Question regarding Datum and tolerances on it as well as feature

[ryandias]

I have a heat sink with 3 plateaus. I used a Flattness call out of 0.30mm and named that my Datum A.

Now I have additional features which I want to have a tighter tolerance of 0.02mm wrt Datum A.

Does the feature tolerance become insignificant because Datum A has a larger tolerance? Or does the feature tolerance become "with respect to" whatever plane datam A lies on?

 

[ewh]

If you want it parallel within .02 to Datum A (which is flat within .30), that means it may be flat to within .32; Is this really what you want?

Believe it if you need it or leave it if you dare. - [small]Robert Hunter[/small]

 

[ringman]

EWH,

Are you sure of that or might you want to rethink it?

If the Standard applied it Y14.5, I do believe that the referenced surface will have to be flat within .02 to satisfy the parallel requirement of .02.

 

[MikeHalloran]

I thought a Datum Plane was a perfectly flat surface upon which your potato chip rests while it's being measured, not an imperfectly planar feature of the article itself.



Mike Halloran
Pembroke Pines, FL, USA

 

[KENAT]

ryandis if I understand correctly your 3 'plateau' can't really use flatness if ASME Y14.5-1994 is in effect, I believe they need to be surface profile see paragraph 6.5.6.1 & figures 6-20 & 6-21.

We had a similar situation recently and I put together the attatched. It's a compiled extract of ASME Y14.5, I hope I'm not massively breaking any copyright issues, if so administrator please delete the link or whatever works.

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

 

[weavedreamer]

The flatness of -A- refers to the 3 'datum targets?' designated -A-. Datum -A- may still require additional definition, as flatness does not establish orientation; unless the 3 'datum targets' are clearly co-planar.

The refinement of 0.02mm would be confirmed while setting on the 'high points' of datum -A-.

 

[Xplicator]

Correct me if I'm wrong, but I don't believe you mentioned targets at all. The three plateaus as you refer to them, should be called out using a Surface Profile of 0.3 without a referenced to any datum and a note under the FCF indicating application to the three planes. This will control flatness and co-planarity of the three datum features. The FOS dim/tol defining the additional features you want to control, probably allows for more than 0.02 in orientation and form. The refinement is the orientation callout with respect to datum A, not to mention that it also controls form. It is not accumulative at this point with the geo tol on the datum and should be acceptable, since the measurement originates from the datum not the datum features. The only thing to be concerned with is validating the Actual Local Size of the features first and I would also make sure to lock the part unambiguously in the Six DOF with a tertiary for R&R.

Hope this clears things up for you all! Additionally, I really have to question the .3mm form on a datum (foundation)when other features w.r.t. are specified to less than 10% of that. Seems like you could be asking for trouble. Upon performing a stack analysis on the assembly this may be also show up as a contributor.

I would advise evening them out a bit and if any thing make the Datum form tolerance tighter.

 

[KENAT]

Xplicator, that's more or less what I was thinking with regard to defining -A-. Not sure where the 'datum targets' came into it.

However weavedreamers last paragraph matches my understanding.

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

 

[weavedreamer]

Just picturing 3 plateaus. While datum targets were not mentioned, it is that imagary which prompted the suggestion with the initial question mark ie: 'datum targets?'.

 

[dingy2]

Flatness cannot apply to 3 plateaus simultaneously but only on each one separately. As Kenat stated, the use of a profile of a surface is most appropriate here. Once the surfaces meet the profile tolerance. they become datum A and are assumed perfect.

Now that we have established datum A, is it practical to have parallelism (assumed) of another surface 0.02 to datum A? It depends on how large of an area that the other surface covers. It would not be practical to have the relatively tight tolerance if the other surface is the same size as datum A. If the other surface is rather small in comparison to datum A, then maybe 0.02 would be appropriate.

I might suggest that datum targets are placed on datum A to at least give some sort of consistency of measurement.

Dave D.

http://www.qmsi.ca

 

[ewh]

OK, I was shooting from the hip...
I re-thought my post just after I submitted it, but was called away to fight other fires, and wasn't able to re-address it until now.
I agree that the actual Datum A is a perfect plane established by the three highest points on the surface, and any feature referring to A is actually referring to that plane.

