DOVETAIL DIMENSIONING 3

[KENAT]

I have a colleague that is designing both the male & female halves of a dovetail joint. The dovetail is used for a detachable optical assembly and variations there on have been used with adequate performance for some time at my employer. However since most of those were introduced the company has become more aware of tolerance/repeatability issues and is now using GD&T (ASME 14.5M-1994).

He has asked me how I would keep the angled edges in the same alignment, ie effectively keep the width of the dovetail the same to a fairly tight tolerance. The term parallel keeps getting used but isn't really correct.

My initial suggestion was just to control the surfaces with surface profile however he doesn't seem convinced by this. He is very concerned about keeping them aligned but doesn't seem as concerned by other factors. My other suggestion was to call out a fairly tight perpendicular tolerance to the end face but while this would help keep the edges aligned it isn't really necessary for the dovetail to be highly perpendicular to the end face.

Now I'm not convinced he's correct as once installed a screw pushes in from one side pushing the dovetail to the other, so I don't see why they need to be tightly controlled but he's convinced.

He wants to call out the machining process, saying they are to be machined in a single step but obviously this doesn't fit well with standard drawing conventions, I'd rather detail the end requirement not how to achieve it.

Any ideas appreciated, simplified sketch attatched to give some idea of the female.

KENAT, probably the least qualified checker you'll ever meet...

 

[KENAT]

Well I've pondered over the above, looked at the figures Ron listed and looked at parts of 14.5.

6.6.1.1 caught my eye and seems like "EACH ELEMENT" might apply to exactly this type of situation. Sadly both examples (6-44 & 6-55) are for cylindrical items but I think it can be made to work.

Take a look at the attatched sketch and see if you think I've got it right. I've put C as the secondary datum since only elements perpendicular to C need to be parallel, this is the part I'm least sure about.

Obviously it will need other dimensions/controls but I think this addresses the specific question my colleague asked.

So, what do ya reckon?

KENAT, probably the least qualified checker you'll ever meet...

 

[powerhound]

KENAT,

I breezed through the threads so forgive me if this has been answered but; Why the insistence on calling the two angled surfaces parallel? Drawoh brought it up but it was apparently ignored. A surface can only be parallel to one datum plane, not two. See 6.6.3. Only an axis can be parallel to two datum planes. Also, all the points on the surface must be equidistant from the datum plane. All the points on your specified surface are grossly NOT equidistant from neither datum plane D nor C.
I think you should use composite profile.

Powerhound, GDTP T-0419
Production Supervisor
Inventor 2008
Mastercam X2
Smartcam 11.1
SSG, U.S. Army
Taji, Iraq OIF II

 

[KENAT]

Powerhound, I mentioned in my OP "The term parallel keeps getting used but isn't really correct." I didn't ignore this fact, I'm well aware the surfaces aren't parallel. On the sketches I put parallel in "" to try and make it clear it wasn't really parallel. However, it has caused confusion to several so I'll try and learn from my mistake in future posts.

6.6.3 (d) a tolerance zone defined by two parallel lines parallel to a datum plane or axis, within which the line element of the surface must lie. See Fig. 6-45

It was this that I was trying to invoke. The line elements perpendicular to the end face C, should be parallel to surface D. However, looking at it again I'm even less confident that I'm correct.

KENAT, probably the least qualified checker you'll ever meet...

 

[drawoh]

KENAT,

There is no such thing as a "line element". Your face can be perpendicular to datum_C and located in profile from datums_C and_D. Angle specifications are valid from datum_C and_D, although they are redundant from_C if there is a perpendularity specification.

May we see the mechanism this is part of? This would all make much more sense if we understood what it was to do, and what it interfaced with. I would probably use the top face as datum_A, for example.

JHG

 

[KENAT]

Drawoh, there is such a thing as a line element, 6.6.3.1 says so and 6-45 shows an example.

However, I believe you're correct that for my situation it's not applicable/nonsensical. I was trying to apply it in a way that I don't think it's really intended to be.

Thanks for all your posts on this subject * for you.

The sketch is much simplified from the actual part, in fact I believe the design may be different from how I'd originially understood it so until I have all the details I don't know how much I can add here (or help him). For you specific point though, A mounts to a 'bridge' over the top of the tool. The optics assy has the male portion of the dovetail which then slides into this.

His basic question though was I suppose how to control the variation in width(/orientation?) of the dovetail along it's length. The actual width had to be controlled to say +-.005 (or equivalent) but he wanted the variation in width to be less. I'm still not sure I'm expressing myself clearly and my second sketch didn't help much.

Say the width is nominally 1" with the +-.005 tol. If one end of the dove tail is 1.005 then he didn't want the width at any other point on the dovetail to be less than say 1.001. However if the end was .095 then he didn't want the width at any point to be more than say .099. Does this make any sense?

His idea was to have a note saying that the female dovetail was to be machined in a single step by an inverted V shape cutter as this would be definition keep the width of each dovetail fairly constant to itself.

KENAT, probably the least qualified checker you'll ever meet...

 

[KENAT]

OK, I think I'm starting to get what a coupld of you put, actually read and started to understand 6.5.9 composite profile.

Sorry for being so slow and thanks for your patience.

Though that said, it would allow the dovtail to be located looser with respect to ABC and then the form of the dovetail itsel to be controlled more closely but I'm still not sure it directly addresses his concern.

KENAT, probably the least qualified checker you'll ever meet...

 

[MikeHalloran]

A note as you describe could cost you a lot of money.

Making a dovetail "one step to size" is not usually done, even in wood. Milling cutters, and I think dovetails especially, have problems with tearing material on exit, so it's customary to cut such a groove by entering from both ends.

Making it in one _clamping_ is possible, but still not guaranteed to get you what you want.

As you clearly understand, it's better to specify what you want, than to specify how to get there.



Mike Halloran
Pembroke Pines, FL, USA

 

[KENAT]

Thanks Mike you've helped confirm my concerns about trying to specify the machining step (which of course contravenes the standard anyway...)

He claimed he'd seen this from an optics supplier or something but it still doesn't sound like a great plan to me.

KENAT, probably the least qualified checker you'll ever meet...

 

[powerhound]

KENAT,
I think what you are trying to achieve is on page 168, FIG. 6-16. I think this is a better representation of the concept in 6.6.3(d) than FIG. 6-35. The fact that the linear elements are 2 dimensional allows then to be used with 2 datums, like an axis. The problem with your sketch is that the callout is in the wrong view.
Honestly I still think composite profile is your best solution but if what I've mentioned is what you were after, I can now see what you were trying to say.

Powerhound, GDTP T-0419
Production Supervisor
Inventor 2008
Mastercam X2
Smartcam 11.1
SSG, U.S. Army
Taji, Iraq OIF II

 

[KENAT]

Powerhound, thanks, I think that image is kind of showing what I was trying to achieve. I'll have to have a think about how exactly to show it clearly/unambiguously but I think that's it.

Thanks, star for you.

KENAT, probably the least qualified checker you'll ever meet...