The TPA3110 (if I'm right, the amplifier chip used on this board) and many class D amplifier chips use a pad-down "PowerPAD" (located on the bottom of the device). If the board is designed and assembled correctly, the heat from the die should be transferred out the bottom of the package and into the PCB. Depending on the board design, there may be multiple layers of the board used to dissipate heat. Your best bet would be to forget the heatsink, elevate the board, and blow that fan horizontally across the top and bottom of the board. See note 3 on page 5 of the datasheet: www.ti.com/lit/ds/symlink/tpa3110d2.pdf (TI blocks these links so remove any junk after the .pdf) For more detail on the way the pad-down PowerPAD is designed to work, and the way the pcb is supposed to be designed: www.ti.com/lit/an/slma002g/slma002g.pdf With a normal music signal, I wouldn't expect that amp to have any issues driving 4 ohm speakers without anything but a reasonable amount of airspace around the pcb (standoffs would help). My daughter has that exact amp board in a VERY small enclosure driving 6 ohm Sony bookshelf speakers. She runs it hard quite often and has had no issues in over two years. Among other safety features, the TPA3110 has built in overtemp protection. The board has the correct thermal vias and lots of ground plane - so why worry about a heatsink?
The potential advantages of back side cooling were unfortunately left unexplored. I mean, you said it yourself, the top of the IC is just epoxy, it's not the most thermally conductive thing ever. The die is closer to the thermal pad on the bottom, and hopefully there's sufficient thermal via stitching. Then remains the question of how to measure. You could measure against uncooled case by placing your existing thermocouple against the to of the IC, but you don't really have a good way to directly measure front vs. back side cooling, although maybe simply comparing with uncooled case and crudely assuming constant offset could be enough of an indication. I recall someone, maybe GamersNexus used some kind of very very thin flat K-junction thermocouple that also acts as thermal tape to measure temperature straight between IC and heatsink, and they got confirmation from one of the graphics board companies that it's the same thing they use in house. Potentially the best way is running the IC to the limit using an adjustable load and measuring when the IC goes into thermal shutdown, which should conclusively tell you whether any given cooling setup is sufficient. But then, an adjustable load is something that likely still needs to be built, as nobody usually just has one in their closet.
The interface to the heat spreader is extremely important and using tape with a high thermal impedence really limits how much heat flows to the heat sink. Easy way to check if there is too much impedence is by feeling the heat sink temp vs the chip temp. If the heat sink isn't hot when the chip is, you need a better interface. I think you would have much better results with thermal adhesive.
Have you ever tested heatsinks on the bottom of a board with the thermal pad area. Mine have a big (twice the size of the chip) silver pad visible on the bottom of the board beneath the chip. I have a chip on the board that has a large heatsink on top but I think it would be better on the bottom, because this chip is supposed to direct 80% of its heat through the board. Assuming you have enough room, could gluing a heat sink on the bottom covering some of the vias do more harm than good?? Thank you. Good video. Very interesting to us novices,
John, excellent video. Could it be that the lack of temperature reduction indicates that a lot of heat is already being dissipated through the board? That would account for the small heatsink not doing much at all.
This Maxim application note linked below is even more to the point. Here are two excerpts if you don't want to bother reading it: 1 - "When a Class D amplifier is evaluated in the lab, a continuous sine wave is often used as the signal source. Although this is convenient for measurement purposes, it represents a worst-case scenario for thermal loading on the amplifier. It is not uncommon for a Class D amplifier to enter thermal shutdown if driven near maximum output power with a continuous sine wave." 2 - "The top side of the IC is not a significant thermal path for the device, and therefore is not a cost-effective location for a heatsink." www.maximintegrated.com/en/app-notes/index.mvp/id/3879
Thermal adhesive pads are near-useless unless the heatsink is being pressed down on to the chip with push-pins or something similar, you will never see more than incremental drops in temps. I have two [edit] -TPA3116s- TPA3118's ( in a bi-amped set-up) as my 'daily drivers', and they are 'slug-down', PCB-cooled chips like the ones you've been testing. Since they're on 19.5 volts and get driven fairly hard sometimes, I've contemplated mounting heatsinks with a spring between them and the top of the amp's case (which lend themselves nicely to this as they're in two halves, and the top half can be lifted off and replaced vertically). BTW, the performance of adhesive pads can be improved by heating them and heatsink (heatgun, hairdryer, warm oven) to at least 80-100°C before pressing and holding them firmly in place until they cool.
