Width variation of a feature

[Burunduk]

Please consider a design intent where the variation on the width of a feature of size is more important than the measured value itself. For example, a part feature needs to be of width 4.1mm Max and 3.9mm Min, but the variation on that feature width can only be within 0.07. For example, if the width along the actual produced feature varies between 4.03 and 4.1 the part is acceptable. But If the width varies between 3.95 and 4.05 the part should be rejected as the variation is within 0.1 which is more than 0.07.

How would you specify this requirement in accordance with ASME Y14.5 (any revision year)? Would you attempt to achieve this requirement by a geometrical control or combination of controls (size and parallelism? Some sort of profile? Size and form tolerances? Size and runout tolerances?) Or would you specify 4+-0.1 and write a custom note describing the variation requirement?

 

[Belanger]

Using the newest standard (Y14.5-2018), the "dynamic profile" concept might be a good fit for you. Can't give a detailed reply right now, but that's my initial thought...

John-Paul Belanger
Certified Sr. GD&T Professional
Geometric Learning Systems

 

[Hallec]

Sounds like you might want to include a geometric control that uses two parallel planes as the tolerance zone?

I'm not a vegetarian because I dislike meat... I'm a vegetarian because I HATE PLANTS!!

 

[chez311]

Theres might be a few different ways to skin this cat, but I'm with JP - 2019's new dynamic profile concept seems like a good fit. If the two planar surfaces could be held parallel to a common DRF then that could be an option as well.

 

[greenimi]

I am not really sure if this discussion helps, but as far as I remember pmarc came up with some good solution.

https://www.eng-tips.com/viewthread.cfm?qid=400008

Maybe you can read it and adapt it to your design intent.

 

[dgallup]

Make one side a datum with a straightness requirement and put a parallelism requirement on the other side. That is unless you are OK with something that can wiggle like a piece of spaghetti.

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

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.

 

[Burunduk]

Thanks everyone for the suggestions. It is all very helpful - both the dynamic profile and dgallup's suggestion.
It is currently intended that this requirement will be checked by simply taking two-point size measurements on the feature and looking for the largest difference in the measured local sizes. If dynamic profile is applied, can it still be measured that way? My gut feeling tells me no, as profile is measured by checking if the entire surfaces falls within a tolerance zone (which is fixed, or the new "dynamic" one, or probably in most cases where the dynamic applied both the fixed TZ and dynamic TZ are needed as in the example provided be chez311? ). Two point measurements have little to do with fitting surfaces into tolerance zones. Correct? It will probably not control the form of the feature as tightly as the dynamic profile prescribes. In fact two point measurements don't control form at all and a separate dedicated check for rule #1 must be performed. Am I right with this assesment?

 

[chez311]

Burunduk,

Both of those suggestions will indeed control form*, if literally all you need to do is require that all two point measurements must lie within a certain allowed variation then I would just put a note that says as much ie "DIFFERENCE BETWEEN ANY TWO SIZE MEASUREMENTS MAY BE NO GREATER THAN (X)" or "VARIATION OF ALL TWO POINT SIZE MEASUREMENTS MUST BE WITHIN (X)" or something to that effect. I don't know that any single geometrical control will do that for you without controlling form. Additionally I would either include the Independency (I) symbol or a note saying "PERFECT FORM AT MMC NOT REQUIRED" as it sounds like you want no control over form** other than what controlling the variation in size measurements will do.


*dgallup - are you implying that straightness/flatness on one side and a parallelism will not control form?

**Edit - perhaps not, as you noted rule #1 would require a separate check - its up to you whether the design requires this.

 

[dgallup]

Straightness on one side and parallelism on the other WILL control form but the OP has not indicated that form is important, only width which is a 2 point measurement. The reasoning behind the importance of constant width is not given.

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

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.

 

[chez311]

dgallup,

Just realized I missed the key word "unless" in your initial comment. Carry on.

 

[Burunduk]

chez311,
Thanks, I suppose that a note like you suggested is the most probable way to deal with it, unless it is agreed by the people involved that a tighter form control needed and other inspection methods should be applied. It's not really a part that is subjected to free form variation to the point that rule #1 can be dismissed, but it is not a very rigid part either. Right now they just want two-point measurements for checking both the limits of size and "size variation" along the feature. In addition, I'm going to push for rule #1 control. In the end, the part will be allowed to "wiggle like a piece of spaghetti" as dgallup described, as long as that piece of spaghetti doesn't violate the MMC boundary and is within limits of size.

 

[drawoh]

Burunduk,

Your problem sounds a lot to me like parallelism.

--
JHG

 

[CheckerHater]

Even more like now explicitly forbidden kind of parallelism:

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

 

[chez311]

In the end, the part will be allowed to "wiggle like a piece of spaghetti" as dgallup described, as long as that piece of spaghetti doesn't violate the MMC boundary and is within limits of size.

CH/drawoh,
That doesn't really sound like parallelism as parallelism would certainly control form within the desired range. OP has said they desire no control over form besides what rule #1 will provide.

 

[Burunduk]

drawoh, CheckerHater,
I try to provide some help to the designers with the drawing. From talking to them it seems that they mainly care about width variation/ thickness uniformity, whatever you prefer to call it. I conclude that parallelism is not the best representation of this design intent but I might be wrong. For example, a wavy part can be very uniform in thickness but fail parallelism inspection when being put on a granite slab on one side and running a dial indicator on the other side. Your thoughts?

CheckerHater, out of curiosity, about that past practice of parallelism without datum - how was that inspected?

 

[CheckerHater]

It was not "without datum" - it was just irrelevant, which side to chose as a datum.

I guess it also introduced some ambiguity - is it enough if it satisfied on one side, or it must work on both sides? (Which may explain it becoming obsolete)

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

 

[chez311]

I conclude that parallelism is not the best representation of this design intent but I might be wrong. For example, a wavy part can be very uniform in thickness but fail parallelism inspection when being put on a granite slab on one side and running a dial indicator on the other side.

For what its worth, I agree - hence my previous statement.

 

[chez311]

CH,

It occurs to me that something like your datumless parallelism could possibly be achieved with dynamic profile. That is, if its allowed to be applied to non-continuous surfaces (ie: two separate, parallel sides), which is possible with a standard profile tolerance (4-22 in 2009 and 7-27 in 2018).

 

[CheckerHater]

If we are looking at something like this:

I don't think there is standard solution. Probably have to resort to verbal description.

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

 

[chez311]

CH,

I believe thats exactly what we're looking at, hence why I previously provided OP with a few options for custom notes that could be applied.

FYI my comment about dynamic profile was sort of tangential to the discussion - it would not fit OP's application, it was only meant to be a response to the fact that as you stated the "datumless parallelism" is no longer explicitly allowed however may be possible in another form with the introduction of dynamic profile in 2018.