LM 4780 heatsinking issues help!

[sonicalleystudios]

So I got the audio sector boards and have set them up to run parallel, and have hassled Peter Daniels enough already so I thought I would call on you wise folks instead for this.

I decided to use the chassis sides to heatsink the chips just to test the circuits and make sure everything works. I was given what looks like mica transistor pads to use in the kit which doesnt seem to be proper so any suggestions for better options would be great. Anyway, the problem.. I put some artic silver heatsink grease in the 2 small pad hole just in case the back of the chip could conduct to ground through those. Then I attached the chip to the chassis. I went to test with a mm to see if v- had any continuity with the chassis right after, and it did. Then for the heck of it, I checked some other pins and they all seem to connect to ground??!

Did I screw up royally? Would the result be different if I powered up the amp circuit? I'm afraid to do that and blow something up!

Any advice would be great

 

[rayma]

I checked some other pins and they all seem to connect to ground??!

This device MUST be isolated from the heat sink, unless the heat sink itself is isolated from everything else.
Most power devices with a metal tab must be isolated from the heat sink (that is itself non-isolated).

From the data sheet: http://www.ti.com/lit/ds/symlink/lm4780.pdf
The NDM0027A is a non-isolated package. The package's metal back, and any heat sink to which it is mounted,
are connected to the V- potential when using only thermal compound. If a mica washer is used in addition to
thermal compound, θCS (case to sink) is increased, but the heat sink will be electrically isolated from V-.

 

[sonicalleystudios]

I put the mica isolation pad with added grease between the chip's back (metal) and the chassis (heatsink for intents and purposes) I thought that should isolate the chip...

 

[rayma]

I put the mica isolation pad with added grease between the chip's back (metal) and the chassis (heatsink for intents and purposes) I thought that should isolate the chip...

The back of the package is internally connected to its own negative supply terminal.
Does your circuit have only a single supply (V+ and ground)? If so, then there would
appear to be a short to ground, because the chassis and the package's V- would be
at the same potential (ground), though not directly shorted together.

 

[sonicalleystudios]

My circuit has a dual supply +/- 34vdc with common ground

 

[rayma]

My circuit has a dual supply +/- 34vdc with common ground

Can you post the schematic, and some clear photos?

 

[Michael Bean]

Here's a link to the ESP site with more information than you need. Pertinent info regarding physically mounting power devices to heat sinks starts about half-way down the page. http://sound.westhost.com/heatsinks.htm

Mike

 

[tomchr]

The metal back of the chip connects to VEE. As is the case with most semiconductor products. Thus, the back of the IC *must* be isolated from the heat sink if the heat sink is grounded or can be accessed by the user.

I'm guessing the thermal pad you have is for a TO-247, TO-3P, or maybe a TO-220 package. That's not going to work. You'll need something that provides good electrical isolation, good thermal conductivity, and fits the package. That turns out to be a tall order.

The best option is to get a sheet of SilPad 1500ST. You get an 8x12 inch sheet for about $30. Then you use a tiny corner of it and now you have a lot of SilPad and no use for it.
The other option is to use a thermal pad intended for use with a relay module. Digikey has them. Run a search for thermal pad and sort/select by size. Do beware that some of the relay pads are electrically conductive as well as thermally conductive. Those are no good. You need electrical isolation. I seem to recall finding exactly one pad that would do the trick.

Expect to spend a little time reading data sheets and checking dimensions on these thermal pads before you find the right one.

Tom

 

[sonicalleystudios]

Thanks for the info everyone, Tom the pads I received are for a to-3 style transistor. It must be that there was not enough isolation through the holes that the transistor pins would stick through even with thermal paste. I was looking at sil pad 400 and even requested a sample sheet from berquist. Is th 1500st stuff you quoted better? What I want is something I can exacto blade cut to the exact shape of the chip with the mounting holes in the correct places

 

[sonicalleystudios]

At the same time, since this test setup went wrong, I may just get the heatsinks set up and do it properly. I was looking at 3" x 5" or 7" ones from digikey but they are 32 and 40$ each. Does anyone have any other cheaper sources for something similar?

 

[tomchr]

Tom the pads I received are for a to-3 style transistor.

Yeah. That won't work.

Is th 1500st stuff you quoted better? What I want is something I can exacto blade cut to the exact shape of the chip with the mounting holes in the correct places

The 1500ST is the best I've been able to find. The SilPad 400 and 800 are good too.

At the same time, since this test setup went wrong, I may just get the heatsinks set up and do it properly. I was looking at 3" x 5" or 7" ones from digikey but they are 32 and 40$ each. Does anyone have any other cheaper sources for something similar?

Heatsink USA. Don't know what their shipping rates to Canada are, but one can find out.

Tom

 

[tomchr]

Maybe something like this would work.

[IMGDEAD]https://media.digikey.com/Photos/T-Global%20Technology%20Photos/DC0022%5E02-L37-3F-0.25-2A.JPG[/IMGDEAD]

Digikey P/N: 1168-1830-ND

Tom

 

[rabbitz]

Had a similar issue with the LM4780 and found that the edge of the mounting holes had some conductive metal protruding so mounting screw would touch and short. This was found with running a DMM around the holes and back plate.

That can be solved by using insulating bushes on the screws in conjunction with some heatshrink on the screw thread in this area. The other issue is it's only a part mounting hole so the screw doesn't sit flat without a thick washer. I ended up using a tie bar that clamps over the chip and keeps the screws square. I also used heatshrink on the screw thread for this method.

The material that Peter supplies for insulation is Kapton I think and on later builds I just used silicone insulators cut to shape. You can buy it in small sheets.

 

[Michael Bean]

At the same time, since this test setup went wrong, I may just get the heatsinks set up and do it properly. I was looking at 3" x 5" or 7" ones from digikey but they are 32 and 40$ each. Does anyone have any other cheaper sources for something similar?

Have a look here: HeatsinkUSA

Mike

 

[mondogenerator]

Toms suggestion seems the best here, unless you buy a set of Mica and cut to size (may be cheaper than silpads)

Buy some heatshrink tubing say 5mm diameter and use.this on the heating mounting screws to insulate them from the metal tab which protrudes into the mounting holes.

Don't forget the siliconE grease on each side of the insulator (I use Corning and ut works well)

Job done.

 

[sonicalleystudios]

I like the idea of a small clamping bar over the chip. Thanks for that.

As for the thermal pad, you'll never believe it, but when I made my first amp, I ordered a Hammond box from digikey and instead of the lid, they sent what I now have come to realize is a giant 8x 16" sheet of a2000 silicon fiberglass thermal pad!

 

[tomchr]

As for the thermal pad, you'll never believe it, but when I made my first amp, I ordered a Hammond box from digikey and instead of the lid, they sent what I now have come to realize is a giant 8x 16" sheet of a2000 silicon fiberglass thermal pad!

That's a pretty nice trade. The A2000 sheet is considerably more expensive than the Hammond lid.

Make sure to read Bergquist's instructions on how to use the stuff. You can find the instructions on their website. Some thermal pads have protective foils that you need to remove before you use it.

Tom

 

[sonicalleystudios]

Stuff worked like a charm. It's pretty simple stuff, just silicon infused with fiberglass, no foil or anything, just a plastic film on either side.

Amplifier sounds great on test!

Thanks for all the tips