2N3904/06 pair max dissipation on heatsink

[oneoldude]

I am putting together a complementary symmetry buffer using the 2N3904/06 combo. I plan to drive headphones with it.

For thermal tracking and some heatsinking, each pair will be epoxied face to face on a separate one in sq .062 alum heatsink.

Measurements have shown me that the internal chassis temp will be less than 40C.

The data sheets show R Theta JC to be 83C/W and R Theta JA to be 200C/W.

My best guess is that R Theta for the heat sink will be a bit worse than 20C/W

I can find no data re: R Theta for case to sink. But I assume the actual R Theta for each complete assembly (two transistors on one heatsink) will be somewhere between 83C and 200C. But where will it be?

So for a single transistor in free air at 40C ambient the calculator at:https://www.daycounter.com/Calculators/Heat-Sink-Temperature-Calculator.phtml shows that with an R Theta JA of 200C/W and .3W dissipation the junction temp will be about 100C. That will occur with IC of about 25 mA (more current would be better).

My first question is whether this is too high for the TO92 series.

Now if we glue the driver and output transistor pair on a 1 sq in sink and assume both will be dissipating .3W, am I headed for disaster? Or will the situation be manageable?

If I use R Theta JC of 83C/W and the assumed 20C/W of the sink then the calculator gives me 71C for one transistor and 101C for two. But this calc assumes perfect heat transfer from case to sink which is not true.

I can reduce the driver dissipation to about .15W if necessary. But I am trying to see how far I can go with these little devices.

So the ultimate question is, What is the max dissipation these pairs can stand (with reasonable life expectancy) if they are bonded face to face on a one sq in .062 thick aluminum heat sink? (I can also use a 1.25 in sq copper sink if that is a significant improvement)

Experience is the answer here and I have none with this problem.

Thanks.

 

[pinout]

I think this is an interesting exercise and something I remember investigating many years ago.
Unfortunately I can't recall any 'numbers' so will watch this thread with interest.

The biggest problem from memory is that black encapsulant is big compaired to the die and does not sink the power well to the heatsink.
In the good old days the glue was not that good and I recall sanding the transistor case and clamping that 'flat' side with heat sink compound the the heat sink.
Regarding the heat sink these work best as convectors and a vertical surface is the best (but not good in a small enclosed case) and matt black improves it for it radiant efficiency dramatically, but it has to be something like anodising and not a paint for best efficiency. I don't think copper will be significant for your application but I stand to be corrected.

An area I went onto investigate (briefly) was making the transistor pads larger (a circle split into three segments) and making these act as the heasink with the transistor mounted flush with the board. It is interesting that chips have evolved to use this method of removing heat obviously with the internal design optimised for this approach although some of the early IC power amp chips did use this method many years ago. From memory this approach did work well with my transistor testing, but again it was all trial and erro without 'design'!.

 

[orbanp]

...
So the ultimate question is, What is the max dissipation these pairs can stand (with reasonable life expectancy) if they are bonded face to face on a one sq in .062 thick aluminum heat sink? (I can also use a 1.25 in sq copper sink if that is a significant improvement)

The datasheet states the max dissipation for the transistors is 625mW, and the max operating junction temperature is 150C. It also states that "These ratings are limiting values above which the serviceability of any semiconductor device may be impaired."
If you plan to go beyond these ratings just use bigger transistors.

Peter

 

[oneoldude]

pinout,

I think I have a reasonable solution.

I talked with a designer of OPAs and he said that they routinely design them with Tj of 100 to 110C and have no problems with longevity.

If that is so, then in compliance with the data sheet, the 2N3904/06 transistors will be able to handle .35W dissipation in an ambient of 40C (measured) with no HS at all. Tj will be 110C in that instance. Putting on a HS will be a safety belt. And yes, I have seen a datasheet that suggested a large pad at the collector to sink some heat. But no numbers were given.

I will probably stay at 300 mA where the Tj will be about 100C for a standalone 2N3904/06.

Thanks