ASME Y14.5-2009 - Limit Dimensioning vs. Plus and Minus Tolerancing 2

[rdjanssen]

2.2 DIRECT TOLERANCE METHODS

(a) Limit Dimensioning. The high limit is placed above the low limit.
(b) Plus and Minus Tolerancing. The dimension is given first and followed by a plus and minus expression of tolerance.

----------

2.4 INTERPRETATION OF LIMITS

All limits are absolute. Dimensional limits, regardless of the number of decimal places, are used as if they were continued with zeros.

----------

I understand that limits are absolute [2.4]. However, ASME Y14.5-2009 doesn't do a good job of defining what limits are. In section 2.2 it clearaly states that "Limit Dimensioning" indicates limits, but never calls "plus and minus tolerancing" limits. Are the upper and lower results of plus and minus tolerancing technically "limits".

My understanding was that limit dimensioning was fundamentally different than tolerance dimensioning.

Limit Dimensioning
6.0
5.0

Tolerance Dimensioning
5.5 +/-0.5

A feature measured at 6.00010 would NOT be acceptable for Limit Dimensioning since the upper limit 6.0 assumes zeros extended indefinitely. However, it WOULD be acceptable for the Tolerance Dimensioning because the tolerance dimensioning is limited to 2 significant digits. Rounding the measurement to 2 significant digits results in 6.0, so it would be good.

Is this bad logic? Are tolerance dimensions also defined as limits? If so, what is the advantage or significance of displaying one way over the other?

 

[CheckerHater]

ALL formats of dimensions are interpreted as if they had infinite number of zeros following "significant digits".

So, there is no difference

"For every expert there is an equal and opposite expert"
Arthur C. Clarke Profiles of the future

 

[looslib]

My take is that in theory 6.00010 would NOT pass in either case.
Just because you haven't displayed the value out to X decimal places, doesn't make it pass. The zeros do extend out on all dimensions.

For the practical side, if your part is dimensioned to a single decimal place tolerance, why are you inspecting that part to 4 place accuracy. The part would pass when inspected with a device capable of only 1 or 2 decimal place accuracy.


"Wildfires are dangerous, hard to control, and economically catastrophic."

Ben Loosli

 

[JNieman]

6.00010 would not pass either case.

Why you're measuring ten-thousandths when your tolerance is +/- .5 is beyond me, though. PRACTICALLY speaking, you'd likely never know that feature was anything other than 6.00 or 6.000, but sometimes, yes, you may be made aware - in which case it's out of tolerance.


ETA: That doesn't make it 'ok' to ignore that it's out of tolerance. Once aware, you'd be dishonest to pass it on as a good part. I just mean that one would likely never know. Using a 10,000th-precision micrometer to measure a part with +/- 1/2" tolerance would be asinine, for instance. A waste of time.

Rounding is never the difference between a good and bad part though. If it were, then the tolerance would have been +.54/-.55 and not +.5/-.5

 

[rdjanssen]

Hey, thanks for all the feed back.

Perhaps it's cleaner to state that the limits or dimension and tolerance set the absolute min and max (zeros to infinity), but that the significant digits on the print represent the precision to which it should be measured.

In the extreme case I listed, a part with an actual length of 6.00010 is out-of-spec, but if the tolerance were set as 5.5 +/-0.5 (or the high limit was 6.0) I would only expect the part to be measured to 2 significant digits, so the measurement would come back as 6.0 and be in spec.

I'm starting to see a circular argument here.

 

[JNieman]

It's not circular. Once you have discovered that it's actually 6.00010, it's a bad part.

It's just that you would have never known that if you used something with a 2-place resolution.

Once that can of worms is opened, and you discovered that it's 6.00010, however, it's done. Bad part.*** This kind of goes along with the phrase "Don't ask questions you don't want the answers to."

***according to every quality policy I've worked under.

 

[3DDave]

If you claim it's circular, then list the elements of the reasoning you think make it circular. As it is, the rule is that the dimension limit is the limit at which the part is acceptable. There is no allowance for saying parts beyond the limit are not beyond the limit.

I'm reminded of a guy who claimed that 6.049999 would be in tolerance because that's what rounding rules said.

I would like the OP and that guy to be in a room to argue their points.

 

[rdjanssen]

I think I misread something that led me to my circular comment. The statement was:

"For the practical side, if your part is dimensioned to a single decimal place tolerance, why are you inspecting that part to 4 place accuracy. The part would pass when inspected with a device capable of only 1 or 2 decimal place accuracy."

That seems to say the 6.00010 part is OK (i.e. in spec) when measured to only 1 or 2 decimal places. But it seems circular in that the part can't be measured to only 1 or 2 decimal places because ANY point above 6.0 (repeating zeros) is out of spec.

So, if I say that 6.0 is my maximum, my manufacturing department has to measure to as many decimal places as they can, because if it anything over 6.0 (with zeros to infinity) it is a bad part. Even if I say it is 5.5 +/-0.5, manufacturing STILL has to measure to 5 or 10 or 20 significant digits.

Maybe a simpler question would be: What's the difference between 5.5 +/-0.5 and 5.500 +/-0.500? Are these identical?

 

[JNieman]

Maybe a simpler question would be: What's the difference between 5.5 +/-0.5 and 5.500 +/-0.500? Are these identical?

By the book they have no difference. Their dimensional boundaries would be the same.

 

[rdjanssen]

Maybe a simpler question would be: What's the difference between 5.5 +/-0.5 and 5.500 +/-0.500? Are these identical?

By the book they have no difference. Their dimensional boundaries would be the same.

Awesome! That's kind of the response I was looking for.

