heatsink layout

This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hi all

I'm in the process of build an Elliot P68 Subwoofer amp with 4 pairs of MJ15003/4 outputs. I have a number of options with regards to heatsinks. Have a look at 3 ideas i've had so far and tell me which you think is best. I will use fan cooling at both ends, and the sink is about 30cm long, 9.5cm wide (12inchx4inch) The light grey bits are aluminum profiles about 6mm thick (1/4 inch). Using Mr Elliots Excel spreadsheet, I get about 0,4C/W for both sinks together with a heatsink temp of 100C which is about right.

I calculated dissipation as follows:

Rails: +-56V
Max dissipation: 56V/2 = 28V. Current is 28/4ohms = 7A
therefor P= 28*6 = 196Watts
For 4 pairs or TR's, each TR dissipates 196/4 = 49W
Tjunction = 0,7C/W Tcase-heatsink = 0.5C/W Total= 1.2C/W
1.2C/W * 49W = 59C.
So max heatsink temp is 200-59 = 141C for junction temp of 200C

Let's assume ambient T is 35C.
Then heatsink can dissipate (141-35/0.4) = 265W before it reaches 141C.

At a dissipation of 196W max, heatsink is at temp of (0,4*196) = 113C if ambient temp is 35C. This seems to work out fine.

Since i will be using fan cooling, does anyone have any formulae for working out the new thermal rating of the sink because of the forced air cooling?

oops, here's the diagram


  • heasinksp68.gif
    17.6 KB · Views: 911
Just so you know, at the maximum junction temperature of a transistor (200 deg c for yours), it's rated to do *zero* work at all.

At a 113 degree heatsink temperature, if you include the thermal resistance of your insulators and thermal grease between the transistors, your junction temperatures are going to be very dangerously high. Your amp will function for a while, but likely not a long while.

For good reliability, you're going to want to keep the sink temperature at 60 degrees c or under.
Thanks for the link - i will look it up in a sec.

Good point about the 200C junction temperature - i completely forgot about that!

Since this is a class B amplifier rated at about 300-350W, what sort of average dissipation can i design for since these calculations are for a continuous dissipation at 28V?
new calculations

I redid the calculations

Based on 1.43W/C derating for the transistors dissipation, If each TR needs to dissipate 49W then the max junction temp is 130C. This means the max heatsink temp is (130-49W*1.4C/W)= 61C like you said.

So i need to mount only 2 pairs on a 0,3C/W heatsink. That means i need 2 x 0,3C/W sinks! That will take up the whole amplifier case! How do the pro amps then manage >1000W per channel without taking up a whole room with heatsinks?

Bear in mind that you can considerably relax the heatsinking requirements when you use forced air cooling. This could mean using a smaller heatsink, or mouting it somewhere convenient for layout but not free airflow. From what I can see you have not yet figured this into your calculations.

When I built my P3A variant, I placed a bimetallic switch on the main heatsink that engages a pair of 60mm fans when the temperature rises above 60 degrees. A further switch activates the manual speaker disconnect on my amplifier protection circuit if the heatsink temperature rises any further - above 80 degrees. The provision of a safety net should the fans fail is very important.

As an example of how effective fan cooling is, when I drive my (2 x 2SC3281/2SA1302 per channel) P3A amp hard, the fans operate for around 30 seconds every quarter of an hour or so - that is all that is required to keep the meagre, horzontally mounted 1 deg. C/W heatsink cool.

Hope this helps,
hey that's really interesting to hear. I have already built a little temp sensor circuit that can trigger a fan above a certain temperature, so now all i need to do is get some general idea of how much heatsink i do infact need with the fan cooling. Since this is not a Hifi amplifier i'm building (it's just for Bass duty) and considering that i love my music loud, having noisy fans at 50dB is not going to be noticed in ave room SPL of 105db+, so i could just leave the fans to run permanently and do away with the temp sensor - but that would take some 'gadget' factor out of the amp, so maybe i'll just include it anyway. :)

Thankyou for your post.
Joined 2002
Paid Member
You don't have to have 50 dB fans...

Some nice 80 mm Vantec computer case fans are at about 20-25 dBA, and will put out about 15-18 cfm, which should be plenty to bring a 1 c/w heatsink down to about 0.5 or so.

Consider ducting fan flow into the heatsink. This may allow for more effective cooling, if properly done. If not then turbulence noise will create problems.

in case the aluminium profiles are made of individual parts, the most critical part in your heatsink layout is the point where the aluminium plate you mount the transistors on contacts the other plate(s). (Have you considered the thermal resistance of the aluminium profiles?) Because it is very difficult to get the edge of a plate smooth and straight, you will encounter severe problems with the heat crossing between the two aluminium plates.
Especially in the second example the right heatsink will not contribute at all.
You should use a One-Part-H-Profile or a different layout.
Or different transistors you can mount directly to the heatsink.

Greetings to South Africa,
Yes actually i have seen that project. Thanks for the suggestion. I built one and managed to fry the opamp by *accidentally* shorting out the pins. Version 2 is on the drawing board.

With regards to my diagram above, the left most heatsink cross-section is a solid H profile, not 3 separate Al bits.
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.