PWM amp with modulated rails

[earthtonesaudio]

I like the idea of building this high efficiency amplifier:

-Audio converted to a PWM waveform
-PWM drives output transistors for Class-D operation
-PWM also used to control a switchmode power supply, so the rail voltages supplying the output transistors are modulated. This part is basically the same as "envelope tracking" or Class G/H amplifiers.

The goal being high peak audio power output (and high efficiency) from a low voltage (battery?) power supply, and reduced standby consumption.


I want to try to build a small scale experimental version on the breadboard first (using low power transistors) but I don't know how informative that will be, since the switching speed will be greatly limited.

Also I'm unclear on what special considerations one should take to measure output power of a class D amplifier...?

Any comments or criticisms welcome!
Thanks,
Alex

[Steve Conner]

Hi Alex

Class-D makes the rail-tracking power supplies redundant. A Class-D amp is a buck converter, so it converts the rail voltage to the required audio output voltage with high efficiency anyway. You just have to make sure the rails are high enough for the peak output voltage you want. They're theoretically 100% efficient irrespective of rail voltage, unlike a linear amp.

Output power of a Class-D amp is measured exactly the same way as any other amp.

I've seen many hobbyist designs for self-oscillating Class-D amps, and I know a guy who built a 500-watt one (with a switchmode power supply, no less!) for his bass rig. Self-oscillating seems to be the way to go.

I personally can't be bothered with the things: I'm already lugging a speaker cabinet, so an extra 5 or 10 pounds of amp doesn't really make a difference. For small battery amps, it might. You can investigate ready-made chips, there are lots available for TVs and so on, from TI, Tripath, and the like. A naively designed DIY Class-D would probably have higher idle consumption and worse sound quality than a Class-B transistor amp.

[earthtonesaudio]

Hi Steve, thanks for the info!

The inspiration for this came from an experiment like you described, a self-oscillating class D amp with a switchmode power supply. The problem I discovered was that the oscillator for the amp and the oscillator for the power supply would radiate into each other and create beat frequencies and distortion. It's possible to reduce the problem with careful layout and/or shielding, but I wanted to try a more robust design to eliminate it altogether.

Using a single PWM signal for both the amp and rails could eliminate the possibility of heterodyning, and modulated rails are then just a side effect.

I think it would be beneficial because it would increase efficiency. For a fixed supply voltage, class D amps are most efficient at maximum power output, but their efficiency drops off dramatically at lower output power. Using a low-level input signal to turn down the power supply reduces switching losses. Granted, this won't make the amp louder, but it might make it more lightweight. The heatsinks then have a chance to cool down more than they would with a fixed supply, so they can be smaller.

Anyway this is still just speculation so your opinion is as good as mine.

[Steve Conner]

I think it would be beneficial because it would increase efficiency. For a fixed supply voltage, class D amps are most efficient at maximum power output, but their efficiency drops off dramatically at lower output power. Using a low-level input signal to turn down the power supply reduces switching losses.

It reduces switching losses in the output stage, yes. But by the same argument, the efficiency of the switchmode power supply will drop off as you turn it down.

The size of heatsinks in Class-D amps is negligible. I had a 400 watt Coldamp one that needed no extra heatsinking. It did need forced air cooling, though, and died a flaming death on the test bench when I forgot to turn on the cooling fan. But from what I've seen, Coldamp's products just liked to catch fire and explode anyway.

Brian Taylor's "Power Mosfet Design" describes a SMPS and Class-D amp combined into a single device for maximum efficiency. He calls it an "Ampliverter" IIRC.

[ame=http://www.amazon.com/Power-MOSFET-Design-B-Taylor/dp/0471938025]Amazon.com: Power MOSFET Design (9780471938026): B. E. Taylor: Books[/ame]

[Nate]

Brian Taylor's "Power Mosfet Design" describes a SMPS and Class-D amp combined into a single device for maximum efficiency. He calls it an "Ampliverter" IIRC.

I've always thought a SMPS + Class D amp seemed redundant.