And those two expressions are the same as the limits?:

6.0
5.0

or even

6
5
?

 

[Nescius]

So, if I say that 6.0 is my maximum, my manufacturing department has to measure to as many decimal places as they can, because if it anything over 6.0 (with zeros to infinity) it is a bad part. Even if I say it is 5.5 +/-0.5, manufacturing STILL has to measure to 5 or 10 or 20 significant digits.

That is jumping to a conclusion. Do you need to hire a surveyor to determine if your living room is located on your property? Of course not. You might hire that surveyor if you were planning on putting up a fence, though.

Following that line of thinking, consider that a set of go/no-go gauges can measure 5.5 +/-.5 just as well as 5.500000 +/-.500000.

 

[pylfrm]

rdjanssen,

Consider the following case: The part specification is 5.5 +/- 0.5, the gage resolution is 0.01, and the gage accuracy (maximum error) is 0.05. You measure a part, and the gage reads 5.97.

Does this mean the part is good? No, it only means it is potentially good. The part could be out of spec at 6.0001 or even 6.02. To confirm the part is good with this particular gage, it must measure between 5.05 and 5.95.


pylfrm

 

[rdjanssen]

So, if I say that 6.0 is my maximum, my manufacturing department has to measure to as many decimal places as they can, because if it anything over 6.0 (with zeros to infinity) it is a bad part. Even if I say it is 5.5 +/-0.5, manufacturing STILL has to measure to 5 or 10 or 20 significant digits.

That is jumping to a conclusion. Do you need to hire a surveyor to determine if your living room is located on your property? Of course not. You might hire that surveyor if you were planning on putting up a fence, though.

Following that line of thinking, consider that a set of go/no-go gauges can measure 5.5 +/-.5 just as well as 5.500000 +/-.500000.

Thanks, Nescius. Again, my whole question is just hypothetical and trying to make sure I understand Y14.5-2009 accurately. My biggest concern is that if I have a feature that is important and I put a reasonable tolerance on it, say 5.535" +/-.005". My manufacturer measures it to be 5.5400" and they tell me it is good. My customer buys the part and measures it to be 5.540045" and rejects the part because it is out of spec. So my corrective action is to increase the precision of my dimension to 5.53500" +/-.00500".

But, like we've discussed, this is not a real solution, because the precision on the limits doesn't matter. So instead I have to tell my manufacturer to measure the part "more accurately" because the customer is doing so.

If this is really what it is, that's fine. I was always under the impression that that was the purpose of the precision of the tolerance of a dimension, but it appears that is not the case. Is there a better way to state how precise a dimension should be measured?

 

[rdjanssen]

Consider the following case: The part specification is 5.5 +/- 0.5, the gage resolution is 0.01, and the gage accuracy (maximum error) is 0.05. You measure a part, and the gage reads 5.97.

Does this mean the part is good? No, it only means it is potentially good. The part could be out of spec at 6.0001 or even 6.02. To confirm the part is good with this particular gage, it must measure between 5.05 and 5.95.

I think I'm starting to get it now. If I say 5.5 +/-0.5, my manufacturer would have to measure this accounting for the error and precision of their gauge. So in your gauge example with a maximum error of 0.05 and a resolution of .01 I'd have to have a measurement of:

6.0 maximum limit
-0.005 gauge error (precision of 0.01 has a maximum error of 0.005)
------
5.995 maximum measurement to be in-spec. (And conversely 5.005 minimum measurement.)

 

[dgallup]

In general, the resolution of the gage must be less than or equal to 10% of the total tolerance. So in this case, with a tolerance of 1" the gage must read to 0.1" or better. I have never seen a case where there is a lower limit on the resolution of the gage. So if you want to use a gage that reads to 0.0001" that is fine.

Also, adding more redundant zeros to the end does not imply that a gage with a finer resolution must be used. If the tolerance is 1" or 1.000" the same gage that reads to 0.1" is acceptable. If you require something better than that you need a note specifying what is required.

In practice, I've never seen anyone argue whether or not something was in tolerance to 5.5"+/-0.5".

----------------------------------------

The Help for this program was created in Windows Help format, which depends on a feature that isn't included in this version of Windows.

 

[looslib]

"So, if I say that 6.0 is my maximum, my manufacturing department has to measure to as many decimal places as they can, because if it anything over 6.0 (with zeros to infinity) it is a bad part. Even if I say it is 5.5 +/-0.5, manufacturing STILL has to measure to 5 or 10 or 20 significant digits."

There is a cost associated with precision measuring of parts. Is it worth the extra cost of using a CMM which will give 10+ digits of accuracy for a part that is dimensioned at 5.5 +/- .5? A caliper will perform the inspection within the limits of your dimensioning scheme. There is no sense in measuring beyond that which is required, within tolerance limits of the inspection device.



"Wildfires are dangerous, hard to control, and economically catastrophic."

Ben Loosli

 

[3DDave]

I'd say that the inspection equipment needs to have 10% or less error than the desired approach to a limit, regardless of the number of places/tolerance on the dimension. And that inspection limit should be reduced by the expected error in the measuring equipment.

 

[pylfrm]

rdjanssen,

For a gage with a digital readout, half the resolution only provides a lower bound for the maximum error. In reality all gages are somewhat worse. Exactly how much worse can vary quite a bit, which is why I provided a value in my example. I stand by my original numbers, not the the 5.005 - 5.995 range you posted.

You shouldn't have to do anything special to keep your manufacturer from passing bad parts that are just barely out of tolerance. If you need measurements of a particular accuracy or resolution, you'd have to spell that out in your inspection requirements. That would typically be on a separate document from the drawing.


pylfrm