partial cone surface as datum

[cjccmc]

Hi all,

I'm the stuckee for the design of a conical shaped part and out of my league for determining suitable datums and tolerances. I attached a simplified pic that covers the main points for now. We work to 1994 ASME std and use Model Based Definition (MBD) which means means that you query the CAD model to get dimensions instead of listing them in 2D drafting style.

Main questions for now:

1) I picked the raised cone shaped surface on the outer periphery as datum A.

2) datum B is a flat surface at a skewed angle to the cone axis.

My reasoning for above is that a cone defines everything but clocking and any hole, surface, etc on part can provide the clocking. No need for a tertiary datum, you get a fixed frame of reference from cone and surface.

Are my datums and logic legit per 1994?

 

[MikeHalloran]

The MBD concept is new to me, and I'm having a hard time wrapping my head around the idea of applying tolerances to dimensions that aren't there.

I'm even more perplexed by the prospect of someday having to prepare an Inspection Report for an article that's (not)dimensioned in that way.

Fairly often, I've run into parts that would pass inspection if a dimension goes from <here> to <there>, but the same part would fail if the dimension went from <there> to <over there>, without regard to whether the part would fit or work.
In light of that, I'm thinking that you are, in effect, making the tolerances tighter than they appear to be, because the stated tolerance applies universally to any arbitrarily selected or conjectured dimension from a particular feature to _any_ other feature.

It just sounds _guaranteed_ to produce conflict between producer and inspector, because they will be trying to compare/contrast an infinity of potential actual dimensions from any one feature to every other feature.

I.e., if MBD was intended to save money, I'm failing to see how it does so. Maybe you save a bit by sending out undimensioned drawings, or maybe even no drawings, but I'd predict you'll lose much more money just talking on the phone, never mind the costs that rack up when you have to go to court.




Mike Halloran
Pembroke Pines, FL, USA

 

[cjccmc]

Our MBD datasets typically have a note that essentially says "all features shall be within +/- .xxx of the model unless otherwise noted". Then we add GD&T to features that need to be tighter and are too lazy to add GD&T to those that can be looser ;-). I like it much more than 2D drafting and dimensioning, reduces tedious work. I think it results in a better understanding by the mfg house when they have a 3D model to look at, especially on complex curvey parts that are hard to visualize from 2D views.

 

[MikeHalloran]

Granted, MBD saves _you_ some tedious drafting time; I'm curious about its affect over the component, product and company lifecycles.

Last night I remembered that I had seen an MBD dataset before; hot air ducts for a business jet. Compound curves everywhere, cross section circular only at the interfaces, etc. The tolerance was +/- half a mm. My shop at the time worked to the line from a dull Sharpie. I had to no-bid.

Later last night I figured out, for the first time, how it was possible to make such a part consistently. ... using a 3 axis CNC mill to hog it out fromm billet in two halves, then weld. It would have been something like 98 pct material removal. I think the aircraft company died before completing a plane.



Mike Halloran
Pembroke Pines, FL, USA

 

[PaulJackson]

cjccmc,

If you know how it assembles and what its function is... then you can determine whether the features on the inside of the conic section ultimately orient and locate the cone or vis-versa. It is always preferable that the datum feature selection and precedence replicate how orientations and locations are determined in subsequent assembly but it is also permissible to depart from that for the sake of measurement stability... given that the tolerance stack accumulation will be complicated by the non-functional substitution.

You said that the conic section constrains all but rotation (about the conic axis)... that is true but... as the cone angle decreases tending more cylindrical the apex becomes increasingly difficult to determine. Error in form, measurement, and error due to limited circumferential sampling may make locating the apex extremely difficult.

So have you captured the features that orient and locate the part functionally?
If the conic section does functionally stop two rotations and three translations is it capable of repeat ably stopping translation along the conic axis?

Inspectors commonly compare profiles of fabricated surfaces to modeled surfaced data... that is not the problem... the problem is how to begin... If the coordinate system is well defined, measurement is repeatable and variation can be mapped to its effect on function then the specification does its job.

 

[cjccmc]

Paul,

Yes, datum A is the primary functional feature, it is a sealing / mating interface for this hatch. The large tab for datum B also has a functional purpose, the mechanism that pulls this hatch into the closed postion attaches to it. I notice now that I did not word my OP very well. I know that my A and B have functional value, I am just not sure they make a practical and legal datum reference frame. I could not find any examples of cones being datums in 1994 std. Also seems that if I put A in second line of a composite callout that it would be same as putting both a primary and secondary datum.

 

[PaulJackson]

cjccmc,

What role (if any) does the perimeter of the hatch play in constraining its location... or does the attachment at actually constrain one rotation and two translations?

For your two hole pattern using [C] as primary... by using [C] as primary the small surface now replaces the two rotations and one translation that [A] was most capable of constraining and now leaves [A] with the remaining two translations and one rotation that it is least capable of arresting due to its limited area and its attitude relative to [C]... is that what you intended?

Basically all that your position specification for those two holes does well is control the spread between the holes and the orientation of the pattern to [C] all else I suspect will be difficult to replicate.

 

[cjccmc]

In actual usage the hatch is held in place by 15 latches equally spaced around the perimeter. What I showed in OP pic is a very simplified version of the real thing.

A mechanism mounts on C datum and is held in place by the two bolts in those holes.

 

[cjccmc]

Paul,

I think I missed the point of your comment:
"For your two hole pattern using [C] as primary... by using [C] as primary the small surface now replaces the two rotations and one translation that [A] was most capable of constraining and now leaves [A] with the remaining two translations and one rotation that it is least capable of arresting due to its limited area and its attitude relative to [C]... is that what you intended?

I never thought of it that way before, picking a secondary datum that has a physical orientation more effective in removing the remaining DOF. I just automatically default to using primary datum for that. In actual use neither the datum A or B features physically constrain the part that mounts to C datum, but if I understand your point, that's another matter entirely.

I was thinking that the cone shaped datum A would have two center planes the way the std shows it for a cylindrical feature. I think datum B would provide the missing clocking but not sure if there is one definitive datum ref frame since datum B is skewed to cone axis. I can envision the cone center planes being in orientations that are efficient in constraining the movement about C whereas the actual datum feature A is not. Which matters more I do not know, that's a new consideration for me.

 

[PaulJackson]

If you get the datum features right the rest is simple... tolerance stacks will tend to better predict deviation consequences. So from what I hear in your words the functional primary is indeed the raised perimeter surface of the conic section and it constrains only 3 of the five possible degrees-of-freedom the it theoretically is capable of. Those three mimic those that a flat plane is cabable of two rotations and one translation (perpendicular to the plane). The remaining three degrees of freedom are constrained by the pattern of 15 latches one rotation and two translations.

Tolerance the form of the primary , the pattern of the of the secondary [position], the of the perimeter to the primary and secondary, then the orientation and/or position of all other features to that (primary/secondary) coordinate system... using any local additional refinements as necessary to constrain fit/function.

Paul

 

[PaulJackson]

= ["profile"] less quotation marks