LM3886 & Others Chip Amp Heatsink Clamp Solution

[MartyM]

Hi guys I think I found an easy way to deal with clamping amp chips to just about any flat heatsink.

I recently picked up some more "mending plates" which I use for making my own custom car stereo installation brackets.

They're commonly found at hardware stores & even Wal-Mart.

All I had to do is drill a hole in the center of the plate and into the heatsink and *bam* finished. Really easy!

It's a great IC clamp with no screw insulator needed and is quite strong. (see photos)

Hope this helps you! Just trying to repay the great help I got with ground loop noise on this forum recently...

[gni]

Looks like it does clamp the chips to the heatsink. Thermal compound?
I would condsider going one step further: longer bars and bolt down
in three places to equalize the pressure on the chips. . .but certainly
a good idea. . .cheap too.

[MartyM]

There are thermal pads behind the ICs & thermal compound to be ordered.

It would be a little bit of overkill to bolt the ICs in three locations...the ICs are already almost 100% cover by the brace, and more than sufficient in strength (I leave a little bit of play for the ICs mouting position to allow it to flex and mate to the heatsink).

[lgreen]

Originally posted by MartyM
There are thermal pads behind the ICs & thermal compound to be ordered.

It would be a little bit of overkill to bolt the ICs in three locations...

Not really, I've done it and you get better pressure, the real problem is that the caps get in the way and you can never unmount the chips. You have to put the caps on the other side of the board which may really mess up the mounting situation.

[MartyM]

I appreciate your post but it's overkill. Why? Because (in this case) the bracket doesn't flex and the pressure is distributed equally across the ICs. Hence, assuming the mounting screw is sufficient, one is enough.

I'd be happy to show commercial examples if needed.

[glennb]

Originally posted by MartyM
I appreciate your post but it's overkill. Why? Because (in this case) the bracket doesn't flex and the pressure is distributed equally across the ICs. Hence, assuming the mounting screw is sufficient, one is enough. I'd be happy to show commercial examples if needed.

The bracket will probably flex a little, unless it is very very thick, or you don't do up the bolt very tight. How thick is it?

Even if it flexes a tiny fraction of a millimetre, the pressure on each chip will be uneven.

I'd like to see commercial examples of this T-bar technique across a comparable width. They probably use concave brackets or brackets made of spring steel.

A test is put a very thin smearing of thermal compound on the chips and mount them using your method. Undo everything. If the appearance of the thermal compound on the chip and heatsink doesn't look the same over the entire surface area, you are not getting even pressure.

Also, the Stanley plates are punched from sheet steel and then zinc plated. The punching and plating may raise the surface slightly near the edges of the plate or slightly deform the flatness of the plate. You need to carefully file the back surface so that it is totally flat before using it to clamp chips.

[MartyM]

Thanks for the detailed reply. Yes the commercial versions I've seend did have a bar 2x the thickness, now that I think about it.

I see your point and will work on it...maybe stack 2 of the mending plates, or use a longer version of the same and use multiple machine screws/nuts/washers instead of one heatsink tap/screw only.

Thanks everybody...

[AndrewT]

Hi,
the bees' knees are belleville washers under the screw heads to maintain nearly constant force as increasing temperatures cause differential expansion.
I wish I knew where in the UK, to buy DIY quantities in 3, 3.5 & 4mm bore.