Seemingly incomplete tolerancing, reality check needed

[Nescius]

A customer's part will be milled from extruded bar stock. The profile of the extrusion is governed by a MIL standard. I believe the attached drawing of this profile dates from 1995. I am very interested to hear what some more experienced folks have to say about the GD&T used on this drawing.

In addition to your general thoughts regarding the tolerancing strategy, I'm keenly interested in the angled surfaces oriented with the basic 45 deg. dimensions. In my mind, these surfaces are free to pivot on the points of the inscribed rectangle. The small flats at the "points" of the angled surfaces would grow/shrink as the angle deviated from 45 deg. Likewise, the width of the top flat surface would grow/shrink.

Rightly or wrongly, I'm more hesitant than usual to call foul since this in an established government drawing. What say you?

 

[JNieman]

I would also not say the drawing is in error as there are probably MILES of 1913 rail working flawlessly across the world.

Bigger context of that clip you posted:

4.1 Dimensioning and tolerancinq. Dimensioning and
tolerancing of the accessory mounting rail for small arms shall
be in accordance with MIL-w-13855, ANSI Y14.5M (using inch-pound
units in lieu of metric.) and Figures 1 and 2,

Maybe some of the statements from MIL-W-13855D provide further control?

3.4.4.2 Concentricity and symmetry. Concentricity and symmetry shall be as specified
on the applicable drawings. Where concentricity or symmetry are not specified on drawings,
surfaces depicted as having a common centerline shall not be eccentric or unsymmetrical relative
to each other by more than one half of the sum of the differences between the actual measured
dimensions and the maximum material conditions specified by drawing dimensions. (Half the
sum of the differences represents half of the allowable TIR.) All surfaces depicted in the same
direction as the centerline without locational dimensions are considered to have a common
centerline.

....

3.4.6 Interchangeability. Unless otherwise specified on the drawings, all parts of
separable assemblies shall be interchangeable without filing or selective assembly.

I do see what you mean about the angular basic dimensions being under defined. However, I think it falls mostly under 'good practice' that they don't deviate from 45 too much, and because that's mostly a "stock" dimension controlled by the dies and manufacturing process of the extrusion. The design and nature of this mounting interface also allows for minor deviation.

Additionally, since no one expects to be able to place any precision optics on one firearm, zero it, and then place it on another firearm and have it be zeroed accurately, there is no recognizable failure due to lack of interchangeability. It just has to fit the mount on the optic base, which are somewhat forgiving. I'm reasonably sure the general open-ended verbiage in the couple MIL documents cover extreme deviations from nominal, though. I'd be interested in hearing from someone more familiar with manufacturing and quality control in the context of this item though.

 

[Nescius]

I agree on all, JNieman. However, many of those miles of 1913 rail do not conform to the specification. The mating mounts usually adapt just fine though.

I would be much happier if those basic 45's were directly toleranced instead. I'm responsible for providing the cross-section specs to the extruders and I believe I'm going to directly tolerance those angles.

 

[JNieman]

When I initially dug through the STD docs, I was hoping to find somewhere that specified a default angular tolerance, or specified a standard that states 'boxed' dimensions are to be treated as something more meaningful. Unfortunately not.

Now that I know what role you're in, I see the problem. You're in the one role where this drawing bites you in the rear! lol. What is a "stock" dimension to some, is a manufacturing/quality requirement for someone else...

 

[dgallup]

I for one don't see that as correctly dimensioned at all. I don't know how you determine where Datum C is, the .744+/-.002 dimension is supposed to apply at two heights .108 apart but there is no guarantee that these lie in the same plane. As mentioned, I see no control on the angles. Is there some additional drawing showing a gauge to check the conformance of the rail? Seems like one of those industry standards that's been around so long "everyone knows how to do it" but nobody really knows what the definition is.

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

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.

 

[Nescius]

dgallup, I struggle similarly. The parts that mate with these rails often register on the angled faces in the vicinity of the inscribed rectangle corners, almost like a datum target line. In that light, I can mostly interpret what they were trying to say. It falls apart if you look too closely though, in my mind at least.

I'm also having trouble understanding their reasoning behind using the MMB modifier on C in the two positional tolerance frames. The parts that mate with the rail CLAMP onto it, often on the angled faces. The features controlled by these positional tolerances have no mating function; they must simply not interfere with the apparatus clamped to the angled faces.

Also, the top "surface" has a flatness tolerance. I'm not sure how to interpret this on a 2D profile. If they had noted that this flatness tolerance be held over a given length of extrusion, I could understand...

I am not aware of any official gauge drawing.

