True Position - Drawing Issue?

[[Ja...]]

I have attached a stripped down drawing of a situation that I am asked to program. The part is symmetrical and they are asking for Position of each bore. Datum -A- is the shank, datum -B- is one of the faces (perpendicularity control). My concern is as follows:

One bore is referenced to -A-, -B- and that makes sense. The other bore is referenced to -A-, -B-, and the first bore. This seems fine as to include the C/L distance. Here's the problem - If you get the right set of numbers, the part will be in tolerance one way but upon rotation of 180 degrees, the part will fail. In other words, the C/L deviation is only considered in one Position callout, so if the bore already deviates from the shank by a lot (but not OOT worthy), it may just fail if that bore is the one with the added C/L deviation.

Management tells me: If you get a bad result, flip the part and it may be good.

This seems absurd because the GD&T should allow it to be either good or bad. Not bood.

B141786.jpg

[[To...]]

Datum A is controlling the Y location, perpendicular to A, and parallel to B. The only thing C is controlling is the X location from C.

When you say flip 180 degrees, what axis are you flipping it about?

[[Ja...]]

Assume X left to right as shown, Y down to up as shown, Z into the page as shown. Datum -B- vector lies in Z.

-C- can deviate in Y from -A-, and aside from perpendicularity to -B-, this is how we calculate position (2 dCy).

-D- can deviate in Y from -A-, and aside from perpendicularity to -B-, can deviate in X from -C-. (2*sqrt(dDy^2 + dDx^2)).

If part is rotated 180 degrees on Z (swap -C- and -D-) then now we have the reverse situation. If dCy was high but not OOT, combining it with dCx may bring it out. Does that make more sense?

[[To...]]

Are both faces, left and right, used to establish B? How parallel are both faces?

[[To...]]

Please sign in to view this quote.

B is only conrolling perpendicularity in one direction, A is controlling the other direction.

[[To...]]

Please sign in to view this quote.

B is only conrolling perpendicularity in one direction, A is controlling the other direction.

[[Ja...]]

There are technically four possible faces for -B-, two with vectors into the page, two with vectors out. The width of the opposite end of -B- is held parallel to -B-. The remaining face is not related to any of them

[[Ja...]]

Tom, let's ignore the perpendicularity for the sake of this argument. The issue that I am having is with the Position. Let's remove -B- from the picture/FCF's and reconsider what we're dealing with. It is possible for one part to be good and bad, depending on which way it is oriented. This doesn't make sense to me.

[[Ja...]]

Maybe this will clarify things:

[[To...]]

Unfortunately, I can't ignore B because with B as part of the structure, I could see a difference in the results if you only used 1 surface for B. However, let's say we can ignore B and C was positioned only to A. C would rotate around A until it was straight up and down. Then, when you checked D to A&C, I would expect to get the same results if I flipped C&D. At least in my mind's eye.

[[Br...]]

Please sign in to view this quote.

Hes saying since the part is symmetrical, either bore can be Datum feature C, and either side can be Datum feature -B-. So depending on which bore and which side he chooses to be -C-, and -B- respectively, the part will either pass or fail. He has 4 possible combinations of ABC datum reference frames.

Y14.5 doesn't really say you can't do that. It recommends physical identification of the datum features on the part in some way in this case. But I don't think its forbidden. I guess you just keep flipping the part until you get it to pass. If all 4 fail, then its a non conforming part. I know its a lot of extra work, but from the snip you have here, there's nothing wrong with the drawing. You also have a datum shift on datums A and C. I would definitely use that to my advantage.

[[To...]]

Where are you getting your X and Y values from? The features? Are you constraining per ISO5459? If so, the data in your features is pretty much useless. Plus, the X and Y values are coming from one end of the cylinder axis (at the feature trihedron) and don't take any angular deviations into consideration. Position does.

[[Ja...]]

Thank you Brett,

My concern is this. One of four combinations measures OK, the other three NG. I am told part is OK (because I have a good printout) and to send it.

