TDA 2030 getting to hot!

[basbr]

i have been working on a project that requires audio output.
to do this i have a dfmini mp3 player but to drive a decent speaker i included a TDA2030 audio amplifier.
i used the circuit from the data sheet and it works good. the only issue i have is that it gets really hot even with no speaker connected.
one solution i found is to leave out the bias resistors b1/b2/b3. this results in a lot of clipping/distortion in the audio but the ic stays cold.
the audio quality isn't hugely important. i only got a single 12V+ power supply on the board to power 2 5W speakers.

the board need to be relatively small because it needs to fit in a DIN rail housing, so i want to avoid a big heatsink if at all possible.

do's anyone have suggestions to make the TDA run more efficient or lower power? did i mess something up in the schematic? or any alternative mono amplifiers that run on 12V?
any input would be appreciated

schematic


V1.0


data sheet
https://www.st.com/resource/en/datasheet/tda2030.pdf

[wraper]

There only way that will reduce heating is reducing supply voltage and thus reducing maximum power. Heatsink is way too small, you need 10+ times larger than that. The only way to really improve efficiency is using class D amplifier.

[moffy]

As wraper said you need a larger heatsink, if you want a cooler solution you need a class D amplifier.

[NETeagle]

Hi, C4 and C3 are in wrong polarity and need to be reversed.
What is the purpose of C5 ? I think it is more likely to make the amplifier oscillate. Maybe it it is worth to try to remove it.

[mikerj]

the only issue i have is that it gets really hot even with no speaker connected.

I suspect the first couple of responders may have missed this, if it's getting very hot with no load then it may be oscillating.  Quiescent current at 14v is around 40mA from the datasheet, giving a Pdiss of around 0.5 Watts.  That's enough that it may feel a little warm with your relatively small heatsink but it shouldn't be burning hot.  I agree with NETeagle that C5 looks like a mistake, adding capacitive loading to the output of an amplifier makes it more likely for instability to occur, also fix the polarity of the other caps.  I don't see any local supply decoupling for the amplifier in your schematic, this could also lead to instability.  If you have omitted this from your design you could try tacking something like 100uf-220uF to the power pins of the TDA2030 to see if that helps.

[mikeselectricstuff]

As others have said - use a class D amp - plenty of options out there, small and efficient

[wraper]

the only issue i have is that it gets really hot even with no speaker connected.

I suspect the first couple of responders may have missed this, if it's getting very hot with no load then it may be oscillating.  Quiescent current at 14v is around 40mA from the datasheet, giving a Pdiss of around 0.5 Watts.  That's enough that it may feel a little warm with your relatively small heatsink but it shouldn't be burning hot.  I agree with NETeagle that C5 looks like a mistake, adding capacitive loading to the output of an amplifier makes it more likely for instability to occur, also fix the polarity of the other caps.  I don't see any local supply decoupling for the amplifier in your schematic, this could also lead to instability.  If you have omitted this from your design you could try tacking something like 100uf-220uF to the power pins of the TDA2030 to see if that helps.

TDA2030 requires Zobel network on its output. I guess 1uF electrolytic can act as a piss-poor version of that. Although ESR of 1uF electrolytic is a bit too much, and neither it will survive ripple current for long. And it should be located before 2200uF cap, not after.

[mikeselectricstuff]

BTW you can get some nice audio player modules with onboard class-D amps, content from either microsc card, or SPI flash ( programmed via USB - they open like a drive and you just drop MP3 files on, no sofwtare needed)

e.g.
https://www.ebay.co.uk/itm/193525989771
https://www.ebay.co.uk/itm/155384472764

[basbr]

hmmm. i chose that heatsink because i saw these modules floating around on the internet and thought, ill chuck in something similar and she'll be right
i fixed the caps in the pcb, but neglected the ones on the schematic, sloppy on my part.
i added cap c5 because it is in the datasheet behind a 1ohm resistor (C7).
ill have to do some research to see if i can incorporate a class D amp via a module.
thanks for the input!

[magic]

i added cap c5 because it is in the datasheet behind a 1ohm resistor

It's the aforementioned Zobel network required for stability and it should be copied exactly: 1Ω and 220nF film in series to ground, before the output capacitor.

[macboy]

i added cap c5 because it is in the datasheet behind a 1ohm resistor

It's the aforementioned Zobel network required for stability and it should be copied exactly: 1Ω and 220nF film in series to ground, before the output capacitor.

