New amplifier...possibly..

[fooboo]

Hi there

Between Now and getting the new prints of the corrected Amplifier PCB done I am reading up on Heat sinks.

Using Rod Elliots, free download, Heat sink calculator it appears my heat sink has a thermal rating of 1.11C/W. This an approximation, Rod advises, but is a good starting point.

Another site uses the equation -

50/SqrRt of the surface area (exposed) In my case 1197Cm2 which comes out at 1.45C/W

Looking at the data sheet for the LM3876T It looks like at +/-30VDC with an 8ohm load I will be looking at 42W OP. Taking this to the Power dissipation VS Output Power suggest a PD of around 33W.

Then referencing this to the max heat sink Thermal resistance at Specified Ambient temp It looks like for an ambient temp of 25C a Thermal rating of 5.1 is needed. SO the allow headroom A smaller thermal rating is needed. 1.2 would see me good for 90C which will never happen in the UK

So I thinking of cutting the heat sink in 2 to allow a different enclosure layout. This would take the Thermal rating of each half to 2C/W Which will be good for an ambient temp of 70C.....Still not going to happen in the UK

If I get this bit right - Thermal resistance X OP +25C Ta as a base would give me 1.11C/W x 33W PD + 25C = 62C X 2 for both IC's on the same sink 123C. Absolute maximum rating for the IC's is 265C So I have half again head room?

For the 2C/W with one IC I would be looking at 91C So again still well under the maximum temp?

Have I understood the above well enough to be correct?

regards

Foo

PS keep it simple please I am horrendous at math.

[AndrewT]

I recommend using the graph that National show in the 3886 datasheet.
Does the 3876 have the same design graph?

Then double the sink rating from what National says you need.

Other than dividing by 2 there is no other arithmetic.

[fooboo]

Hello Andrew

The Heat sink charts are identical for the LM3876 and LM3886. So it looks like I'm within specs with either variation.

I attach the relevant graph marked for the 60V total VDC input with an 8ohm load.

regards

Foo

[AndrewT]

I would go all the way to the left temperature value Ta that your heatsink/s may experirnce in normal operation.
For external heatsinks in the UK 25degC to 30degC is a normal maximum.
For internal heatsinks in the UK allow 40degC to 50degC.

Let's suppose you use external sink and Ta <=25degC.
You would extract somewhere between 3.8C/W and 5.1C/W. Let's guess at 4.4C/W.
Doubling comes to 2.2C/W for each chipamp.
For a stereo pair you should use 1.1C/W

The reason I don't use 70degC nor anything higher is that the chip will run hot. So hot that the protection systems will do their best to prevent damage to the chipamp. That means triggering of the protection circuits. That does not sound nice.

[fooboo]

Hi Andrew

Thanks for catching my 'errored thinking'. It seems I've been misunderstanding Thermal rating for quite some time..

I'm now trying to check I have rated my heat sink correctly as there is a little ambiguity with the data input.

Off to the printers on Monday to get my corrected Amplifier PCB done.

regards

Foo

[danielwritesbac]

Here's the considerations:
1). The thermal dissipation size of a big sheet of aluminum is fantastic, but the thermal interface speed is insufficient.
2). The thermal interface speed of a thick chunk of aluminum is fantastic, but the thermal dissipation size is insufficient.
Neither case 1 nor case 2 will work as an heat sink.
But thermal pasted and bolted together they can work.

Your enclosure is at case 1. So, add a nice big thick heat spreader, so easily available as bar stock from the hardware store.

And you might want some sort of thermally efficient decoration, such as U-channel onto the sides of the enclosure.

Many homebrew heat sinks are successful, but you need speed and size and ventilation.

[fooboo]

Hi Daniel

Thermal compound or Sil-Pads will be used never fear.

The enclosure situation is a very amorphous part of the build to which end I have purchased a large heat sink to use directly with the chips.

As you can see in the drawings it is a fairly sizable chunk of extruded Ali'.

It may well get cut in half along it's widest measure to allow a shallower enclosure but until I am sure I have actually worked out it's thermal rating I'm not sure what's happening.

See also the ESP heat sink calculator screen shot. I think I have the correct measures in the correct places.

regards

Foo

[AndrewT]

That 5mm backplate will allow two devices to run hot at their contact/interface locations, but leave the extremities of the sink relatively cool. This will increase Tc to above the predicted temperature.

Note that ESP does not ask for number of devices/locations, nor the thicknesses of fins & backplate.

What's the plural of thickness?

[danielwritesbac]

That cute little heatsink should work for one chip if you have 8 ohm or 16 ohm speakers.

[fooboo]

Hi there

Andrew - Rods site does indicate that the actual math is more involved and the spread sheet is merely a 'rough guide'. For those with experience I'm sure it's enough.

I think the plural is Thicknesses.

Daniel - Enough for 1.................. Oh..... FIDDLE!

But both of you indeed indicate it's not enough for the job of both chips.

I have a CPU Sink. Although I understand this may be no good due to the density of the fins designed as the are for forced air flow. But it is 70mm x 82mm x 30mm High. There are 28 fins which are 25mm high. The base is 7mm thick near the centre section and thins out towards the edges to 3mm. The surface area is 1246.4cm^2.

Would this be any good? Failing that could one of you wise gentlemen point me at a known heat sink.

regards

Foo..... The aghast!