TPA3116D2 Amp

If the soldermask isn't a great heat conductor then I suppose it might be worth trying. I'd say just try it first, see how hot it gets.

Yeah, I'll do that. After all what matters most to me is to learn how much better cooling gets at each step: stock board vs heatsink on soldermask vs heatsink on bare copper. It's all for the science 🙂

I saw xrk971 on here used RTV silicone to mount four of these boards to a big heatsink.

Neat! I'm searching for the relevant thread, but the advanced search function doesn't seem to work. Do you have a link by any chance?
 
Yeah, I'll do that. After all what matters most to me is to learn how much better cooling gets at each step: stock board vs heatsink on soldermask vs heatsink on bare copper. It's all for the science 🙂

Neat! I'm searching for the relevant thread, but the advanced search function doesn't seem to work. Do you have a link by any chance?


Sorry, I couldn't find it when I posted earlier. These threads are massive and I've just been reading TPA3116 stuff that's 5 years old, which makes me feel very behind! I've got it now, it was in my browser history:
Cheap TPA3118D2 boards, modding them and everything that comes with it
 
I was thinking the four corner standoffs would clamp the lump of aluminium between the pcb and the chassis? This will probably work though!

Yes, the plan is to mount the aluminium block to the chassis, then the pcb with the 4 standoffs at the same height pressing against the PCB, plus a dab of thermal compound between PCB & aluminium block. Wasn't that what you had in mind, except perhaps screwing the block to the chassis?

Well, we'll see how that goes. In principle it appears to be pretty simple and I like the fact that it's not permanent
 
Yes, the plan is to mount the aluminium block to the chassis, then the pcb with the 4 standoffs at the same height pressing against the PCB, plus a dab of thermal compound between PCB & aluminium block. Wasn't that what you had in mind, except perhaps screwing the block to the chassis?

Well, we'll see how that goes. In principle it appears to be pretty simple and I like the fact that it's not permanent


Yes that sounds good. The screw hole looks countersunk so it should sit flush against the board.
 
Hooked up the "Nobsound" NS-01G to two 8" speakers, & drove it from my pedalboard.... with no pedals on, sounds dull, so will need some sort of guitar type preamp. When I switch on the distortion pedal, there's a large amount of 50Hz noise/hash (I don't get this with my regular amp), so I think this will need an earthed power supply, or an earth somewhere in the chain. Using a 14v wall wart for the amp & 9v switchmode supply for the pedals. It's loud enough, but I doubt whether it could keep up with a drummer.... haven't been inspired enough yet to give it a serious thermal workout
 
Hooked it up to my main speakers & fed it bass heavy Chemical Bros for a couple of hours... the amp & PSU got a bit warm, but seem to be coping OK; I think I need a higher Voltage PSU to give it a serious test, saw one at work yesterday I'll borrow & try next weekend. The low end doesn't seem to be as solid as my class AB Sonab, but that's hardly surprising from a wall wart, still pretty impressive though...
now got to watch the Wallys being defeated by the All Blacks, cheers,
 
I am using my home-brewed TPA3118 boards in my guitar stage amps for several years. They are powered with 24V notebook bricks. To make it louder on a bigger gig I fired two 8Ohm speakers resulting in 4Ohm load. Before that I did a lab test with 40Watt into 4Ohms dummy resistor for one hour - which is far more heating than real music. No thermal overload occured.
All in all I encountered no single prob over the years. These little TI-chips are real workhorses😉
 
Hooked it up to my main speakers & fed it bass heavy Chemical Bros for a couple of hours... the amp & PSU got a bit warm, but seem to be coping OK; I think I need a higher Voltage PSU to give it a serious test, saw one at work yesterday I'll borrow & try next weekend. The low end doesn't seem to be as solid as my class AB Sonab, but that's hardly surprising from a wall wart, still pretty impressive though...
now got to watch the Wallys being defeated by the All Blacks, cheers,

Apparently that Nobosound has a nice heatsink on top and lots of vias underneath, so cooling shouldn't be an issue.

341946_img_20180124_191540-2976x2232.jpg


Maybe lowering the gain / increasing input impedance would help boost that lower end?

I am using my home-brewed TPA3118 boards in my guitar stage amps for several years. They are powered with 24V notebook bricks. To make it louder on a bigger gig I fired two 8Ohm speakers resulting in 4Ohm load. Before that I did a lab test with 40Watt into 4Ohms dummy resistor for one hour - which is far more heating than real music. No thermal overload occured.
All in all I encountered no single prob over the years. These little TI-chips are real workhorses😉

Impressive! Thanks for sharing, voltwide. Do you have a heatsink in your board design?