+1 to this. Amp is almost certainly oscillating.
Also where is the supply decoupling? Put at least 10uF electrolytic across the supply pins.

[basbr]

yep, that fixed everything!

a 1 ohm resistor in line with a 220fu cap before the 2000uf cap
I was sceptical that such a small resistor would do anything but hey, im a digital jockey, analog amplifier stuff is like magic to me.
so i guess oscillating was the problem?

Massive thanks to macboy and magic for pointing pointing out the problem. now the audio is crystal clear and the amp stays nice and cool.

this amp is part of a larger PCB where the supply decoupling is (not shown on the schematic).

[wasedadoc]

a 1 ohm resistor in line with a 220fu cap before the 2000uf cap

There is no unit of capacitance denoted by "fu". If that was a typo and you actually fitted a 220uF capacitor, that is 1000 times more capacitance than you should have.

[srb1954]

a 1 ohm resistor in line with a 220fu cap before the 2000uf cap

There is no unit of capacitance denoted by "fu". If that was a typo and you actually fitted a 220uF capacitor, that is 1000 times more capacitance than you should have.

"fu" is the unit of capacitance you use when put a electrolytic in backwards and it blows up in your face

[basbr]

your right, i accidentally added 1000 time the capacitance i needed. i swapped it out for the correct one and the results are the same surprisingly.
220 fu's is way to much, i only have so many
thanks for pointing it out

[Doctorandus_P]

You can get plenty of small Calss-D modules from Ali / Ebay / China, but al long as it's not battery powered, a few Watts should not matter much. (Although, even with a few watts of difference, you will earn a more expensive module back over some years of (constant) use, depending on who pays the electricity bill).

The proper way to proceed is to check for oscillations. Just bad PCB layout can for example be a cause of oscillations in audio amplifiers.

If you do not have an oscilloscope, you can build a simple circuit with an DC blocking capacitor, some diodes, resistors and capacitors. The goal is to rectify any AC oscillations, put them in a capacitor to measure them and use a resistor to discharge the capacitor if there are no oscillations. This should be easy to build on a breadboard and I'll leave the details for you to figure out.

As a beginners project, you can also make a more elaborate "AC milli volt meter" with active rectification and built in meter (or as DMM adaptor).

[GigaJoe]

if it idle the current around 20-30 ma, if input short. with your size of heatsink it about +30-35C
if it hot without any signal - it oscillation.

small and big resistor - give you K=10 , that on the edge of stability  and possibly oscillating, 2003 designed for  K about 22. 
as well as resistors 10K and 100K - very high in value , consider 680 and 680 X 20 = 12-15K approx ...  C3 at least 20uF + 1uF ceramic , c6 about same function , probably OK, but 10uF would be better ..

pin 5 and pin 3 -  power feed , should be capacitor least 200-300uf and ceramic 0.1-1uf  -  close to pins as possible ;   otherwise you get oscillation as well .

in overall chip designed to specific schematic and not so much tolerable to random value of element , getting oscillated ,  you need a scope to make sure it not ;  or at  least measure idle consumption current - in your case should be 30ma  approx,  if it more then 60-70 - most likely oscillation.

[magic]

your right, i accidentally added 1000 time the capacitance i needed. i swapped it out for the correct one and the results are the same surprisingly.

Too much is probably not a problem for stability, but it's an additional load on the amp which would increase power dissipation in the chip and the 1Ω resistor when playing high frequency at high volume. It can be a particular pain with sinewave testing.

However, electrolytic may be a problem for stability due to its inductance and the proper choice is film. The schematic does show a nonpolar capacitor here.
Case in point, it appears that the series combination of ESR/ESL of two capacitors was too much. Or maybe it was the lack of 1Ω.

[floobydust]

I've worked with the TDA2030 and newer LM1875
I had a TDA2030 (and LM1875) running hot and found it was oscillating at ~12MHz. Of course the electrolytic decoupling caps didn't help with that, so I added 0.22uF to at least clean up the rails.
They can be a real bear to stop oscillating. I found I have to add a small cap across the feedback resistor to limit bandwidth such as 47pF across the 100k "big1" OP has, has always worked well.
Gains below 24dB (Av=16) for TDA2030 (or Av<10 for LM1875) the IC's can go unstable. OP's gain is below that 100k/10k is Av=11 or 20.8dB I think so it could be the problem as well.