Power stage topology

I am looking into power stage topologies.

Currently there are two alternatives that I am aware of.

Full H-Bridge and a Half bridge.

The both seem to be as good as each other but the half bridge requires a -ve supply where as the full bridge only requires a 0 to +ve.

I have also read some information relating that with a full bridge you only need half the maximum voltage to produce the same output power. This doesn't seem correct as I thought the magnitude of the voltage would be the same for each topology.

If anyone can help I would be greatfull.

Does anyone have proof of better quality sound from either topology. I am planning to opted for the best quality sound.
 
Topology

Half bridge is good because you have no DC voltage on the output. Plus, you can later bridge for more output. Somewhat cheaper to build. This is what Hypex do.

H bridge is good because you only need a single supply, of a lower voltage for a given power output. Also, the balanced output stage cancels out many of the distortion products and has what some describe as a mellower sound, somewhat "tube like" but without the output transformer artifacts. It also cancels out much of what noise or ripple is on the power supply. This is what ICE Power and D2Audio use.
 

BWRX

diyAudio Moderator Emeritus
2005-01-17 5:29 am
Pennsylvania
frozenpod1 said:
I have also read some information relating that with a full bridge you only need half the maximum voltage to produce the same output power. This doesn't seem correct as I thought the magnitude of the voltage would be the same for each topology.

If anyone can help I would be greatfull.

Look at the otuput stages of both the half bridge and the full bridge.

The half bridge has two transistors with one transistor tied to the + rail and one to the - rail. The speaker output (forget the filter for a moment) is the connection between the transistors and ground. Theoretically, this allows the voltage across the speaker to be as high as the + rail or as low as the - rail, with respect to ground.

The full bridge has one + rail and a ground and two pairs of output transistors tied between the + rail and ground. The speaker is connected between the outputs of the transistor pairs. This allows one pair of output transistors to pull one side of the speaker to the + rail while the other output pair simultaneously pulls the other side of the speaker to ground. This allows the voltage across the speaker to be as high as the + rail, or the same voltage swing as in the half bridge scenario but with only a + supply rail.

Try to find a picture of what I described, it's much easier to understand when you see what's going on.

edit: Maybe I misunderstood the meaning of your question. By half the maximum voltage they probably mean only one supply rail instead of positive and negative supply rails.
 
And much simpler to compensate.

Hi Lars

What do you actually mean by compensate ? Are you talking about the carrier suppression ?

I ask because there are actually more than one things that can be/must be compensated/supressed.
E.g.

- supply pumping: Easier on full-bridge i.e. non existent
- NFB: Easier on half-bridge
- Output filtering: Easier on half-bridge

Regards

Charles
 
frozenpod1 said:
I understand exactly how both works I have built several switching h-bridges for other power applications

I was trying to find out if anybody nows which produces a better quality of sound.

Has anyone done comprehensive testing of each topology using the same feedback and control loop.

I have and I get *exactly* the same sound from full or half bridge.
 
janneman said:
Maybe not 100% OT, but this article gives an interesting overview of the issues, including power supplies for class-D:

http://www.edn.com/contents/images/526331.pdf

Jan Didden

This article shows how much digitally controlled class D amps are all about creating real analogue problems (power supply sensitivity, jitter sensitivity etc) only to solve an imaginary problem (the DAC and analogue signal processing).

A simple but well designed analogue class D delivers performance (measured and sonic) that makes this article's subject a joke, contains no fancy noise shapers or filters and does not need anything special by the way of power supply and is cheap to build.

I do use noise shapers and digital filters extensively, but only in DAC's. It's always better to do the D/A conversion in the small signal domain instead of trying to pull it off using power FETs.