Parts over engineered?

[[Er...]]

I had to submit a part to a customer that failed a profile .010", measuring .036" in it's worst location. This was checked using the Duramax. I created a OK/NOT OK evaluation plot with the highs and lows and sent it for a graphical representation that I have uploaded here.

This is the response we got back in return that I thought many of you would appreciate:
Here are the comments from Engineering, FYI:
"I’m quite impressed and pleased that they are actually measuring the P.O.S. requirement. It appears that they’re using a vision system to check this, and the graphical representation is very clear and helpful. I have examined all three plots and determined that the build up in the corners will present no problem to the assembly. Please process a deviation."

Apparently, many parts have callouts/tight tolerances that are in many cases unnecessary. 🤣

[[An...]]

We also very often have to produce parts that are over-toleranced.
Many designers work with very small tolerances because they do not know the true requirements of the part and this results in what I call fear tolerances.

[[Ma...]]

And many of them don't know even limits of machining machines. These obvious deviations should be considered in pre-drawing

[[Ri...]]

Most parts are over-engineered. I've seen Profile callouts on weight saving features. 🙁

[[Ow...]]

The Profile characteristic to an engineer is like the internet to modern society, it’s quick and easy, saves a lot of time and you think it covers everything only to find out it’s fake news. 🤣

[[Ri...]]

I just noticed that the majority of the deviations occur at the corner radii. Is your stylus tip small enough to measure these features? Otherwise tell the tooling guys, or the machining guys to cut the correct radii next time.

[[To...]]

Could also be due to milling too fast and getting tool deflection in the corners.

[[Er...]]

@Richard,
No, the stylus is a 1mm probe in the smaller .0625" corners and a 2mm for the larger .125" corners. I start with very slow speed to get a baseline of what it measure and then gradually increase my scan speed from there to know i'm not losing any accuracy on the 3dcurve measurement.

@Tom,
You are the winner. Typically, and too often parts are programmed using the same size tool as the radius and then machined too fast so this is all from tool deflection. I have had the programmers literally ask if I would change my inspection programs to not inspect the corners. My answer always is to get the customer to change the print to exclude corners, or a document stating the corners don't matter, and then I will change my program.

Thanks for all the feedback guys!

[[Er...]]

@Richard,
Talking with the customer we found the most of these pockets are weight saving features just as you suggested. Which is why they themselves called this, "a P. O. S. requirement" 🤣

[[Me...]]

A true Design Engineer does his stack up tolerances to avoid this kind of garbage. If I had to guess, I'd say that most engineers don't even bother and are guessing what tolerances and GD&T FCFs should be.

I'm working with a genius Design Engineer that speaks heavily on stack ups. We're working together with PMI & this other stack up software and I've got to say its amazing.

[[Cl...]]

Please sign in to view this quote.

That's when you start looking at fit, form, and function.

[[Er...]]

Exactly. Unfortunately, we don't do the assemblies and I have not seen everything what goes into them to know for sure. So my hands are tied, and I can only follow the print requirements.

[[No...]]

Please sign in to view this quote.

Best comment I've read in a longt time! 🤣 🤣

[[Pe...]]

Please sign in to view this quote.

CNC programming the same size tool as a corner radius seems to happen more often than not, and I always have the same issue with corner profiles.

One useful trick I know is that I'll ask the programmers to plunge/feed the endmill straight down into the corners after they sidemill. It doesn't take much programming or cnc runtime, and it always gives a good profile result measurement afterwards.