LM380 heatsink

Started by disorder, February 23, 2019, 05:45:29 PM

Previous topic - Next topic

disorder

The datasheet for the LM380 power amp IC shows a once-available add on heatsink. It gets soldered to pins 3-5 and 10-12. I am going to construct my own out of flattened copper but I can't figure out if I want the heatsink to also contact the top side of the IC package itself? Would this bring unnecessary heat from the ground plane pins up to that part of the package? Or would it be good for the package itself and the pins to be thermally connected?

http://www.applelogic.org/files/LM380.pdf


Rob Strand

You don't *have* to use that heatsink.    The size of heatsink required depends on the power dissipation.  The power dissipation depends on the supply voltage and the load impedance.  So maybe you don't need such a large heatsink.

The graph at the bottom of page 4 of the datasheet shows more options.  However it indicates a 2 oz/sq-in copper thickness on PCB.  If your PCB copper thickness is unknown it could be 1 oz/sq-in or even 1/2 oz/sq-in, which would have an impact on the heatsink degC / W rating, but maybe it's still OK for your load and PSU.

The ratings on page 3, note 4, indicate using the PCB heatsink.

QuoteI want the heatsink to also contact the top side of the IC package itself? Would this bring unnecessary heat from the ground plane pins up to that part of the package? Or would it be good for the package itself and the pins to be thermally connected?

Page 3 note 7 Phi(j to pins) = 15 degC/W  and the Phi_JC 30 degC/W (14-pin) means that heatsinking the top surface will do *something*.  The problem I see is attaching the heatsink to the pins might will introduce problems trying to get good thermal contact to the top. If there's a large air-gap it makes the effort a waste of time.
Send:     . .- .-. - .... / - --- / --. --- .-. -
According to the water analogy of electricity, transistor leakage is caused by holes.

Mark Hammer

I managed to score a neat little heatsink for my 380-based amp 15 years back, that screwed onto the PCB over the chip.  If you can unzip the file, there should be a pic in here: http://hammer.ampage.org/files/Miniamp.zip

disorder

Quote from: Mark Hammer on February 23, 2019, 09:43:51 PM
I managed to score a neat little heatsink for my 380-based amp 15 years back, that screwed onto the PCB over the chip.  If you can unzip the file, there should be a pic in here: http://hammer.ampage.org/files/Miniamp.zip

That's a cool build! Probably sounds better than most other newer chip amp based guitar amps as well. Did that pcb mounted heatsink also contact the ground plane pins?

bool

Before committing to metalwork, try to see if you can get by with making a dual-layer pcb with large copper pads at top and bottom layers to heatsink the chip.

Similar "technology" was used for tba800/810 chips ... back in their era.

PRR

Good on you for using LM380. It is a far better chip than the '386 cut-down, far-far better than recent '386 made for the modem market.

Figure your actual dissipation. The curves are in the datasheet. Up to 16V and 8r load, you do not need a sink.

As Rob says, you probably can't get both side-legs and top-surface with a single sink, though I don't think it would matter if you did.

As Mark says, there are now other options for DIP sinking.
  • SUPPORTER

disorder

Quote from: bool on February 24, 2019, 07:27:01 AM

Quote from: PRR on February 24, 2019, 02:04:19 PM

Thanks for the suggestions and info! My girlfriend happens to be a metalsmith, she has some flat stock copper and a desire to help me with this project. I might work with her on bending and cutting up a simple heatsink. Will post results if we go that route!

bool

Power to the metal chicks!