Dimensioning for Interchangeability 1

[SwinnyGG]

Thats very clearly not what I said. Not even remotely.

I agree, you didn't say that- but that's the conclusion one reaches if your line of thinking is followed to the end.

 

[chez311]

that added subphrase serves only to aid in the ability of a Y14 compliant manufacturer to alter their processes as necessary without blowback from an engineer or designer

Thats literally all I said from the beginning. Go back to my initial reply to you (19 Oct 21 22:44) - you said a nominal value tells manufacturing "where we want the center of the statistical distribution of feature sizes to land" and I said no, its up to manufacturing/quality to determine their own targets/process limits as long as its within the drawing spec limits. Maybe we have been talking past each other somewhat, but thats all I said from the beginning.

I do not believe it was intended to create a fundamental shift in the way drawings are created or interpreted

I don't either, besides from releasing manufacturing from any perceived manufacturing targets suggested by a nominal value. I didn't mean to give the impression that it did.

that's the conclusion one reaches if your line of thinking is followed to the end

Its not at all, though its difficult to explain without getting into the weeds of Y14.5 and Y14.5.1 minutiae - a whole other can of worms really. And I don't think it adds much to this particular discussion.

 

[SwinnyGG]

you said a nominal value tells manufacturing "where we want the center of the statistical distribution of feature sizes to land" and I said no, its up to manufacturing/quality to determine their own targets/process limits as long as its within the drawing spec limits. Maybe we have been talking past each other somewhat, but thats all I said from the beginning.

We are talking past each other, I think.

This is a gross oversimplification, but in simple terms:

We put a value on a drawing, and then tolerance that value with a bilateral tolerance (most of the time). That nominal value tells manufacturing what we think the ideal dimension is. So for a hole that's 4.000 +/- .010, if manufacturing and materials are perfect, every single hole will be exactly 4.000. We obviously know that's not going to happen, so the tolerance allows any hole between 4.010 and 3.990. So we have communicated to the manufacturer what we believe is ideal, and what we believe is allowable.

If the manufacturing guy wants to set his boring bar to cut a 4.009 diameter, and then run 24,000 parts such that tool wear at the end of the run results in holes that are 3.991, I have no issue with that. My tolerance has given 'permission' for that approach, but it has also not dictated that that approach is what must be done. I do not care. The adjustment made to the manufacturing process is still a black box to me, which is fine.

But maybe he's going to cut those holes with an EDM which is capable of extreme accuracy. Now he's going to command that EDM to cut a 4.001 hole, or whatever, and he's going to make a bunch of parts that are bang on 4.000.

By providing that nominal value, I'm telling the manufacturing guy what is ideal, and if he has means and methods to make a lot of parts that are ideal, he can do it. If he doesn't, the tolerance means he doesn't have to.

If I spec that hole as MAX 4.010 MIN 3.990, and he's using a process that is capable of way more than .020 accuracy at that diameter, where does he start? He doesn't know, because I have removed information from the drawing. All I have done by removing the nominal value is reduce the amount of communicated information, and make the job of interpreting the drawing a tiny bit more difficult. Multiply this by tens or hundreds of features and it adds up.

More complicated features make this whole interaction... more complicated, but the principle remains the same.

 

[jassco]

Hi, SwinnyGG:

Sorry! I have to disagree with you.

3.990/4.010 is identical to 4.000+/-.010.

Both of them means as follows:

LSL: 3.990
USL: 4.010

Best regards,

Alex

 

[chez311]

First off, not to be nitpicky (I guess I have been this whole time really) and I'm not sure if you meant that as a literal example of a drawing specification but MAX and MIN should not be used together like that. The only thing necessary for a limit dimension is the limits 4.010/3.990 would suffice. MAX and MIN have special meaning and should only be used with single limit tolerances.

I can understand arguing for using bilateral/symmetric tolerancing from an ease of readability standpoint (ie: since 99% of time a machinist will try and target the middle of the tolerance zone, so providing this value means less math is necessary 99% of the time even though they are NOT required to use this value), but it does not communicate what our "ideal" part looks like. At least not by Y14.5 definitions.

That nominal value tells manufacturing what we think the ideal dimension is.

