Heatsink problem

I have some equipment which uses a power amplifier (15 lead TO-220) mounted on one of these on a PCB...

formatting link

No problems until now, it runs cool because the duty cycle is low, in fact it would be fine without a heat sink, I just need the power for a few percent of the time.

However, a new customer wants to use the equipment in an unusual manner such that the duty cycle for the amplifier is high, above 50%. Consequently, it gets hot and goes in and out of thermal limit. It's not far off working and takes several minutes or tens of minutes to manifest the problem.

I can't spare much extra height, don't want to change the board, fans are ugly.

So I'm thinking a right angle of aluminium with a large horizontal surface area and a thick vertical tab for the amplifier. Or copper. Or maybe leaving the heat sink as it is and screwing a large flat plate on top (the heatsink is the highest thing on the PCB). Heat pipe?

Just casting around for ideas.

Cheers

--
Clive
Reply to
Clive Arthur
Loading thread data ...

t's

r. Or

Hi Clive!

Have you room for a larger heat sink -

formatting link
? Or a small fan?

John :-#)# flippers.com

Reply to
John Robertson

Is it in a closed box? Maybe just some air holes would help. (Does it work w/o the box top on?) If it's an aluminum box, can you make contact with the surface?

How many are they going to buy? Just wondering if you could stick the amp on the side of the box and air wire.

George H.

Reply to
George Herold

Is the board mounted vertically or horizontally? What is above it? If the sink is hemmed in above and below, you won't get too much natural convection. If the board is mounted vertically, you want the fins to go the other way.

Increasing the volume of the finned part will help.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
 Click to see the full signature
Reply to
Phil Hobbs

Are you using an insulator? The sil-pads are terrible.

The thick large-area right-angle aluminum thing is good. Copper probably wouldn't help much, and it's a pain to machine.

A gap-pad can conduct heat from the aluminum thing, and/or the bottom of the PCB, to an enclosure. The heat has to get out to the world somehow.

Spread-and-insulate is better than insulate-and-spread.

What IC is it? How many watts do you need to dump?

--
John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  
 Click to see the full signature
Reply to
John Larkin

Those are designed to be mounted vertically so if it is horizontal it won't quite perform to specs. If it is vertical, many punch holes or slots in th e PC board under it to facilitate convection.

I wouldn't put a plate on the top as it would inhibit whatever convection y ou have going on now. It would also be a feat to get good thermal contact b etween a plate and the top of those fins. You're better off with a differen t heatsink. If it takes awhile to overheat you apparently have thermal mass , but not enough dissipation. If you can get one thinner and with more fins it might just do the trick. If not at least you'll know sooner.

Another trick that might work is a fan. You don't need the fan right by the part, all you do is make it blow out and put a vent near the part. It'll suck in a nice breeze.

It would help to have pictures of the thing to see the layout. That determi nes the options more than anything else.

Reply to
jurb6006

Almost any way to get more ali area would work. Almost. There's nothing magic about heatsink designs, their optimisation is more to get you to pay several times the price of a bit of ali than, in 99% of cases, anything more useful.

Black coatings aren't very useful either, just avoid shiny metal.

Yes yes, if you must have absolutely every last drop of heatsinking buy a big fat commercial heatsink, it'll be a bit better than an ali bracket. And a lot more money.

NT

Reply to
tabbypurr

I just noticed it uses a mica insulator, presumably left over from the construction of Noah's ark. There must be something better.

Machining isn't a problem, or rather it's someone else's problem. But coper has nearly twice the thermal conductivity as aluminium, so I thoughtthat would help

It's a NatSemi LM4765 stereo audio amp used as a bridge, it's obsolete but we have stock. The standby function on these is useful. They do make what appears to be an identical part in every spec except it has an isolated tab and lower power, so I guess the same die. I wonder if the isolated tabs are better than metal tab + insulating washer?

Running at about 15W.

