I recall that there was an audio power amp topology that used two transistors and two power supplies for each polarity output. Low voltage swings used a low voltage supply but it cut over to a HV supply for big swings, with diodes or something.
What was that called?
Class E through G, iirc. Class H makes the supply rail approximately track the output. (We use that for TEC controllers. )
Cheers
Phil Hobbs
About ten years ago someone asked me if I could repair an audio amplifier, which I did by replacing a shorted output transistor and a few associated components.
The topology was that a conventional class B circuit had the + and - halves of the output supplied initially by a few volts from a linear regulator followed by a regulator switching synchronously with the audio waveform to keep the collectors a few volts above the required audio output.
Unfortunately I can't remember enough to find an online reference. Not sure it was called anything specific.
class-G (or H) depending on who you ask
Looks like G or H. Thanks.
Cool. Thanks.
I need a power amp that runs near ground most of the time but spikes to higher output voltage occasionally. And runs 4 quadrant.
I have a switcher design that looks good, but it's slow.
I vaguely recall the G/H topology having someone's name too.
Our TEC controller is like that, 4-quadrant asymmetric bridge, but current mode. Too slow for audio, but that’s mostly parts values. Cheers
Phil Hobbs
My current design uses a TI DRV8962, which is a quad halfbridge. I'm using two sections to make a full-bridge bipolar class-D amp. It's a very cool chip, with lots of current limits and suppy lockouts and stuff.
One of those chips would make a nice 200 watt stereo amp.
But they only recommend running it at 200 KHz max. I could go a lot faster with a class G/H linear amp, of course at the cost of a heap of complexity.
Why does a TEC driver run in the goofy quadrants? Does the cooler ever act as a generator?
I'm designing a dummy load that looks resistive+inductive, like a solenoid, so it usually runs in the odd quadrants, where the user is pushing power into us. Sometimes we pretend that we stored some energy and have to return it.
Looks like a nice part for reasonably noise-tolerant applications. Class H allows us to use Class-A noise rejection with very good efficiency. Normally VDD is less than four tenths of a volt more than the differential output voltage.
Yup. Not a very good one, of course—max thermodynamic efficiency of a couple of percent.
Cheers
Phil Hobbs
The Proton AA-1150 is a nice example of this... 50 watts/channel continuous into 8 ohms, but it'll do short peaks of up to well over
100 watts/channel without clipping.Have something to add? Share your thoughts — no account required.
Ask the community — no account required