Lets define what we mean by "ideal." Are you using it to mean something as simply theoretically exact, or do you mean it as "perfection or excellence" as in best satisfies design intent? If the former, we agree. If the latter, thats often subjective and also not always the case. First there are so many variables going into most designs that I doubt one could confidently say one dimension reflects the most perfect part taking into account all these variables. Secondly, I imagine theres plenty of times a drawing and corresponding model is adjusted to some value other than what the designer/engineer might consider "perfect" a simple case being the engineer desires a fit at MMC but the company's internal standard/standard practice requires all models be at the median of the tolerance zone.

I think you're right in that many people will read it the way you are saying, and that many shops reading a .500+.003/-.007 will likely target that .500 value even when a target of .498 would provide them more working tolerance. What I am saying is that they really should just be concerning themselves with getting the most parts within the tolerance zone - and that clause frees them to do exactly that. They can use the stated nominal as a suggestion if they like, but without specific instruction to do so it just (potentially) makes their job harder.

 

[SwinnyGG]

If the latter, thats often subjective and also not always the case

It's the latter.

With that said, in this type of argument, anyone is going to come up with an edge case which 'breaks' the rule. I acknowledge that there are unique cases where certain approaches need to change; that's the entire reason why Y14 is not contained on a single page.

But we're talking about basic, fundamental practices which apply most of the time.

What I am saying is that they really should just be concerning themselves with getting the most parts within the tolerance zone - and that clause frees them to do exactly that.

Agree with this 100%.

They can use the stated nominal as a suggestion if they like, but without specific instruction to do so it just (potentially) makes their job harder.

And disagree with this completely. I cannot imagine a scenario what a nominal value with a symmetric bilateral tolerance is more difficult to understand than any of the alternatives discussed in this thread.

While 'we've always done it that way' is almost never a valid reason to stay the course, 'after century or so of trying different things this method won out and became the standard' is, at least, informative.

If, as you seem to be saying, nominal values with symmetric bilateral tolerances make life more difficult, how did they 'win' against the alternatives after a century of humans drawing things on paper to hand to someone else to tell them what to make?

 

[chez311]

I didn't come up with an edge case. Theres typically many variables which designs have to balance - to say that one dimension perfectly satisfies them all is unlikely. Hence why I said this could be subjective - heck you've got someone just a few comments up (OP) that claims the part at the MMC limit is always the "ideal" part.

My other example was also not an edge case, I'm almost certain theres some companies out there that have an internal requirement to model at the median value so regardless of what the design target might be, or what the engineer/designer imagines is "ideal", those models and therefore the drawings (unless they manually edit the drawing value to not follow the model - which is usually a bad idea for a whole host of other reasons) will always be at the middle of the tolerance zone. Hence why I said "not always the case" - the specified nominal won't even line up with what the engineer/designer views as "ideal" in this totally plausible case.

If, as you seem to be saying, nominal values with symmetric bilateral tolerances make life more difficult

Again, I did not say that. I'm trying not to get frustrated, but you've claimed several times that I've said something when I clearly didn't and/or cherrypicked my statements without context. I clearly provided an example right before that statement that was unequal bilateral NOT symmetric and said that it POSSIBLY might make their life harder - not necessarily from a readability standpoint but that they're missing out on the wider available tolerance if they target the specified nominal. Obviously its not desirable for every process to target nominal and thats another reason why I said POSSIBLY. I clearly don't think that a symmetric/equal bilateral tolerance would usually make their job harder - I even said as much just above that!

since 99% of time a machinist will try and target the middle of the tolerance zone, so providing this value means less math is necessary 99% of the time

 

[SwinnyGG]

Dude. This:

a simple case being the engineer desires a fit at MMC but the company's internal standard/standard practice requires all models be at the median of the tolerance zone.

Is an edge case.

claimed several times that I've said something when I clearly didn't and/or cherrypicked my statements without context

I haven't done that. I've followed your logic to the end, in an attempt to illustrate why I believe you're not correct.

At this point I don't believe this conversation is productive, either personally or for the thread. At the end of the day neither of us agrees with OP, we're just arguing with each other about minutiae. I'm moving on. Good luck out there.

 

[chez311]

At this point I don't believe this conversation is productive, either personally or for the thread.

My thoughts exactly. Best of luck.