On Wed, 21 Mar 2012 14:04:08 -0400, Ed Huntress wrote:

On Wed, 21 Mar 2012 12:21:57 -0500, Tim Wescott
wrote:

So, for my day job I'm working on a circuit board design to replace
something which features a heatsink made from aluminum angle. It's 1" x
1" x 1/4", with nice square corners everywhere (radius 0.02").

Up to now I've been blithely assuming that this is an off-the-shelf item
that I can get anywhere -- but it looks like it may be harder to get
than that. Worse, I'd really like to extend the heat sink another 1/2"
or even 1" under the board, while keeping the outside leg at 1".

So I want to specify something that won't have their mechanical
engineers muttering under their breath _too_ much about @#$% EEs with
time on their hands...

McMaster carries aluminum angle, but it describes the inside corner and
the inside ends of the legs as "rounded", without saying what the radius
is. I can handle a radius on the inside corner, but that radius on the
leg takes away from area that I want touching my board. Furthermore,
McMaster only carries angle with even-length sides.

So my questions a

Is there any commonly-available aluminum angle that has corners that one
wouldn't describe as "rounded"? From who?

If I must go with rounded corners, can I expect that there is a
standard? What is it? Is there a place I might find it on the web?
(Machinery's Handbook doesn't seem to list anything like that).

Is there any commonly-available aluminum angle with uneven leg lengths?
I'm specifically looking for 1" x 1.5" x 1/4", or 1" x 2" x 1/4". From
who?

Any notion of how much it might cost to have a machine shop take a
larger angle and whack it down? These need machining anyway: they have
to be cut to length, then drilled on both webs and tapped on one -- so
it would be a case of "while it's in the machine anyway, make one or two
additional cuts". Precision is nearly nonexistent: +/- 0.05" would be
fine, and finish wouldn't be a huge issue: as long as the edges are
deburred and the cut side isn't so rough that it draws blood when
handled things would be fine; I would expect that a decent shop with
even a minimal sense of pride would insist on a much better finish than
necessary to get the job done.

For an application like that, note the variation in thermal conductivity
for different grades and hardnesses of aluminum. 2024 T4 is around 120
W/m-K. 1199-O is twice that: 240 W/m-K.

I found out about the hardness/temper issue the hard way, semi-annealing
a piece of 2024 to almost double its conductivity, only to find it drop
back to the original value a couple of weeks later.

Fortunately the thermal design is way over-specified -- but thank you for
the reminder; I'll make sure to take hardness (and self-hardening) into
account.

I was kind of thinking that hard enough for easy machining, but no
harder, was what I wanted -- now I just need to make sure to pick an
alloy that _stays_ that way.

--
My liberal friends think I'm a conservative kook.
My conservative friends think I'm a liberal kook.
Why am I not happy that they have found common ground?

Tim Wescott, Communications, Control, Circuits & Software
http://www.wescottdesign.com