? This one is probed as a cylinder and I am taking 12 hits per level at 3 different levels.


auto feature.JPG

And a picture of the feature, the cylinder is the convex bottom of this pocket...

pic.jpg

Gheez this post got a bit stupid... hopefully my question is clear among the rest

[RIDER]

https://support.hexagonmi.com/s/article/How-to-measure-a-small-arc-less-than-90-degrees-1528360401117

Here's what Hexagon says to do.

[Mike Ruff]

+1. I recommend downloading and saving that Hexagon file somewhere on your server

[Douglas]

thank you, I did download it but pretty much read it quickly and understand, the alignment would be the same as my part alignment except for the measured deviation in x,z datums. Given this is the case should I not be able to plug the measured x,z values into a best fit least sqr and get the same result without using polar coordinate system or changing the alignment at all?

[SingularitY]

Interesting. I've never used the fixed radius option but it seems like it would be very useful!
I usually just use a slow, high point density scan (or several scans) and construct the feature for dimensioning. Has worked for me so far. Auto Measurement Routine scanning has also done me really well. (I just haven't used it for anything that wasn't a full arc yet.)

[Roberto]

If the surface vector of the cylinder is -Y (0,-1,0) change the value for starting angle to either X or Z (i.e. 1,0,0)

[Douglas]

yes that should be done... my code increase the risk of it flipping vectors on me I think

[InspectorJester]

I don't know if tolerancing X & Z fulfills the same requirements as the blueprint; check with engineering before going this route, at least to cover yourself if they ask (which, I mean, they probably won't...)

I can verify I have had much success using the Polar Vector Compensation method; this ideology should allow you to see any deformation in the material, and, more importantly, where that deformation occurred

[Douglas]

Engineering has improper datums on the drawing, current engineering manager blames past staff and doesn't care as the product is in service now. Guess I'm just looking for a shortcut and I better stop that...