True position off by 0,5mm

[[Ar...]]

Hello,

We have following part under inspection and there is position tolerance for hole and two slots.

Part is good, but I keep getting 0,5mm error when using true position. Is there any tips and tricks for this?


BR Artturi

[[Da...]]

Please sign in to view this quote.

Not necessarily the reason, but this primary datum (A), doesn't look very stable, so you are probably getting non-repeatable measurements also?

[[Je...]]

Neither Datum A nor Datum B will likely produce repeatable results due to projection error. I am curious how you have determined "Part is good" but CMM reporting shows non-conforming.

[[Jo...]]

Please sign in to view this quote.

Datum A is the surface.. not a cylinder (origin, axis) correct ? Should this be accomplished using a best fit alignment ?

[[Ar...]]

Please sign in to view this quote.

Okay, maybe it's too much to say part is good. But it is definetly not off by 0.5mm.
There is additional dimensions for centerhole. These dimensions are out of tolerance by 0,01mm
Datum A diameter is under tolerance 0,05mm.

I doubt that these will together add to 0,5mm error in position?

Please sign in to view this quote.

Datum A is cylinder, but its very narrow one, 8mm widht.


-----------

Is there any effect if I choose my datums according to the drawing? Should True position feature "create" new alignment?

BR Artturi

[[Ma...]]

Please sign in to view this quote.

I would clarify - result of TP is doubled deviation - so you are off by 0,25695mm.
If you have curve / FF you can lookup how it looks.

[[Jo...]]

In ASME symbology it would mean the surface.. there is no diametric symbol, or feature of size, or leader associated with Datum A

The geometry may be cylindrical..

leader.jpg

[[Mi...]]

With a cylinder angle less than 180° the cylinder inserted in the first datum is not the best solution. Unfortunately, incomplete scanning of this cylinder could lead to the cylinder center and diameter being moved to the wrong area. You can change the cylinder scan path and see that the different true position result increases when the cylinder path range is small.
In any case it is possible to see on the result which direction is out of tolerance. The error is probably along the Z direction.

[[To...]]

As I see the CAD file the datum does not have the same angle as the circle. Could that be the reason for the error on the position?

If look at the nominal angle on the feature and rotate the datum in that angle. This could give you a better position.

[[Ja...]]

The datums A and B are locking only 5 DOF here. Just a thought that choosing primary, secondary and tertiary in the TP characteristics with features that are not 3-2-1 systems might be the culprit here.

As suggested by John, use of best fit alignment is one of the solution here. Create a best fit alignment using datum A and B with eval constraints selecting the 5 DOF that dat A and B are locking. For your hole features in this case that are coaxial to the datum A you can use this BF in the GDT characteristics directly as last DOF will have no effect.

For features that are dependent on the last translation DOF, simultaneous requirement comes in to play. For those you need to create another best fit alignment referencing the first BF alignment and selecting all dependent features that you need to output some GDT with dat A and B (hole in this case) with eval constraints selecting that last DOF (last remaining translation) and use that in the GDT characteristics.

[[Je...]]

Evaluate the location of the hole in terms of straight line distance from -A-. This will give you an indication of the issue is Feature-A relationship or a Feature-to-B/A-B relationship.

If linear distance from single hole to -A- is well within .25 (over .25 would force TP be over .5 right off the bat). Then, assuming -B- is a straight line to center of -A- (this only works if it is), create a point on -B- and connect it to -A- with a 3d-line. Use that 3d Line for your rotational constraint. Re-evaluate your position using this method. If it is well under the previous reported value then the issue is the related angle of -B- to the center of -A- (and this is VERY likely the cause). The datum structure should be using -B- as a rotational stop about -A- but CALYPSO will be using the A1 of -B- unrelated to -A- as its axis of rotation.

You can confirm that -B- is not on center to -A- by reporting a perpendicular distance. It should be nearly 0.0000 as that is what CAD indicates. Anything above 0.0000 is having a substantial effect on your position reporting.

Laymen terms - Just the form/sigma of -B- using LSQ can have a huge effect on your position reporting due to projection error.

[[Ar...]]

Please sign in to view this quote.

I Have not yet tried (never) BF alignment, so I need to get familiar with it. As I understand the real benefit using BF alignment is when there is multiple holes to be measured?

Currently I try to get good results from only centerhole. When I get that figured out I will move on.

Please sign in to view this quote.

Linear distance error from -A- to feature is 0,7mm. Angle of -B- Is not straight to center of -A-. It's off by 0,3 degree. However, I did try it anyway, see attached picture. Error decreased significantly.

Part is made from 3D-printed inconel and machined afterwards. There is really much distortion from inner stresses in part.

I do have similar parts with thinner machining margin. I think those have less distortion.

I'll check back when I have time to inspect them.

BR Artturi

[[Ar...]]

And one more picture...

There clearly is something odd going on with this part as you can see from diameter no.9. It's off by 0,2mm.

[[Ar...]]

Please sign in to view this quote.

This is really weird. I did run scan with GOM laser scanner for the part and it shows different story than CMM. Which one is right?



Here is picture of comparison of scanned profile and 3D-cad file. As you can see there is not massive 0,2mm deviation. I did use local best fit though. I guess it's going to distort results a little.

Edit: It's very different thing to compare scanned model to actual model and actually measuring the part. Even if the diameter of section would be off by 1mm, it wouldn't show up in scan inspection because of how small the scanned angle is.

[[Ja...]]

Please sign in to view this quote.

That is different you are talking about Best fit Bore pattern that is found in the true position characteristics.

I am talking about Geometry best fit alignment (in your case.) that is found in the resources menu. This will output an alignment by best fitting selected features based on selected evaluation constraints (5 DOF in your case leaving 1 translation DOF) with remaining DOF (1 translation DOF in your case) falling back on the selected alignment in the drop down menu of the dialog box.

[[Ar...]]

Please sign in to view this quote.

Thanks. Dropped the true position from 0,2mm to 0,08. Which I think correlates well with fact that diameter of reference -A- is much smaller than nominal.

Thanks to everyone helping me out with this.

Br Artturi