Heatsink Flatness Specification

[JaPaBo]

Hello All,

I'm hoping for some input on flatness specifications for extruded heatsinks (typically with a machined mounting surface) for electronics applications. Module vendors will recommended flatness over a length of 100µm (e.g.,

http://www.infineon.com/dgdl/Infine...n.pdf?fileId=db3a30432a651cf2012a79e0c9042d3c

) or per inch (e.g.,

http://www.pwrx.com/pwrx/app/mount pwr modu.pdf).

The heatsink vendors will tend to have either a total flatness spec for the whole extrusion face or on limited basis (e.g., table B of

http://www.aavid.com/product-group/extrusions-na/dimtol).

My impression is that modern drawings using GD&T methods call out flatness for the total surface (e.g.,

http://www.gdandtbasics.com/flatness/).

My question: is there a straightforward way to equate these callout methods or to scale the flatness for differing dimensions? I'm a little confused on how to consider this when accounting for tolerance build-up, and this doesn't seem like a straightforward unit conversion exercise. If a surface met the 0.001 in/in requirement, it would seem to easily meet 0.002 in/2 in. More concretely- What would 0.001 in/in equate to in µm/100µm? What's the total flatness of a 6" by 12" plate with flatness of 0.001 in/in? Maybe I'm overthinking this?

Many thanks!
-Jack

 

[dgallup]

It is permissible to specify total flatness, per unit flatness or both. Converting between imperial and metric dimensions is simple unit conversion like any other dimension. Note that the denominator of a flatness per unit specification must be an area (i.e. square or circle), not a straight line. Thus 0.001 in/in is incorrect, that would be a straightness specification. It should be 0.001 inch per 1x1 inch.

It is somewhat difficult to compare a total flatness and unit flatness, you have to work out the worst case surface allowed by the per unit tolerance. It will be more than just the flatness per unit multiplied by the size. Hopefully someone can dig up an example, I don't have one handy.

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

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.

 

[JNieman]

Y14.5-2009
Section 5.4.2.2

 

[3DDave]

This may be of use:

http://www.alexandriaindustries.com/technicalguides/dimensional-tolerances.pdf

It appears to be part of the Aluminum Extruders Council publication available at

http://www.aec.org/?page=lib_designmanualjump

This article is also interesting - different style of heatsink, but interesting

http://www.aec.org/?page=lib_almagledheatsink

You may also come across ANSI H35.2, bur I think it does not apply to custom aluminum shapes; instead to industrial standards.

 

[3DDave]

Flatness per unit length is setting the curvature of the surface if it accumulates in a constant fashion. Go far enough and a uniform deviation will produce a complete circle.

I think the result is not linear - that a delta/length for a give length <> 2delta/2length.

The proof is that one can take a circle and divide it into 4 equal arcs. Pick one vertex as an origin and then draw a line from the origin to each of the other vertices. Use that as the baseline and measure the deviation to the circle. The first is a small amount; the second delta will be the radius of the circle; and the third is almost the diameter of the circle.

It is likely the effect is less pronounced on smaller amounts of the arc, but I would not be surprised to see that it is noticeable.

 

[JaPaBo]

I just wanted to thank everyone for the replies to date- all useful perspectives. I'm currently working on getting a draft of Y14.5-2009 to see what that turns up.