Since most of these power chips have a square or rectangular heat pad at the bottom, why don't people route out a hole the exact same size and shape of that pad into the pcb, and over-solder it to the copper pour on the backside, instead of using many small vias? OR one large enough via to cover the heat pad? I've never seen that.
Because that would require a separate step during manufacture, a very uncommon one at that, which would significantly increase the price, and may not work any better than many small vias
has anyone used the motor salvaged from a DVD player running at 3 or 4 vaults turned into a fan used it in an amplifier build? the motor is so small and so easy to get from old DVD players. not sure about sticking a fan on it but maybe it's worth researching.
at 3:15 you say that you get your money's worth! lol We are all guilty of just wasting our money just to make something work or to just see how it works, in the name of our beloved hobbies!!!
I came across this very interesting video about heat dissipation for the class D amps, just that the guy pushed the poor class D amp just a little bit too far: czcams.com/video/Kv_Zn8HB_NY/video.html
Wow even though the heat sinks were small I thought they would make more of a difference. Glad I added a fan in my last build.
The TPA3110 (if I'm right, the amplifier chip used on this board) and many class D amplifier chips use a pad-down "PowerPAD" (located on the bottom of the device). If the board is designed and assembled correctly, the heat from the die should be transferred out the bottom of the package and into the PCB. Depending on the board design, there may be multiple layers of the board used to dissipate heat. Your best bet would be to forget the heatsink, elevate the board, and blow that fan horizontally across the top and bottom of the board. See note 3 on page 5 of the datasheet: www.ti.com/lit/ds/symlink/tpa3110d2.pdf (TI blocks these links so remove any junk after the .pdf) For more detail on the way the pad-down PowerPAD is designed to work, and the way the pcb is supposed to be designed: www.ti.com/lit/an/slma002g/slma002g.pdf With a normal music signal, I wouldn't expect that amp to have any issues driving 4 ohm speakers without anything but a reasonable amount of airspace around the pcb (standoffs would help). My daughter has that exact amp board in a VERY small enclosure driving 6 ohm Sony bookshelf speakers. She runs it hard quite often and has had no issues in over two years. Among other safety features, the TPA3110 has built in overtemp protection. The board has the correct thermal vias and lots of ground plane - so why worry about a heatsink?
The potential advantages of back side cooling were unfortunately left unexplored.
I mean, you said it yourself, the top of the IC is just epoxy, it's not the most thermally conductive thing ever. The die is closer to the thermal pad on the bottom, and hopefully there's sufficient thermal via stitching.
Then remains the question of how to measure. You could measure against uncooled case by placing your existing thermocouple against the to of the IC, but you don't really have a good way to directly measure front vs. back side cooling, although maybe simply comparing with uncooled case and crudely assuming constant offset could be enough of an indication. I recall someone, maybe GamersNexus used some kind of very very thin flat K-junction thermocouple that also acts as thermal tape to measure temperature straight between IC and heatsink, and they got confirmation from one of the graphics board companies that it's the same thing they use in house.
Potentially the best way is running the IC to the limit using an adjustable load and measuring when the IC goes into thermal shutdown, which should conclusively tell you whether any given cooling setup is sufficient. But then, an adjustable load is something that likely still needs to be built, as nobody usually just has one in their closet.
The interface to the heat spreader is extremely important and using tape with a high thermal impedence really limits how much heat flows to the heat sink. Easy way to check if there is too much impedence is by feeling the heat sink temp vs the chip temp. If the heat sink isn't hot when the chip is, you need a better interface. I think you would have much better results with thermal adhesive.