 

[dtmbiz]

Nescius

Not sure what version this drawing refers to, however what I see is 'basically' hosed. (yes pun intended )

Datum feature C is not a feature of size.

The Basic dimensions do not originate from a Datum Feature.

Pretty much a free for all interpretation.

 

[Nescius]

I'm glad that I'm not the only one who thinks it's highly questionable.

For my drawing, I'm moving forward with an all-around profile of a line tolerance. The straightness and twist of the extrusion are controlled by themselves in notes, on a per foot basis.

 

[JNieman]

That sounds reasonable and practical.

Since the MIL-STD drawing shows the final product, not the raw material / extrusion, the flatness would (correct me if I'm wrong) only apply to individual flats between slots - not the continuous top surface comprised of n-surfaces.

 

[Nescius]

Well, the standard says this:

"3.5 Profile. In this document, profile refers to the cross
sectional outline as depicted in Figure 1 which establishes a
dimensional tolerance zone extending the length, width and depth
of the considered feature."

...and this:

"5.1 Accessorv mounting rail Profile. Standard dimensioning
and tolerancing for the profile of the accessory mounting rail
for small arms weapons shall be as shown in Figure 1."

...and this:

"5.3 Lenqth of accessorv mountina rail. The length of the
accessory mounting rail is dependent upon the particular
application of Small arms weapon it is to be mounted on."


So, I really don't know what the .005 flatness applies to. Nothing definitive, I suspect.

 

[Nescius]

Please forgive the typos in my previous post. I copied and pasted the text from a crude PDF of the MIL standard and it appears that things got a bit scrambled.

 

[JNieman]

If they mean the flatness applies to the length of the rail, then I know of a great many manufacturers that would be shocked to hear their 20" long rail isn't "mil spec" like they'd so boastfully purport.

 

[Nescius]

Indeed. "Mil-spec" has largely become a useless marketing buzzword. Case in point, the 1913 rail drawing itself was presumably to "mil-spec" and we can all see how great that turned out.

 

[powerhound]

I pretty much agree with teh others that teh drawing is bad. Datum feature C is bad. It looks like a picatinny rail. Knowing how they function, there are a multitude of features that could have been used more effectively than some imaginary lines.

John Acosta, GDTP Senior Level
Manufacturing Engineering Tech

 

[pylfrm]

I have been involved in the manufacturing and inspection of such rails in the past. I agree that the drawing seems somewhat lacking, but I'm not as familiar with the older drafting standards in effect so I won't pass judgement. Instead, I will offer the following possible interpretation:


The 4X 45° BASIC angles, .108 BASIC dimension, and .748 +/-.002 dimension all combine to create four tolerance zones that the angled surfaces must simultaneously fall within. This requirement could be equivalently defined by making the .748 dimension basic, and applying a profile of a surface tolerance of ~0.001414 to each of the four angled surfaces (as a simultaneous requirement and without datum feature references).

Datum feature C is an irregular feature of size composed of the four angled surfaces. The maximum material boundary is clearly defined by the profile of a surface tolerance interpretation already described.

The .164 +0 / -.020 requirement creates another tolerance zone in a similar manner. This requirement could be equivalently defined by changing the dimension to .154 BASIC, and applying a profile of a surface tolerance of .020 to the top surface. Whether it should reference datum feature C at MMB or RMB is not clear, but RMB would be the conservative choice.


I'm not saying this interpretation is correct, or even that a correct interpretation exists. However, I do feel that this approach is fairly conservative and has some basis in the standards.

See ASME Y14.5-2009 Fig. 2-21 for an example where basic angular and linear dimensions combine with a toleranced linear dimension to create what is essentially a profile of a surface tolerance.


All this still leaves the question of what length the tolerances apply over. Note that the .617 width and the lower two angled surfaces extend continuously for the entire length of the rail. Per MIL-STD-1913 paragraph 3.5, the tolerance zones for these features would extend the full length also. This includes half of datum feature C, so you could argue that the other half (upper two angled surfaces) should be included as well.

A useful compromise might be to control the entire profile continuously (ignoring the slots) for some intermediate length between the slot pitch and the full rail length.

- pylfrm

 

[Nescius]

pylfrm, thank you for taking the time to write that up. Since we know exactly how these rails function, I think your reasoning is defensible.

My extrusion consists of barely more than the angled surfaces themselves, so I was able to "cheat" and use an all-around profile tolerance, .0015 inches. More than the calculated .001414213..., but I'm comfortable with it. If my extrusion was more complex and the 1913 was a mere feature on the whole, .0015 would've been far too tight to apply all-around. In that case, I'd have to roll up the sleeves and undertake some deeper GD&T wrangling.