Believe it if you need it or leave it if you dare. - [small]Robert Hunter[/small]

 

[dgallup]

Flatness can not have a datum reference, so flatness can not be used for the secondary surfaces. The flatness of the surface used to establish Datum A is irrelevant, Datum A will be perfectly flat. The other surfaces should be controlled by parallelism to Datum A since they are flat. Profile can be used but in my experience it will only confuse the issue since profile is just a general case of the parallelism concept extended to non planer surfaces. No need for the more complex definition when the simpler parallelism will suffice for a flat surface.

 

[axym]

Some important issues have been raised, but I'll put my two cents in regarding the original question.

As a general statement, the feature tolerance does not become insignificant when the datum feature has a larger tolerance. If Datum Feature A has a flatness tolerance of 0.30, the considered feature can still have a parallelism tolerance of 0.02. There is not really any relationship between the two numbers.

As has been pointed out, the datum is a perfect plane that contacts the high points of the datum feature (i.e. the tangent plane, physically represented by a surface plate).

The form of the datum feature becomes significant when trying to establish the datum. If the datum feature is convex, the part will rock on the surface plate (have three-point contact in more than one way). There will be more than one tangent plane, and hence more than one possible datum (these are called "candidate datums" in ASME Y14.5.1). The parallelism tolerance zone is defined to be parallel to the datum, so its relationship to the considered feature changes as the part is rocked.

The form of the datum feature is often controlled in an effort to minimize the possible extent of rocking and datum instability. Alternatively, datum targets are specified. The part would rest on three pins or tooling balls instead of a surface plate. This allows a very stable and repeatable datum to be established from a very imperfect datum feature. So you could have a datum feature that was flat within 0.5" and machine a surface that was parallel to it within 0.0001".



Evan Janeshewski

Axymetrix Quality Engineering Inc.

http://www.axymetrix.ca

 

[KENAT]

dgallup, the point of the profile was that the OP appears to want to have multiple surfaces "flat" to one another. This is not done by flatness but by surface profile.

For the surface being controlled to the tighter tolerance I agree with others that parallelism wrt -A- is probably the way to go.

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

 

[drawoh]

ryandias,

Your primary plane datum is defined by ASME Y14.5M-1994 as the three points of your surface that make contact with a flat plane. I am not sure what happens if your face is convex.

Given the above definition, and a datum flatness of 0.3mm, you can apply a profile tolerance of 0.04mm (+/-0.02mm) to a surface, with respect to the datum. It makes sense to me, and it is inspectable.

Whether or not this is good design practise, is another matter. The geometry is a little weird, and it may not be what you think it is. What are you trying to accomplish?

JHG

 

[ryandias]

I have a heat sink. Three pads will dissipate heat created on a PCB. The PCB must be isolated from the heat sink. Hense we use small islands that extend 0.07mm above the pads. The function of these islands is to create the "bond line" and ensure it does not decrease to allow shorting between the PCB and Heat sink.

I am not terribly picky about where my 3 pads are with regard to height. I am more concerned that they are all coplaner. The 3 gap setting islands on the other hand, I have to hold a very tight tolerance on since the dimension itself is 0.07mm. << hense the 0.02mm tolerance.

I'm not sure how "right" this is with regard to GD&T, but after speaking to a collegue about it, he understood what I wanted from the drawing. I am going to try to chat with the supplier to ensure they know what " I want..."

Thanks for all the answers >> a lot just confused me more, but most made sense.

The whole problem pretty much arrises because I do not want to put any tight tolerances where they are not necessarily needed. I would like to get away with no machining on a Cast part. Weather this works out or not...

 

[KENAT]

Sorry if my bringing up profile instead of flatness caused confusion.

Attatched a very simple sketch showing what I meant.

By the sounds of your last post too simple to be much help.

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

 

[ryandias]

KENAT >> yes your illustration could apply to my case. Does your illustration make sense wrt GD&T guidelines?


Basically once the datum feature is created, the tolerance for other features are independant of the datum tolerance?

 

[KENAT]

Except for my deliberate mistake (3 not 2 surfaces), yes as I understand it my example complies with ASME Y14.5M-1994, if it doesn't others here will I'm sure point it out.

See the attatchment to my first post above which explain parallellism in this context.

Attatched is a corrected sketch.

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