LMxxxx Thermal Simulation

[zenon]

I'm looking @ doing some thermal simulations on LM4872 and LM3875 chips. I can get absolute peak thermal loads from the .xls file they provide. What's a fair "average" output to use? 75% of that value?

TIA.

[SpittinLLama]

Tough question because there are lots of variables. Depends on what you want as well. Do you want a tough as nails amplifier that can run hard all day in summer in a hot room or outside and be fine? Do you want an amplifier that can handle hard speaker loads without issue? Want it to be able to run sine waves at the peak Pd all day too? Balancing these design goals will give you the trade offs that you want or don't want. I have undersized some of my designs because they were for me and I wanted a smaller physical size. Music is much less demanding thermally than sine waves or even pink noise so it just all depends.

Short answer, I'd say 75% is reasonable, so if the sheet tells you 40W max Pd then sizing for 30W would be fine. Hard to say for sure without more info.

Good luck.

-SL

[zenon]

I'm looking at making a 7.1 power amp. Stereo channels will be bridged LM4780, as will the centre channel. Surrounds will be LM4780's in a stereo configuration. All these amps will be on a shared heat sink with the two bridged stereo channels a far away physically as possible. I'm modelling these in ANSYS and trying to get a reasonable estimation as to how well it will all work.

The heatsink will be Aluminum plate 0.190" and approx 15"x15" IIRC

Do you have or know of any actual data, such as a given chip with a given voltage and a measured heat sink temperature? Something I can use as a simulation benchamark.

Thanks.

[zenon]

BTW, the worst load would likely be a pair of Seas Thor TL's on the Stereo.

[SpittinLLama]

I don't know of any actual data like you want. That is a lot of power in one box. I'm not sure your plate will be enough, maybe if one on each side of the box and you can split channels between two plates but that is a lot of power to dissipate. That is the right choice, to put the bridged channels away from each other.

-SL

[jackinnj]

Originally posted by SpittinLLama
I don't know of any actual data like you want. That is a lot of power in one box. I'm not sure your plate will be enough, maybe if one on each side of the box and you can split channels between two plates but that is a lot of power to dissipate. That is the right choice, to put the bridged channels away from each other.

-SL

after a few moments the thickness doesn't matter -- you are just loading up a "body" with energy -- it's surface area and air velocity which move the heat --

[poobah]

Go to the Aavid Thermalloy website. Look at some various profiles... large ones. Then study carefully the formulas regarding length, orientation etc...

From looking at other designs you can extrapolate a bit to something else... I know it is blaspohome, but a small fan works wonders... especially if you use a 12 Volt fan at say 8 volts... you get a cubic reduction in noise.

Jackinnj is dead on about the thickness. Thickness is valuable, but not what sheds the heat. And, I too would guess your 15 x 15 plate is a wee bit small. Vertical is the most important thing, then fins, then forced air... even a bit is a world of difference.

[zenon]

Interesting site poobah!
I 2nd your idea of lowering the voltage of DC fans, my computer is liquid cooled and I have two 12v case fans wired in series. They are inaudible.

Even if I had a benchmark of a well defined (enough so that I can model it in 3D-CAD) pre-existing safe design, I can easily throw it into Ansys and see how it performs to use as a benchmark.

[poobah]

Glad I got you thinking... as far as modeling goes, hot air can and will find its way upward and out if you give it half a chance. Cold air finding its way in is a different story though. Think about the cold getting in... just a tip!

[dfdye]

For the LM3886 based amps that I have built, Slot A CPU heat sinks have worked great for me simply because I don't really crank on the amps that hard. I have an aux. fan to cool the sinks if they need it, but I have never had it kick in.

The point is that in addition to th other parameters, I would keep in the back of my mind how loud I plan on listening to music. The amp I built for my lab (which has to play hard out to crank over noise from fume hoods and other equipment) is in a more robust package than the amp I have for my office since the latter doesn't ever play that loud.

Of course this is just anecdotal and can't really be modeled well, but it is just a little experience I have had with different design parameters.

Good Luck!
David