Cheers

--
Clive
Reply to
Clive Arthur

Horizontal PCB (so vertical heat sink) in a 1U rack box, hence the height restriction. I couldn't readily connect it to the (steel) box, but there's plenty of room to dissipate the heat if I can spread it out.

What's your opinion of copper vs aluminium? Cost is not really an issue, though diamond is out.

Cheers

--
Clive
Reply to
Clive Arthur

** A small BLDC fan ( say 40mm dia) blowing directly across the heatsink will do the job. No need to cut holes to open up the rack box, just spread the thermal energy around using the fan.
** You got one, cheap and easy to carry out too.

I've used the same idea stop 3 terminal regulator ICs from going into thermal shutdown in hot conditions.

.... Phil

Reply to
Phil Allison

Yes, I'm coming round to that view, at least until we do a new PCB. There's plenty of room in the box.

But how reliable are fans? What's a good make?

Cheers

--
Clive
Reply to
Clive Arthur

Seem remarkably reliable - how many of your PCs died from fan failure? I've only seen one in forty plus years.

piglet

Reply to
piglet

In natural convection a beefy heatsink like that, hemmed in above and below, will be as near isothermal as makes no odds. Steel would probably be about as good.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
 Click to see the full signature
Reply to
Phil Hobbs

Ball-bearing fans from reputable manufacturers are pretty good. Rotron and EBM are names I've heard good things about.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
 Click to see the full signature
Reply to
Phil Hobbs

The best insulator is no insulator. Just a bit of grease.

As Phil says, either metal would be isothermal. But make sure it's flat where the transistor lands.

Bolt the transistor to the metal and isolate the metal. Like this:

formatting link

15 watts dissipation?

--
John Larkin         Highland Technology, Inc 

lunatic fringe electronics
 Click to see the full signature
Reply to
John Larkin

Read the data sheet. typically several years

Most can be run well below rated speed which will reduce pressure, flow, noise and wear but not lubricant or plasticisr evaporation.

Sunon seems good, That's what I mostly see inside servers. Thier website is broken but mouser holds copies of their data sheets. The one I looked at claimed 70000 hours (about 8 years continuous)

--
  When I tried casting out nines I made a hash of it.
Reply to
Jasen Betts

Reliable? meh. Of the various PC failures I've encountered a lot were down to fans. HDDs being the other weakpoint. And caps to a lesser extent.

Fans last longer on reduced voltage - some start up reliably low, some don't. And of course they last longer if you only turn them on when the IC is or is likely to be hot.

One good trick is to run them at reduced voltage but let them have a good starting kick of max voltage by using a big fat lytic. Otherwise they can fail early due to low torque starting & increasing stiction/friction.

But really do you need to do that? Just get more aluminium area in there. However bad the airflow it's still losing heat, and more area should still get you more heat loss.

There are other options but more metal is normally the simplest.

NT

Reply to
tabbypurr

I bought a piece of Cu busbar, 50mm x 100mm x 3mm which I bent (lump hammer and vice) into an L with the larger part horizontal and the smaller part vertical mounted to the amplifier. The chip body temperature is down from 125'C with the old heatsink to 95'C. The furthest part of the Cu is only 3'C lower, so yes, almost isothermal, so Al will do almost as well and could be bigger too.

I'm hammering it a bit harder than the customer does, so I reckon this will fly, though a fan blowing on the old heatsink would be easy too. Shame I don't have 5V or 12V spare, but there's plenty of room for an extra PSU.

Cheers

--
Clive
Reply to
Clive Arthur

Watch out for vibration. That big assembly hanging off the leads of the poor TO-220 is liable to wind up broken off and lying on its side when the box comes off the truck.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
 Click to see the full signature
Reply to
Phil Hobbs

o

Nice one. Ali would be more robust too as it weighs less. And of course you could go thinner.

you could shape it so the other side of the heatsink meets something & gloo p it in place.

NT

Reply to
tabbypurr

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.