Hilariously cheap car audio amp (Cougar C-300.4)

[tom66]

Re: Hilariously cheap car audio amp (Cougar C-300.4)

I'm an idiot!

But, also, the manufacturer is one too.

I assumed (rather sensibly or not?) the REMOTE input was a 12V signal to turn on the amplifiers. But NO, it powers the TL494, which powers up the amp chips. I had a 1k in series to limit current (It simply said "to car stereo" so I made a guess), hence the 5.3V supply.

Now the amp powers up with a 17.5V supply.

Observations:
- Amp is now much louder. However, it still has a limitation of around 35W per channel. Better than before, but still not 75W per channel.
- I can get the amp to draw about 7.2A from a car battery. (Set up with the bench supply as a parallel 4A charger @ 14.4V.) However, the inductor/toroid converter starts saturating and whining, and the output clips at around 15V; it looks like a square wave. I imagine one channel could do 60W, but if you tried two channels, the combined limit would kick in.

Thermocouple says amp IC hits 70C on tab, FETs run relatively cool at 55C, inductor gets pretty hot (hard to probe but around 65C.) I'm not sure if there is a current limit on this converter as I can't see a sense resistor. Could be using FETs. Each FET gets a different gate drive which looks ugly (will upload pics later.)

Got annoyed with the sound from it, so using a 4 ohm rheostat for now.

[tom66]

Re: Hilariously cheap car audio amp (Cougar C-300.4)

OK, started tracing through the circuit for the SMPS. Definitely some type of push-pull converter. Feedback is accomplished through a 15V zener diode, so it has regulation. Unfortunately, as the blue trace shows in the below video, it is simply incapable of regulating under any substantial load. I don't even want to think how it would cope with two loaded channels, let alone all four - in fact I did notice when running before with the 1k how badly it coped with four channels, it was very quiet. At peak amplitude, it's not much better than a ~15.5V input, a high side car battery gives 14.4V so it's a lot of expense for no real gain; and with all four channels loaded? Who knows! It could be even worse than just running off the battery, as it is a semi-isolated topology (with shared ground.)

Wonder how this could be improved - could it simply be the toroid is incapable of supplying the demanded load? Maybe better feedback compensation? Better output capacitance? (currently just a single 2200u 25V "Jyco", probably not even low ESR.)

http://www.youtube.com/watch?v=Ljy23...ature=youtu.be

[tom66]

Re: Hilariously cheap car audio amp (Cougar C-300.4)

OK, here's the traced SMPS. Couldn't trace the base of Q5. It looks like the top blue traces are just silkscreened on; the actual traces are on the bottom side. Cut one of the blue "traces" and I still get a solid short across it.

[tom66]

Re: Hilariously cheap car audio amp (Cougar C-300.4)

Fixed an error - thought that string of resistors was strange!

I've noticed this converter seems to mostly operate at around 5% duty cycle, and under full load about 20%. Seems like a bad design to me.

[Th3_uN1Qu3]

Re: Hilariously cheap car audio amp (Cougar C-300.4)

Yes, the top traces are just silkscreened, this is a single sided board.

A few observations: Are you really sure C4 is 100p and not 1n? With 100p, the frequency would be 733kHz, the TL494 can't even do that. Most likely C4 is 1n and the frequency is 73.3kHz.

That's a really weird arrangement with the zener there. Also, what Q5 does is really interesting. Assume Q5 is off (open switch), then the divider to the inverting input of the error amp in the 494 is formed by R12 and R16+R17, which gives ~4v Vref. With the transistor on, you get also R4 in parallel with R6+R17, which gives ~2.2v Vref. It appears this converter has a switch that "steps" the output voltage for some reason. You should check the C-E drop of Q5 during operation to see whether it's on or off, if it's on, you should see near zero since the currents involved are very small.

[tom66]

Re: Hilariously cheap car audio amp (Cougar C-300.4)

C4 is marked 102. Now I realise it's the first two digits, times 10^multiplier; in other words, 1n. The snubber cap is also wrong, I'll update the schematic shortly. And take some pictures of the waveforms.