GD&T bilateral profile - doubling the deviation

[[Je...]]

.

There's a post from 2018 about a similar topic, but I'd like a more condensed explanation.

Your task:

• Explain, in two sentences or less and 40 words or less, why the measured result of a bilateral profile is doubled when reported (min./max. version is not used).

Explain as though you had both engineers and machinists in the room, and everyone had to understand the answer.


Jeff Frodermann
Meier Tool & Engineering
Brooklyn Park, Minnesota

.

[[Ri...]]

Take your nominal Profile, and create parallel bands that squeezes down until it contains all of the deviations. The distance between that parallel band is your Profile result. The Profile result will always show as 2x the largest deviation.

[[Ri...]]

It's easier if you can show it was some illustrations.

[[Th...]]

I think now I would say "because ASME Y14.45 says so" and see what happens.

[[To...]]

I always try to take an approach from a different direction. Let's say the profile tolerance is .010". Do they understand that a point that is +.005 or -.005 is technically in tolerance. (work with me bro) Then, I try to get them to see that a point in either of these positions times 2 equals .010.

[[To...]]

Please sign in to view this quote.

Unfortunately, when they say "show me",..... oh crap! I'm late for a meeting.

[[To...]]

Please sign in to view this quote.

I don't like your statement because it sounds like the 2 parallel planes can squeeze to different amounts. I might add something like "the 2 planes must squeeze equally until either plane hits the first point." Like a self centering vise...lol

[[Ro...]]

Please sign in to view this quote.

Why? There's no logical reason that I know of other than ASME says to do so. ISO has it correct. That doubled value has absolutely zero analytical value, IMO. Min/Max actually gives you data you can use. My 2 cents and I didn't follow directions. 🤣

[[Ro...]]

As far as the technical reason why it's done, it's based on the same principle of True Position. The result is supposed to tell you the minimum tolerance zone needed for the feature to conform. That makes a lot more sense with True Position, IMO, because you're dealing with predominantly 2 axis and a cylindrical tolerance zone. When calculating assembly fits, that makes sense when you need to calculate for patterns, etc. For profiles, I just don't see it as logical. Profile is dealing with 1 vector offset direction. Doubling that result tells me nothing most of the time. Added to that is that it doesn't work for unequal distribution of a profile. The standard should cover all profiles, IMO.

[[Ro...]]

See graph

profile basics2.png

[[Ro...]]

Out of Tol

profile basics3.png

[[To...]]

My entire issue with this is that nowhere in ASME Y14.5 does it actual state how to calculate the profile, i.e. max deviation times 2 That would clear things up instantly, right?

[[Ro...]]

Hi Tom,
I agree. I've been told this is a legacy reporting thing based on the True Position theory I posted, but I don't know where it's at. Yet, Calypso and other software keeps insisting to report it as 2x the max deviation.

ASME Y14.5.1M, 6.5 tells you how to calculate Profile. It's says just the opposite of what Calypso and other software is doing per ASME. It makes zero sense to me.

"Note that no single actual value may be calculated for comparison to the tolerance value in the feature control frame, except in the case of unilateral profile tolerances."

Pretty clear to me. Yet, here we are.

[[He...]]

Please sign in to view this quote.

ASME Y14.45 describes how to calculate it. See attached picture.

Personally I think ASME Y14.45 (and Calypso) is doing the right and logical thing.
I like that when you have a tolerance of 0,10 and get an actual value of 0,12 the part is nonconforming. It is obvious because 0,12>0,10.

I do not like a situation where you have a tolerance of 0,10 and two actual vales of 0,06 and -0,01 that the part is nonconforming. If you don't know how profile works there is no reason for you to suspect that a value of 0,06 will be out of tolerance when the tolerance is 0,10 on the drawing. That is why I think it is logical that the software calculates the value as the double of the actual deviation. Otherwise the tolerance in the feature control frame should be written as ±0,05 instead.

ASME 14.45 9-1.PNG

[[Je...]]

.

Please sign in to view this quote.

.

I TOTALLY agree. Min/Max plays nicer with everybody. I'm requesting the advice because... "that's what the customer wants." 😡


Jeff Frodermann
Meier Tool & Engineering
Brooklyn Park, Minnesota

.

[[Ro...]]

Please sign in to view this quote.

I get what your saying, Henrik. The problem I have is that it doesn't make sense to double the result when you have a profile with unequal distribution. Calypso actually breaks the ASME rule if you report unequal distribution as 1 result, because otherwise it could report a number smaller than the tolerance zone size.

