Class D amp and the PSRR

[rtarbell]

Just an idea:

One disadvantage of Class-D amplifiers is that they have 0dB PSRR (the level of the power supply rail is directly associated with the gain of the amplifier, so any hicup in the line will be seen at the output). This is not necessarily a bad thing...

==> I was thinking, if one were to put a modulated power supply rail for a class-D amplifier, you could make a kind of multiplier. The input signal to the class-D amplifier would get multiplied by the modulated signal on the power supply rail of the class-D amplifier. I know that there may not be a big demand for high power multipliers (many class-D amps are higher power than a single multiplier chip), but it's just an idea...

Would it work?

[phase_accurate]

Easier than using a multiplier is making the tringle wave proportional to the supply voltage which can be implemented easily.
This works only for carrier-based PWM however and it is already patented.

Regards

Charles

[classd4sure]

Originally posted by phase_accurate
Easier than using a multiplier is making the tringle wave proportional to the supply voltage which can be implemented easily.
This works only for carrier-based PWM however and it is already patented.

Regards

Charles

Hi Charles,

Seems to work with self oscillating ones as well according to the reference design thread where the feedback was tied to the rails with a voltage divider.

Can you tell me who patented this offhand? Just curious.

Thanks,
Chris

[phase_accurate]

Can you tell me who patented this offhand? Just curious.

Will try to find it again.

Regards

Charles

[phase_accurate]

Here you are:

http://v3.espacenet.com/origdoc?DB=E...&QPN=EP0503571

Regards

Charles

[classd4sure]

Hi,

Thanks Charles,

Chris

[dmfraser]

Not all Class D amps have 0 db PSRR. Only the non-feedback designs like those using the TI chips. I use many of the ICE Power modules in my work and they have PSRR essentially as good as a Class AB amp. The UCD amps have great PSRR as well. However, the amps I use from D2Audio, have no PSRR and need a regulated power supply.

So remember that class D does not automatically mean there is no PSRR. You have to check.

[Gorilla]

Correct me if im wrong, but unless some sort of feedback path exists wouldn't any amplifier designed with any topology have 0db PSRR?

[hermanv]

Originally posted by Gorilla
Correct me if im wrong, but unless some sort of feedback path exists wouldn't any amplifier designed with any topology have 0db PSRR?

No. Visualize a push pull emitter follower driving a load, even though the plus and minus rails may have ripple the output will have very little PSR if the bases have a clean signal. Other topologies also reduce PSR by varying amounts.

Even the simplest class D will have significant reduction if plus and minus rails that have mirrored ripple (the normal case) are used.

[dmfraser]

A non feedback design with a H Bridge output can have some PSRR as both sides go up and down at the same time. However the amount or rejection depends on the match of the Rds and several other factors. Regardless, it will be poorer than a feedback design and a regulated supply should be used with a non-feedback class D amp.

The point I was trying to make is that one cannot jump to the conclusion that because an amp is class D that it had bad or no PSRR. The two do not go together automatically and that the PSRR is an independent spec that must be checked before building one design or another.

I have used both methods in my five years as a Class D amplifier designer and I find both methods have their sonic faults and favors. And the feedback and non-feedback designs definitely sound different if compared side by side with an audiophile grade system surrounding it. However, which is superior is, as always, a matter of personal taste.

However the feedback designs tend to measure better and my experience is that which measures better usually, though not always, sounds better.