Secondary Datum Repeatability

[[Du...]]

Hi all,

I'm programming a part on an O-inspect that is a simple distance measurement that's turning out to be a nightmare when we do a Gauge R&R on the fixture. The part in question is a molded plastic part. I'm not sure I can post prints so I'll post a very expertly drawn Microsoft Paint Version of the part with the Datum in question below.

The Datum the print wants is on an edge that has a radius and draft since it's a molded part. When trying to find the datum I have to use Front light since there is a feature in the way of back lighting. I try to get a z focus on the part and then use a 2D line with 500 points to create the datum. Getting the Z focus for the datum is hard since it's on a radius and has draft, it's also a molded part so the surface is somewhat bumpy at 6.3x. I also use a point to locate the edge and use a formula to feed a Y value into the Z focus so it can get the edge better. I've also tried using filters including edge filter to help create a better image for the evaluation.

I have a base alignment set on the fixture and a secondary alignment set for each part using the datum as a Planar rotation.

So far repeatability from run to run without touching the part is repeatable but when we reposition it for the GR&R we get a variance of .02mm.

PXL_20221005_161200445.jpg

[[Ow...]]

I know nothing regarding the O-Inspect but, are you looping the secondary alignment?
With what little I've worked with optical systems, lighting, distance and part color introduce a lot of variable/variation so, it might be best to use the probe to align each part and loop the secondary alignment for each part?

Right click on the alignment and choose loop. Right click in Break Condition
and choose formula, enter it in the space directly under Boolean expression.
Then click insert and enter (in the space named "END") the number of times
you want to loop it.
Secondary alignment formula.
differenceSystem().valueA<0.0005.
If the parts are flexible in any shape or form, you might never get it to repeat.

[[Je...]]

.
I completely agree with Owen. If you can use a touch probe to find the part, you will be more successful in locating the feature of interest with optics. Perhaps construct an intersection from two features that you can reliably probe.
.

I haven't had great success with z-focus points, but maybe others have?
.

[[Ri...]]

A lot to unpack here.

Would really have to see it run. Optical measurements are always specific to the part, so general suggestions aren't always helping.

What is your reasoning for using the 6.3x zoom? Are you actually measuring z-heights with the optics, or are you just using it for autofocus?

[[Ri...]]

What is the accuracy of the part?
.02mm repeatability on plastic than can change if you breath on it. 😱
What is your humidity?
Do you know the water molecule absorption rate of this material?

[[Du...]]

We ended up using a probe for all measurements. The probe offers "mechanical filtering" of the cavities that the lights pick up and augment your alignment/datums. Sometimes you think you have this program figured out, and sometimes you spend a lot of time on something "simple".