If you have a tolerance of 0.010 U .009 (+.009/-.001), for example, and your deviations are +.00076 & -.00179, reporting Min/Max give you those exact numbers. Reporting it as a single result gives you .01158. Calypso is taking the entire tolerance zone and making it Bilateral, which would give you a nominal location of +.004. Then it adds the negative deviation giving you .00579. Then it doubles it for a result of .01158.

What? 😕

The tolerance zone isn't bilateral. So it's breaking the rule because it would be ridiculous to report .00358 as a single result, which is 2x the greatest deviation from the nominal tolerance zone. 🤣

At the end of the day, an unequal tolerance zone is bilateral just by moving to the center of the zone. Which means you could make everything bilateral and just get rid of the silly unequal distribution concept all together. 😃

[[Je...]]

.

OP TAPS GAVEL VIGOROUSLY

POINT OF ORDER! Two sentence / 40 word max. breached. You do not get to double the word limit.

...unless the engineer revises the print ...or sends an email ...or slaps a sticky note on it ...or is on vacation, in which case you can do whatever you want.


Jeff Frodermann
Meier Tool & Engineering
Brooklyn Park, Minnesota

.

[[Ro...]]

[[Ri...]]

Please sign in to view this quote.

I think you are confusing nominal with true profile.

In a typical bilateral profile, nominal is the same as the true profile because the tolerance is equally disposed around true profile. On a unilateral or unequal distribution the nominal cannot be the same as the true profile because the tolerance is not equally disposed around the true profile.

If this were the case, with a unilateral tolerance zone if the linear deviation was say 0.001mm above the true profile then your profile would only be 0.001mm, but we know that cannot be true because you are actually approaching the edge of the tolerance zone, so would it not make more sense to display it as 0.099mm of profile?

Regarding unequal distribution, the principle would still be the same.

I do agree though that you should report min/max in all cases, but just remember that depending on what your deviations look like you could miss the part showing dangerously close to being out of specification by just looking at the min/max if the tolerance zone isn't bilateral to the true profile.

Having said all of that, I want to reiterate something that I've said multiple times - Profile is one of the most misused tolerance schemes today. It typically does not help someone manufacture a part. It was mainly meant for final inspection/assembly to tell if you if the part will function/work. Seeing a unilateral profile or unequal distribution on a manufacturing drawing/model is silly because the manufacturing person is going to attempt to make it true to the model/drawing, and if they don't pay close enough attention they can unintentionally manufacture parts out of specification.

It's like showing fit diameters on a drawing - say Ø50.00 +0.010/+0.030. The manufacturing guy will probably only look at the Ø50.00 and not the tolerance band.

Profile has great applications, especially in the aerospace world where there are undefined shapes, but it is so common to see it misused, or used in cases where the engineer was just being lazy.

[[Je...]]

Please sign in to view this quote.

.
.


Jeff Frodermann
Meier Tool & Engineering
Brooklyn Park, Minnesota


.

[[Ri...]]

Oh man. Why I'm glad we own a 3d scanner. Eat that up in no time.

On a CMM it is painful. And then you find out if they really care or not because they will just say, "oh just take a point here and there".

[[Ro...]]

Hi Richard,

Please sign in to view this quote.

I disagree. Nominal doesn't mean equally distributed. It's the exact dimension which we would like to achieve ideally. Otherwise, why would anybody put unequal distribution on a drawing, of any dimension type? Here's one snippet from Y14.5...

That's my understanding of nominal. If there's a document that defines nominal for a profile, or any dimension, differently, then I would be very interested in seeing it. I've certainly been wrong before. 🤣

[[Er...]]

OK, here is my stab... 😃

It's like true position where the tolerance is 1.0, or roughly +-0.5, and we double the radial distance from nominal to compare a diameter to a diameter. A profile of 1.0, or +-0.5, the largest distance from nominal has to be doubled.

[[To...]]

Please sign in to view this quote.

I think people often mistakenly interchange the terms mean and nominal.

[[Ri...]]

Please sign in to view this quote.

That's a good point. It appears that I might be misspoken.

I still think my argument stands that leaving "nominal" at the true profile on a unilateral or unequal distribution is silly, and would only make sense if looking at something in a pure min/max function. Once you try to report one value, is where it would get confusing. How do you know the difference if the deviation is 0.001mm from "nominal" and 0.099mm from "nominal" on a unilateral tolerance with a tolerance of 0.1mm (if you report only one number).