Hi Jeff...can you recommend any tape? Manufacturer and number? Thank you!
Yeah, as far as i remember, when the heat sink is attached, the heat sink itself gets/ supposed to get hotter than the ic itself
Have you ever tested heatsinks on the bottom of a board with the thermal pad area. Mine have a big (twice the size of the chip) silver pad visible on the bottom of the board beneath the chip. I have a chip on the board that has a large heatsink on top but I think it would be better on the bottom, because this chip is supposed to direct 80% of its heat through the board. Assuming you have enough room, could gluing a heat sink on the bottom covering some of the vias do more harm than good?? Thank you. Good video. Very interesting to us novices,
Seems like the addition of a heatsink with a fan blowing on it would be better than the chip with just a fan blowing on it?
John, excellent video. Could it be that the lack of temperature reduction indicates that a lot of heat is already being dissipated through the board? That would account for the small heatsink not doing much at all.
This Maxim application note linked below is even more to the point. Here are two excerpts if you don't want to bother reading it: 1 - "When a Class D amplifier is evaluated in the lab, a continuous sine wave is often used as the signal source. Although this is convenient for measurement purposes, it represents a worst-case scenario for thermal loading on the amplifier. It is not uncommon for a Class D amplifier to enter thermal shutdown if driven near maximum output power with a continuous sine wave." 2 - "The top side of the IC is not a significant thermal path for the device, and therefore is not a cost-effective location for a heatsink."
www.maximintegrated.com/en/app-notes/index.mvp/id/3879
Thermal adhesive pads are near-useless unless the heatsink is being pressed down on to the chip with push-pins or something similar, you will never see more than incremental drops in temps. I have two [edit] -TPA3116s- TPA3118's ( in a bi-amped set-up) as my 'daily drivers', and they are 'slug-down', PCB-cooled chips like the ones you've been testing. Since they're on 19.5 volts and get driven fairly hard sometimes, I've contemplated mounting heatsinks with a spring between them and the top of the amp's case (which lend themselves nicely to this as they're in two halves, and the top half can be lifted off and replaced vertically).
BTW, the performance of adhesive pads can be improved by heating them and heatsink (heatgun, hairdryer, warm oven) to at least 80-100°C before pressing and holding them firmly in place until they cool.
Why does the perfomance increases if you preheat it?
Tpa3118 is the exact reason im looking if these mini heat sink even work or not 🤣 they should have given us the place to mount a heat sink tho 😑
Yep, interesting. I would like to have a infrared camera to visualize the heat on a board. But they are still very expensive.
Since most of these power chips have a square or rectangular heat pad at the bottom, why don't people route out a hole the exact same size and shape of that pad into the pcb, and over-solder it to the copper pour on the backside, instead of using many small vias? OR one large enough via to cover the heat pad?
I've never seen that.
Because that would require a separate step during manufacture, a very uncommon one at that, which would significantly increase the price, and may not work any better than many small vias
has anyone used the motor salvaged from a DVD player running at 3 or 4 vaults turned into a fan used it in an amplifier build? the motor is so small and so easy to get from old DVD players. not sure about sticking a fan on it but maybe it's worth researching.
cool dude. How about watercooled, dry ice or nitrogen ?
What about results for 8 ohms load 😊❤
at 3:15 you say that you get your money's worth! lol We are all guilty of just wasting our money just to make something work or to just see how it works, in the name of our beloved hobbies!!!
Heatsink is no good without a fan unless you have a passive heatsink.
I never knew their was a tape spefic for heat sinks
I had knew it back 2 months ago I wouldn't have tied a heat sink by string,😐😫
It's cheap, let it bake. If it starts thermally clipping while playing Music then worry about it.
I came across this very interesting video about heat dissipation for the class D amps, just that the guy pushed the poor class D amp just a little bit too far: czcams.com/video/Kv_Zn8HB_NY/video.html