I take it you used 20V for your 40W test into 4ohms, I reckon one can exceed those 40W when playing overdriven guitar, so unless you play clean I'm guessing you're getting more than 60W on your TPA3118 guitar amps @ 24V / 4ohms 😱
 
As I used TPA3118 - there was no heatsink.

Only the pcb did the cooling.

40W of output is just a rough approximation.
The supply was indeed 24V=, but as I am using some soft clipper to limit output level below class-d-clipping the max output level is reduced by about 10 %.
 
As I used TPA3118 - there was no heatsink.

Only the pcb did the cooling.

40W of output is just a rough approximation.
The supply was indeed 24V=, but as I am using some soft clipper to limit output level below class-d-clipping the max output level is reduced by about 10 %.

Awesome, thanks, do you have pics of your board by any chance?
I have no idea what that soft clipper is TBH (does it have anything to do with the soft clipping of opamps with diodes across the negative feedback loop? I guess not). Off to google again!
 
Awesome, thanks, do you have pics of your board by any chance?
I have no idea what that soft clipper is TBH (does it have anything to do with the soft clipping of opamps with diodes across the negative feedback loop? I guess not). Off to google again!
Yes, the result is similar to these diode clippers. Instead of diodes I use a pnp-differential (long tail pair) amp that is fed by some current source. I will look for some fotograph meanwhile
 
This is a pic of a recent version.

You see a bunch of additional circuitry around it:
-lead acid battery charger
-deep discharge protection
-a "deadman switch" shutting down after 15min music signal pausing
 

Attachments

  • 01_power_module_IMG_3173.JPG
    01_power_module_IMG_3173.JPG
    74.4 KB · Views: 698
Last edited:
Yes, the result is similar to these diode clippers. Instead of diodes I use a pnp-differential (long tail pair) amp that is fed by some current source.

Thanks, reading on that right now. Ensuring that the power amp doesn't clip harshly sounds like a great idea indeed

This is a pic of a recent version.

You see a bunch of additional circuitry around it:
-lead acid battery charger
-deep discharge protection
-a "deadman switch" shutting down after 15min music signal pausing

Very cool mate. The board I ordered is smaller than yours so it might not dissipate heat as efficiently as yours, we'll see
 
Some Bourns 2101-V-RC 10uH 10.8A inductors arrived today so I swapped the ones on a Sanwu red board. It's quite a tight fit but the solder pads are big so it's not too hard to solder them in. Getting the 'feet' bent just right is the trickiest part. I thought these bigger inductors would reduce the heat but they're heating up much like the little black ones! Oh well, I'm glad I didn't buy 12 of them at 2 pounds each from RS.

I realised that for the PBTL mono board, the two 10uH inductors are joined with the solder links to make 20uH ones. If I was swapping those, two 22uH ones would be cheaper and easier to fit in. I'm not sure I'll bother though. Maybe larger inductors would get the heat down. These ones I bought are 22mm diameter, the next ones up are 32mm.

Jamming these inductors onto these boards has made me think the capacitors will end up getting cooked over time. I might fit two caps either side of the board instead.


Edit: I'm running these on 19V. A YJ blue/black board I'm running on 15V runs a lot cooler, as do these Sanwus on 15V.


Attention Lost Eden! I've just found a vid on this 2.1 board which looks like it might have a proper crossover on it. There's bass volume, bass frequency, then treble freq and volume. There's a switch on there that offers "All Freq" and "Normal", so that could be a high-pass for the stereo amp. I don't think the treble freq knob sweeps the high pass point though.

YouTube

Fifteen pounds on Ebay.

HIFI 12V-24V 50W*2+100W TPA3116 2.1 Amplifier Board Digital Super Bass Speaker 788988565963 | eBay


Or Banggood TPA3116 2.1 DC 12V-24V 50W+50W+100W HIFI Digital Audio Amplifier Board Sale - Banggood.com
 

Attachments

  • inductor_swap.jpg
    inductor_swap.jpg
    410.7 KB · Views: 629
Last edited:
Hello all.

My first post in the amp section. Getting into this diy lark.

Just took delivery of one of these boards. Think it is a PBTL?

TPA3116D2 DC24V 2x 100W two-channel stereo Class D digital audio amplifier board
Online Shop TPA3116D2 DC24V 2x 100W two-channel stereo Class D digital audio amplifier board | Aliexpress Mobile to clipboard

It has the hiss at 50pc volume so I'm going to maybe swap the 50k pot out. I haven't been able to suss out the resistors for the gain yet and I also need to look into the mod that Fcossagg did to the slave/ master setting resistors to ensure the clocks are synced.

It's all a bit over my head to be honest but for the money I am pleased.

Does anyone know more about this board? The big black components either side the heatsink....one of them has the housing coming loose. Not sure what it is.

I managed to take the sinks off....they were just superglued on, hence the cyano white bloom over everything!
 
Last edited: