Hex versus octagon

[gabimo]

One end of a part is hexagonal—left end: called datum feature A, dimensioned with basic angle, basic size across the flats, basic size across opposite corners and an all-around datumless profile.

The opposite end: right end: profile all around to A primary. (same as the left end: everything else is shown basic).

Questions:
How to relax the clocking between the hexagonal side of the part and octagonal one?
We have some arguments that SIM REQT added under the octagonal all-around profile will only brake the mutual relationship between the eight sides of the octagon, but won’t accomplish anything about relative rotation between hex and octagonal shape.

How to make the octogonal shape to freely rotate about its centroid?

 

[CheckerHater]

OK, I think we are paying too much attention to relatively simple issue, but I will try to answer.

First, The hex was never a datum in the OP:

This led me to the idea to look for datum somewhere else.

Second, you often see direct-toleranced dimensions (or not toleranced at all) in catalogs, but we can agree, those are not working drawings.

Now, direct-toleranced dim's with position tolerances. Yes, they will position flats wrt OD of the tool or axis of the bolt, etc. To position flats to each other, we have to use basic angles, just like we use basic coordinates to position our holes. I see no difference.

On the matter of angle, I don't care as long as it's basic, so there is no tolerance accumulation (I think we agreed that direct-toleranced angles have no use).
You may get away with using only (3) 60 deg. angles with (3) positions.
By the way, is "equally spaced" basic or directly toleranced?

Not everyone will agree, but I think 3 toleranced dimensions and 3 geo tolerances will imply more then one measurement.
And if we decide to make our hex into a datum, it will be acting as a pattern.
At the same time, applying profile to the entire hex will make it a single feature. Which is always comfortable.

So, there is always more than one way (as usual)




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

 

[Kedu]

CheckerHater,

Could you please show how the hex dimensioned with plus minus look like? Not sure I understand it.

 

[CheckerHater]

I do not suggest dimensioning "hex" by plus/minus. I agreed that "wrench size" could be dimensioned plus/minus, thus becoming FOS and located using basic dimensions (angles)

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

 

[drawoh]

CheckerHater,

The OP wants to locate as sloppily as possible, an octagonal feature to a hexagonal feature. The hex must be defined somehow as a datum.

--
JHG

 

[CheckerHater]

drawoh:
1. I'm afraid you haven't follow the discussion closely
2. Actually OP desired no rotational restriction at all between hex and octagon, as in "freely rotate".

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

 

[chez311]

While I don't necessarily agree that the hex MUST be a datum feature, it was clearly stated by gabimoang in the OP text.

One end of a part is hexagonal—left end: called datum feature A, dimensioned with basic angle, basic size across the flats, basic size across opposite corners and an all-around datumless profile.

That being said, I did state previously that holding both to an OD datum feature with SEP REQT to allow relative rotation was an alternate option, albeit one that maybe does not utilize what I assume are the mating/interfacing features as datum features. CH your comments about the nuances within the concept of "functional" bring up a good point - its definitely worth considering for the designer whether the added difficulty of inspecting/fixturing to a complex feature such as this and the added complexity of something like a customized DRF really adds anything to the design. If so then you have your answer - if not and the OD is sufficiently accurate and all the relevant features on the part can handle the additional stack-up by referencing the OD instead of the hex then I can see a case being made.