Use of Filters & Outliers

[[Ke...]]

Theoretical Question (based on real life events):

Say you have a cylindrical bore on a component made of alimunum (or aluminium, if you prefer).
Lets also say that the specification of the diameter of the bore is 50mm +/- 0.02, with no seperate roundness call out.

Note:Due to the depth of the bore, and the only way to fixture the part is by aligning on the exterior of the part, which is a rough casting, I am using an 8mm dia stylus, to ensure that the stylus is not shanking.

What Outlier settings would you choose to apply, and why?

Thanks in advance, heroes!

[[Da...]]

We do lots of cylinders here. I would check it as 3 separate circles. Ask for roundness at 1/2 the diameter tolerance and set filters at 50 upr. Outliers would include 5 adjacent points and use 2 iterations. Pre filter at 15 upr. You should be using a fairly high point density.
Then check the part a few times and look at the roundness charts. If they are acceptably smoothish ( 🤣 ) then you're all done , If not adjust accordingly.

I only recall points into a cylinder to analyze max inscribed or If the customer wants Cylindricity. Using circles allows you to see taper, roundness, and size variations that are important to making a good part.

[[Ke...]]

Please sign in to view this quote.

Do you use the default of 3 standard deviations for the Outliers? Not sure if/when to use tighter settings.
I do always include 5 adjacent points (with 0.05 step width for most features).
I tried 2 iterations once r twice, but when I did it looked like it filtered out more data than I was comfortable with... maybe it depends on the form of the part(?)

[[Ja...]]

Let me preface this by saying I’m definitely not an expert, but I have been programming for a fair amount of time. I sometimes use individual circles as Dave Scott mentioned. Depending on the length of the bore and tolerance, I may use several circles. For any tight tolerance bore (we make industrial and commercial pumps) I always evaluate as a cylinder as well because there’s going to be a tight-fitting mating part that’s probably cylindrical. I would suggest using a bare minimum of 2000 points per circle for your size bore. For me personally, I’d probably use twice that many. The more data Calypso has to work with the better. For UPRs I would set the filter at 196, which should be in the ballpark for ISO 12181. I’d set pre-filters at 4 and 984. For that tolerance window I’d set sigma at 2.8. It will only be deleting somewhere around 0.3% of the data, and it’s another good reason to collect data at a higher rate. I almost always use 2 iterations of “To computed feature”. In your case, I would use it because you just don’t know how good the relationship is going to be to the casting from run to run. I verify the diameters on larger bores by using a 2-point or 3-point bore gage. The bore gages should always show the bore larger than the CMM (I use max inscr obviously). If it doesn’t, I don’t have the CMM coverage high enough and need either a higher data rate or more circles, or both.