500W... O RLY? #
Kinter MA-150. Image credit: ebay
Let's say that you want a small, cheap amplifier for automotive use, and you go and look on eBay. At the time of writing these Kinter MA-150 jobbies are available for around £8 direct from China, or around £10 from UK sellers. Sure you think, that "500W" and the accompanying "150W RMS" is going to be an outright lie (if you don't think this, well, you need help that I'm not able to give but see note 1 anyway), but it's a neat enough package and it should be able to do what a cheap head-unit can do, right?
Well, you're half right. The amplifier module that it contains is an STA540, which is a perfectly capable chip from ST and considering that you would probably pay £3 for the chip alone, this is a decent enough way to get one mounted on a board with a little aluminium case that will serve well enough as a heatsink.
HOWEVER, for some inexplicable reason the PCB layout is so poor that the Kinter pretty much has 3% distortion from low levels all the way to the onset of clipping, at which point of course it gets much worse. Okay so it works, as in sound comes out when you hook it up to some speakers, but that sound isn't very nice.
The problem is all to do with grounding; they have created a nice loop of low-quality ground between the two ground pins of the chip (power and signal) and connected the supply voltage decoupling cap to the wrong leg of that loop. This is allowing nasties introduced by the output stage to be coupled back into the input.
To fix: #
Cut the grounds where they enter the signal area (the end nearest the phono sockets) and make your own connection to pin 8 of the chip, going via the ground pin of the supply voltage rejection cap (C12). Cut the ground that runs down the side of the chip. Beef up the ground that arrives orthongonally across the middle of the board to pin 9, and join the two ground pins together right at the chip (I soldered the legs together right under the chip but I suspect that simply soldering the pads together will do just as good a job).
Bingo, onset of clip is now 8.17V RMS at 0.1% distortion with a 14.7V supply (really a 14V supply as far as the STA540 is concerned as there's a diode in the box), which is about 16.6W into 4 ohms. That'll do nicely. For reference, this level is achieved with 450mV RMS signal at the input to the DC blocking caps.
I have also changed the DC blocking caps to 680nF polyesters, stripped off all the un-needed parts (filter and its switch, USB socket and its regulator) and added some more supply capacitance though to be honest that didn't make a huge difference. I may also remove the pot and replace it with some precision resistors, as the chip's 26dB of gain is always going to be more than I need and the pot is quite high impedance. Add a little dab of heatsink compound to the back of the chip as you reassemble to help transfer heat into the case, the factory didn't bother with this which is a bit cheap of them.
All testing done with a dScope, bench supply set to 14.7V, and a resistive 4R load.
1. How do they get to 150W RMS? I suspect it's like this: If we discount all thermal considerations, feed the thing with a stable, high-current 22V supply (which is right on the limit of blowing up the amp chip and well in excess of the 16V power supply capacitor contained within), and run both channels with maximum-level square waves into 4 ohms, the total power transferred into the load is right about 150W. I suppose this is fine if you want to use your speakers as heaters, but it doesn't really apply to real audio and anyway you can't get a stable 22V supply from your car. How do they get to 500W? I have no idea whatsoever.