Hello, have been reading this forum for a while now, great source of information. Recently question came up about measuring method so I would like to get advise from the experts. Based on the drawing datum A is the surface of the the stamped part, datum B is is a hole without any positional tolerance and than there is a round slot. Length and width of the slot are dimensioned separately. Length of the slot is marked as datum C and has non diametrical TP 0.25 MMC, A, B MMC with the associated basic dimension of let say 4.5 mm in X direction (looking at the drawing length is in X and width is in Y). Width of the slot has also non diametrical TP 0.25 MMC, A, B MMC, C MMC with the associated basic dimension of let say 17 mm in Y direction. Both hole and the slot are on the same plane A . At some point insert stopped fitting so measurements started at both ends (our company and supplier). My results were showing TP out with center of the slot being only 3.7 mm in X and 17.15 in Y. Company which makes inserts admitted their punch was loose but said position is still in spec. I was setting alignment by measuring 2 radiuses of the slot, creating line between their centers and rotating alignment to it, they simply used 4.5 mm basic to create offset alignment which also automatically made 17.15 in Y shorter and almost perfect. They are also reporting TP of the slot's center in only one direction since in the other direction would be perfect based on the offset. I presented position in 2 directions (thanks to all the posts regarding generic features and variables), also tried explaining that TP callout is not only for the center but any point along slot's axis. I need some confirmation or correction to my method.
In your first post you said that length is clocking, now you are saying width? As for the rest, isn't that what I said earlier that all the points along each axis should be within +/- 0.125? Doing 4.5 offset makes slot skewed and points at the end of L axis are outside +/- 0.125. I feel like we are going in circles here.
No. You're mixing things up.
1) The hole and slot length are your rotational constraint. The two, together, stop the rotation of your DRF.
This happens BEFORE you check the TP of the width (aka, the clocking of the slot).
2) The TP of the width controls the clocking
of the slot. To pass the TP of the width, yes, all opposed points would need to be +/- 0.125, but
you don't base your DRF on that! You check to see if it's right
after you're set your DRF.
These are two completely separate things.
Yes, this does mean that if the slot is at a 45 degree angle from how it should be, that it won't measure very well
because that means the part is BAD! It doesn't mean that something is wrong with the print.
No, there's one way that it should be measured. You just don't seem to want to believe that you are wrong
Not exactly, I am opened to learn. Just don't understand why TP along the full length is not being checked but only in relation to center point. Since width is the shorter distance than length checking only TP of the width (along full axis) can show part as being good, when points at both ends of the length can still be outside +/- 0.125 and part simply will not fit where it supposed to. And that is exactly what happened in real life. Part fits on top of the pin where datum B is and slot fits on top of the diamond shaped pin where the slot is. At the same time outside profile of the part has to fit inside reverse profile. At some point one of the 2 cavities stamped by the supplier stopped fitting, but measuring with the offset alignment and checking only TP along the width shows part as good even though supplier admitted that the punch was loose.
If it fits on the diamond pin, that doesn't tell you anything about whether the width of the slot measures good or not.
The fixture, with a diamond pin, is doing EXACTLY what I've been describing
It has a diamond pin and I agree with you. All I am saying that 1 of the 2 parts stopped fitting (it actually has to fit inside injection mould not a fixture). Punch making this part was loose. But when the part is measured with offset alignment, position along W axis is still good because it is shorter axis (and actual slot is off more to the right than down). Points at the end of L axis are no good but they are not checked, only center along L is checked showing perfect position since it is based on offset from the drawing.
It has a diamond pin and I agree with you. All I am saying that 1 of the 2 parts stopped fitting (it actually has to fit inside injection mould not a fixture). Punch making this part was loose. But when the part is measured with offset alignment, position along W axis is still good because it is shorter axis (and actual slot is off more to the right than down). Points at the end of L axis are no good but they are not checked, only center along L is checked showing perfect position since it is based on offset from the drawing.
Ok then. First, inspect the part to the datum structure. If the slot
is shifted, then the entire rest of the part
should fail.
After it fails, you go back and add more data where needed to prove that C is the failure point.
How do you know that "Points on the end of L axis are no good" if they "are not checked"? If you are only checking a single point on the width of C, of course you won't be able to prove that the slot is shifted. Like I said, If you're trying to prove what is failing, then you need more data.
Measure 3 sets of opposed points. One set towards each end, and one set in the middle. If you are right about the slot, the sets of points on the ends should fail. Viola, there's your proof.
The center point of the width can also potentially fail, if the slot is shifted far enough right/left.
Can I take opposite points on the radius (since it will be skewed to CS) or only on straight edges of the slot? Those are very short and midpoints based on sets from straight edges only will still be good. I could just use math to show that if the midpoint of the set 2 mm to the left from the center is so much off in Y than the midpoint 6 mm to the left will be that much off based on ratio but I would like to avoid this.
Well, if the slot is skewed so far that taking opposed points would mean that it was measuring into the radius, that would still give you really bad results (the cosine error from the slot being skewed would pretty much cancel out because it would be opposite on each opposed point). I would place the points on CAD, and just let them measure as they are.
You could also do a linear closed scan of the slot and dimension it as a profile with graphical analysis to show how it is deviated.
Or you could measure the radius on each end and dimension them. Their centers would basically be equivalent to the midpoints of the opposed sides at the tangent.
