Open question about profiles with touch trigger probe.

[Jim Poehler]

I find myself more and more frequently needing to do profile measurements of multiple features/surface areas on some very small parts. I usually use a linear scan to accomplish this but sometimes the results are not all desirable. I usually end up measuring the basics of the features as well as more and more customers want to know the actual values as well. When those are well within expected boundaries to CAD nominals I find it difficult to see why profile form and location can be so far off. While I know a good portion of the problem lies with the design engineers' call outs of FCF datum selection, ie using an extremely small surface area as the primary datum, I am looking for some feedback as to the best approach for profile dimensioning with offline CAD models. Combining adjacent features does not produce a good profile and I have has some luck with taking individual point hits along an entire surface area but this takes a good deal of program and inspection time to accomplish. I use a TP 20 as I don't have a TP200, scanning probe. Any ideas as to best approach which would give me the most reliable information would be greatly appreciated. I don't find much about profile measurements in any help files. I am also unsure of unilateral profile call outs as to whether to add a + something value and -0 or the other way around. Not sure what PC-DMIS is considering the boundary area for unilateral profile. Thanks for any discussion that this might lead to.

[RLMorris]

Unilateral

Unilateral.jpg

[ir a programmer]

I find myself more and more frequently needing to do profile measurements of multiple features/surface areas on some very small parts. I usually use a linear scan to accomplish this but sometimes the results are not all desirable. I usually end up measuring the basics of the features as well as more and more customers want to know the actual values as well. When those are well within expected boundaries to CAD nominals I find it difficult to see why profile form and location can be so far off. While I know a good portion of the problem lies with the design engineers' call outs of FCF datum selection, ie using an extremely small surface area as the primary datum, I am looking for some feedback as to the best approach for profile dimensioning with offline CAD models. Combining adjacent features does not produce a good profile and I have has some luck with taking individual point hits along an entire surface area but this takes a good deal of program and inspection time to accomplish. I use a TP 20 as I don't have a TP200, scanning probe. Any ideas as to best approach which would give me the most reliable information would be greatly appreciated. I don't find much about profile measurements in any help files. I am also unsure of unilateral profile call outs as to whether to add a + something value and -0 or the other way around. Not sure what PC-DMIS is considering the boundary area for unilateral profile. Thanks for any discussion that this might lead to.

Jim, here are some color-coded thoughts to address two areas in your post. I report Profile quite a bit in PCDMIS:

We have some parts where the datum structure used and the nominal point origin(s) are not always identical. If the incorrect origin is used, the Profile will show to be "far off". Check the drawings to determine if this condition exists.

PCDMIS reports Profile as +/- material condition. Ex. - A bore with a Unilateral Profile tolerance of 0.01in in which the bore Profile can be larger but no smaller, PCDMIS Profile must report this as MINUS material, even the tolerance itself may seem to be PLUS. As materila is removed, it becomes a feature "minus" material. The Profile tolerance must be set to -0.01in. for PCDMISD to report correctly.

HTH

[RandomJerk]

Here's a reason why the basics that comprise a profile tolerance are irrelevant. You can draw this out on CAD if you need to.

Say you have a basic of R2.3 with only 5 degrees of arc and a bilateral profile tolerance of .02 (these numbers are sort of off the top of my head). It's entirely feasible that a straight line will fit comfortably inside of that zone (and if it doesn't, increase the tolerance a little bit).

So, what's the radius of a straight line? Infinity? Is the part good or bad? The answer is good, because, per the standard, the basics do not refer to the feature but to the tolerance zone.

8.2.2 Profile Specification
The profile tolerance zone specifies a uniform
or nonuniform tolerance boundary along the true profile
within which the surface or single elements of the
surface must lie.

So go ahead and report dimensions, but go to F10 and set your columns to report Nominal and Measured, deviation if you want. A customer can do what they want, but technically you can't have a part rejected because a basic dimension on a profile tolerance is a certain value.

[Josh Carpenter]

(snip)... I find it difficult to see why profile form and location can be so far off. While I know a good portion of the problem lies with the design engineers' call outs of FCF datum selection, ie using an extremely small surface area as the primary datum, ...(snip)

You just answered your own question. Insufficient datum land can make a very good part look very bad, very easily. I made a long perfectly lathe-turned part show end-to-end concentricity off by 0.125" by using too-narrow datums.
In these cases, I recommend "real worlding" the datums: find out what mating features really control the part's end use and use those to align to. Often the designer datums are theoretical and not practical - we need practical.

(snip)... Combining adjacent features does not produce a good profile and I have has some luck with taking individual point hits along an entire surface area but this takes a good deal of program and inspection time to accomplish. I use a TP 20 as I don't have a TP200, scanning probe. ...(snip)

First, programming time drops when you use Point Only Mode.
Second, take a good long beep-beep-beep program and track your runtime measuring all those points. Get us to take the same part and do a demo program on it with a scanning probe and track our cycle time. Do the math of time-savings and present that as an ROI business case to get your CMM upgraded to a scanning probe.

(snip)... I am also unsure of unilateral profile call outs as to whether to add a + something value and -0 or the other way around. Not sure what PC-DMIS is considering the boundary area for unilateral profile. Thanks for any discussion that this might lead to.

Example callout: profile of a surface 0.010 Circle-U Symbol 0.002
What this means is: total zone allowed is 0.010, plus side is 0.002... some quick work with my fingers reveals that it must be +0.002 -0.008.
So, it's always: total zone #, circle-U, plus side #, do the math to find minus side.

If you are using Xactmeasure mode it does it for you when you just build the callout as shown on the print.
If you are using Legacy mode you get to type in the plus & minus.
HTH!

[Jim Poehler]

Jim, here are some color-coded thoughts to address two areas in your post. I report Profile quite a bit in PCDMIS:

We have some parts where the datum structure used and the nominal point origin(s) are not always identical. If the incorrect origin is used, the Profile will show to be "far off". Check the drawings to determine if this condition exists.

PCDMIS reports Profile as +/- material condition. Ex. - A bore with a Unilateral Profile tolerance of 0.01in in which the bore Profile can be larger but no smaller, PCDMIS Profile must report this as MINUS material, even the tolerance itself may seem to be PLUS. As materila is removed, it becomes a feature "minus" material. The Profile tolerance must be set to -0.01in. for PCDMISD to report correctly.

HTH

The information on profile interpretation relative to CAD programming is exactly what I was looking for.
Trying to understand the datum structure used and nominal point origin is a little more challenging. When I import models I always transform them to fit the inspection orientation on the CMM. I use read point exclusively which means that I have redefined the nominal coordinates of the model to reflect where I set my 0/0/0 position. I will always align the part to the most logical datums as suggested by Josh. When I begin a scan, or a set of points I am usually in that alignment origin. However, when I call for profile, even if the datums required are ridiculous and sometimes not even feasible (pc-dmis lets me know if it can't be done, although I already know, just need to try and please the design engineers) and I use the specified FCF datums does this relate to the statement of incorrect origin? Do I need to create an alignment using the specified FCF datums before I take the hits or begin a scan? Perhaps some of the problem lies within this unclear area. Any clarification is always appreciated.

I do thank all for the input so far; it helps to make this a little easier to do. I understand that some of the profile envelopes are not toleranced well for the parts we are making but that is another story.