Customer (75% chance) measures NG. They are not happy. Do I tell them "Flip it and rotate it until it's good"?


This is clearly not well-defined.

[[To...]]

My bad. I thought you were questioning "How can it be good one way and bad another way?"

[[Br...]]

Please sign in to view this quote.

I would agree with you. I would ask them how you know which bore is datum feature C and which plane is datum feature B. If the part is completely symmetrical there's no way to tell. Maybe they will have an answer for you. Maybe they will want some sort of a marking to identify the datum features. Until then, I don't see how they can argue with you over it.

[[Ja...]]

Tom,

I clearly see many ways that it can be good one way, bad another. That is my problem. This shouldn't be allowed.

A part is either in tolerance or OOT. My car tires are never full and flat.

I just don't know what course of action to take. Is this a fundamental GD&T issue, should I petition to have a composite DRF to strictly control C/L?

[[Ja...]]

Thank you Tom and Brett,

I will try again to take this up with Engineering. I apologize if I have misled you.

[[To...]]

The other option is to push back to manufacturing and say "Make it so it's right any way you check it." Not a popular stance. LOL

[[Br...]]

The issue from a design perspective you have here now is at assembly, how does the assembler know which way this part fits? Many cases when you have a symmetrical part like this, both of those bores fit simultaneously, and therefore should Both be datum feature C. Datum Feature C would be a pattern then, and it wouldn't matter which way it was oriented at inspection or assembly. But who knows? Maybe there's a reason for this.

[[Aa...]]

Please sign in to view this quote.

Exactly.

From ASME Y14.5-2009, Section 4.8,

Please sign in to view this quote.

It could be that at assembly, one hole is slid onto a longer shaft before the other hole is slid onto a shorter shaft. But how would the assembler (is that a word?) know which?

[[Ja...]]

Please sign in to view this quote.

You're right. I had seen that earlier in ASME but we do not identify one side vs. the other. This may be the reason why they tell me "flip it and try again" and they may tell the "assembler" the same thing. But we would save inspection/assembly time by marking one side over the other if this is the case. Maybe this is just unimportant to others.

[[Me...]]

Please sign in to view this quote.

This ^

If you're allowed to flip the part then the holes are a pattern and should be checked as a pattern. Same thing with -B-. I'd use a symmetry plane created from all 4 surfaces for -B-.

Robert

[[An...]]

Please sign in to view this quote.

If 3 fail, then it's a non conforming part ...
If 2 fail, then it's a non conforming part ...
If 1 fails,then it's a non conforming part ...

If 0 fails,then it's a conforming part ...

You have to measure all possible combinations and all results
must be within spec!
The requirements regarding the complete part geometry might
be very high. (designer's concern)

[[Br...]]

Please sign in to view this quote.

Absolutely incorrect. If you have a passing part in one single orientation, you have met the drawing requirements. You can not prove that a different orientation was intended by the drawing due to its symmetrical nature. It doesn't even make sense from a design standpoint. To say all of those tolerances apply in all orientations is actually an over-definition of the drawing. Its like saying a diameter has two different tolerances. It doesn't make sense. There's nothing in the standard to support your claim. In fact the standard says the datum features need to be readily discernible. So in this case, they are not.

[[De...]]

The real question is what is the receiving inspector at the customer going to say?

You can argue all day about the spec says this or that, but at the end of the day what is going to matter is whether the customer agrees or not.

Most customers whether it is their bad print or not are not going to be willing to re-run their own part 4 times, they will just run it once and then reject it and then its right back to being your problem to fight with them why you say its good and they say it isn't.

For us the hit we would take in our quality rating for even having the argument is not worth the chance.

It is much easier in my opinion to just make the program handle all the configurations, I have a part like this for a very large well known corporation where I have to have the C Datum 12 times and have 12 times the number of characteristics related back to it in order to assure no matter how they check it they will always have a conforming part.

Its stupid, and the datum feature should be marked or something, but the day you can get engineers to actually do things right all the time, they stop making drawings and they hire someone new.