Filters for Diameter, suggestions?

[[AG...]]

The Diameter is 38.038", i am checking it at the Radius, the length of the block is 11.000". As you can see its just a small area of the whole callout. I am looking for some filtering for this particularly crazy area to get the correct dimension. My dimensions are all over the place just looking for some advice. I have messed with speeds and and points and scans i am just not convinced/comfortable of the output the CMM is giving me. If there is any advice it would be greatly appreciated. The engineers have already told me that CMM's don't give good Diameter/Radius readings. I have never had problems in my career with diameters but they seem to think so. 

[[Ma...]]

Since it's small area, then you will always get unstable results.

Much better way would be surface profile callout. You will see those tiny deviations which will result into unstable measurement.

[[DW...]]

Please sign in to view this quote.

Please sign in to view this username.

 Large diameters from small features are inherently unstable. My recommendation is switch to single points (4-6) and constrain the location in the appropriate axes (for example X and Y).

[[RP...]]

With large incomplete radii like this, my go-to method is constraining a radius and recalling the points to another circle feature.  Then constraining the location of the recalled feature.

[[na...]]

Please sign in to view this quote.

[Like button pressed hard]

[[To...]]

Question to the group.  If you measure a short arc segment and the form results are very, very low, but the radius is off, which value is more important?  Just to add, when the location is constrained, the form gets worse.

Edited April 17, 2025

[[Ma...]]

Please sign in to view this quote.

Tom, I think the answer would depend on what information you are looking to obtain from the measurement.

If you are trying to find the best approximation of the actual radius then the result with minimum form is best.

If you are measuring for compliance to a specific radius tolerance then evaluating the radius values of the points to the constrained location obtained from the constrained radius would be the best.

[[Ow...]]

Attached is simple power point with video's (click on illustration to show start button) I created several years ago for internal beginners training. 

It seems like last time I shared it there was some devils advocate comments on it but, it is what it is and helps some understand how and why small arcs of circles/radius's measurements can be challenging 

Small Radius limitiations on a CMM.pptx

[[Ma...]]

If you constrain location, then you are defining radius value within a few 0.0x mm.

Bad thing is that unconstrained radius will have form in 0.00x mm and constrained in 0.0x mm - which is debatable.

Whenever i am dealing with this i often take into consideration how it is machined and what can go wrong which can result in bad machining,

[[AG...]]

I really appreciate all the responses to my question. I ended up after finding that scanning the Radius wasn't repeating i then took points over the majority of the radius and constraining like was suggested to me. This worked out better and more repeatable. Yes like Tom is saying when i constrained the radius the form got worse but my numbers were more repeatable. This helped a lot, thank you. 

[[AG...]]

Please sign in to view this quote.

Thank you for sharing this, i understand and see how all this works. I actually had a previous question about a project that was a huge Radius that was split up into 3 different blocks. I had a lot of information that helped me through that also. This is similar to that and now i see it.

[[To...]]

Please sign in to view this quote.

It's not CMM's specifically but rather any measurement process that calculates the circle using points, i.e. using cross-hairs on an optical comparator. It's just math.

 

Edited April 18, 2025

[[Cl...]]

For what it's worth. I was shown this 25 years ago. It's a simple, quick sanity check.

You can either use a theoretical line for the construction or construct intersections if the geometry allows/dictates.

In the case of the latter scan/probe the radius and the two lines Intersect each line with the radius. Construct a

mid-point from the two intersections. Create a secondary alignment using that mid-point (XY origin). Now probe a point

on the radius face at X zero. Translate that point -Y by the nominal radius. Switch to polar coordinates and probe points

on the radius. This will best show the actual radius.

 

 

Edited April 18, 2025

[[Ma...]]

For this I would Radius Measurement Points. Take 3 or 4 points on the radius the radius measurement points will report the 2d polar radial measurement from the nominal of the circle 

[[Cl...]]

I was shown that when I was using Virtual DMIS 25 years ago.

[[AG...]]

Please sign in to view this quote.

If i constrain my radius at the size i have on print for first feature. I then recall that feature ( points ) and constrain the location i am getting in print Radii.

[[Ke...]]

Please sign in to view this quote.

Out of curiosity... what is the tolerance for the diameter?

[[AG...]]

Please sign in to view this quote.

+/-.005"

[[DW...]]

Please sign in to view this quote.

Please sign in to view this username.

 This is not related to the initial post, but if you need to write the (plus and minus sign), hold down ALT and type 0177 which gives you ± 🙃

[[Ke...]]

Please sign in to view this quote.

My 2 cents... they are equally important: the form is getting worse, because the location is being forced toward nominal, due to the constraints. While I agree that constraints may be necessary, at times: it is best practice (in my opinion) to use constraints as little as possible.

A good practice might be to develop a rule of thumb, such as:

• "if the form deviation of an unconstrained arc segment is "X" microns, then the form deviation of the same arc segment, while constrained shall be no larger than 2 times "X" microns (the form deviation of the unconstrained arc segment)"
• ... or using another metric, simply "the form deviation of a constrained arc segment shall be no larger than the allowable form deviation of the feature being evaluated"

...thoughts?

 

Edited April 30, 2025

[[Ma...]]

Please sign in to view this quote.

That's what i wrote in my second post here.

I don't fully understand logic behind reporting location with constrained radius and then report radius with constrained actual position.
It makes no sense because you are forcing those values by this method. Not sure if it's correct way. I would rather ignore wrong values and use profile to show deviations to prove, that it's machined properly.

[[Ke...]]

Please sign in to view this quote.

I 100% agree with the use of profile... the only problem is when that avenue is roadblocked by a refusal to update/modify the print; although, I still may report the profile as a reference characteristic.

[[To...]]

Please sign in to view this quote.

That's what I do.