What every Class AB builder needs to be constantly aware of

[amplifierguru]

Class AB amplifiers are so popular because they can produce substantial power output far cheaper and without the size weight and cost penalties of class A. They also consume substantially less power than equivalent Class A and as such are substantially more environmentally friendly.

But designing one is not as simple as just turning down the output stage bias of an equivalent Class A design. Many of these are the picture of simplicity and it's relatively easy to get a good sound and that's why they've become a fallback position for some manufacturers unknowledgeable of how to deal with the distortions generated by the output stage switching and end up in desperation turning the bias up maybe half way, fitting huge heatsinks, continuous duty regulated supplies and promote largesse as the way to audio nirvana.

I worked for one such company and it was a real eye opener. In my short time there I designed and prototyped a lightweight low bias Class AB power amplifier from parts lying around there and it blew those high bias behemoths away. How did I do it? Simple.

I had this in the FRONT of my mind when designing the topology, the PCB layout and the chassis geography.

[amplifierguru]

Don't know what went wrong the zip file was almost as large as the JPG. This ones a BMP zipped.


This is what each half of the output has to deliver to the load in producing a simple 1KHz sinewave output from the amplifier.

[amplifierguru]

Thanks guys.

Now with each output half having to deliver half the waveform, just look at the spray of harmonics that are generated in and around the outputs and power supply lines

[lumanauw]

How did I do it? Simple.

What's the schematic looks like? Is it something different than ordinary low biased classAB looks like? Or it doesn't have to do with schematic, just pcb and chassis geography?

[amplifierguru]

Lumanauw,

It's not a specific topology. It's more about approach. It's about awareness that such voltages and consequent currents are lurking everywhere around the output stage and drivers radiating fields and modulating power supplies. They are there waiting at the periphery of your design looking for a weak point to intrude.

And most people give them numerous opportunities starting with the simple miller comp Vas. Straight down the PSRR route or in through that too close input stage or coupled through that sloppy wiring loom.

It pops up on the spectrum analyser as harmonic emphasis as you juggle the wiring loom for best results.

[mikeks]

I think you'll find this problem was addressed by Cherry in one of his early AES papers...

[amplifierguru]

Thanks Mikeks . I have Cherrys paper and he does deal with the issue of inductive stray pickup of such hash, one of the issues.

[mikeks]

so...what is your solution...

[lumanauw]

Hi, Amplifierguru,

I'm very curious. What's your design (that you consider as good) looks like. Could you post an example here?

[MikeB]

Don't you get easily rid of this by using a full H-Bridge ?
At least none of these half-waves are disturbing the psu this way,
as the load is always symetrical.
Of course you still have these pulses in the wires to the outputbjts...
Does it help placing a big cap from collector of upper device
to collector of lower device ? (For EF) Very close of course...

Mike