are isolated supplies needed for high-power class-D?

[darkfenriz]

And how about power factor?

 

[Pafi]

Who cares? (However rectification, PFC, and idle potential are 3 different things, so I don't see the relevance.)

 

[iand]

And what are the technical benefits of non-isolating a high power class D amp?

I assume that the amp's outputs are close to earth at idle, hope this is undisputable.

In the end you need around +-100 to +-150V for the full bridge audio PWM stage, capable of several dozens of amperes at least. As far as I understand to achieve that you may use either:

1. boost PFC, full bridge transformer coupled H-bridge PSU, audio amp (isolated)
2. active front end (AFE) PFC for symmetrical rails, symmetic bucks, audio amp (complicated)
3. AFE PFC and duty cycle limited audio amp (you'll need 1000V switching transistors for audio)

So why bother? My feeling is that this is an idea inspired by the synchronous motor driver circuits, which use nonisolated approach for obvious reasons, one can quarrel taht a speaker actually is a motor, but...

For really high power outputs the only sensible approach is to use fast-switching 600V MOSFETS (CoolMOS) with series Schottky diodes and antiparallel hyperfast diodes -- preferably in one package like this:

http://ixdev.ixys.com/DataSheet/IXKF40N60SCD1.pdf

These can run straight off a centre-tapped half-bridge with up to 500V across it (+/-250V relative to centre tap), and all this then needs is a single-stage PFC boost stage -- this can use CoolMOS and SiC diodes to get very high efficiency (much better than a 2-stage PFC/PWM).

Put 4 of these IXYS devices in parallel for high and low switches (8 devices total) and you have a half-bridge amp which can put out 100A and 225V peak, and can be bridged if you need truly ludicrous single-channel power -- hang on, this sounds very like the Powersoft K20...

Combine both these and you should be able to exceed 90% wall-to-speaker efficiency, which is what you want to do for really high-power amps -- 95% would be even better...

I wouldn't be at all surprised if this is also what Eva's working on

Ian

P.S. Outputs don't need to be close to earth at idle or any other time, since you won't be able to touch them it doesn't matter...