Class H amps?

Just came across 'Class D' (after years of A, AB, not really B). 

So what is Class H? How does it differ from Class D?



The output transformers are valves between 2 voltage rails. Say, + and - 10 V.

In A, B and A/B those rails are fixed values. Classes G and H vary those rail voltages. Carver and NAD got famous on G. The idea is that minimizing the difference between the rails and raising them only when needed minimizes heat of the output transistors.

NAD continues to dabble with their hybrid Class D amps today.

Wikipedia is your friend:


As an aside, an audio buddy is using a pair of Benchmark AHB2 Power Amplifiers to drive his SoundLab electrostatic speakers to great effect.

Benchmark's specs state it: “combines class-AB, and class-H topologies, using a feed-forward error correction system” (  My buddy is using two AHB-2’s that are in bridged mode to output enough power for his SoundLabs.

H is good.


D is a muddy mess.


My Sunfire sig 600 is class H, and smokes much if not most of the newer amps at lo local shop. They are sick of me showing up with it, and besting much of their high class amps. Im thinking's soon they will ban people from bringing in their stuff for speaker auditioning.


It’s all in the implementation. Making a generalization about amps by class is . . . well . . . unfounded. It’s like saying all blond(e)s are dumb. Every one I’ve met is different.

I’ve heard the AHB2 (class H) in my own system (with Janszen electrostats), and it sounded great.

I’ve heard Apollon amps made with PuriFi 1ET400A modules (Class D) in the same system, and the sound was great also.

It’s true that some early Class D didn’t sound so good. Development has been intense, and it’s paid off. The good ones are more than competitive with Class A and AB. Choice then becomes a matter of individual taste.

@arcticdeth arcticdeth "D is bad, H is good". That is the analog vs digital opinion just played on a different field? 

Just to clarify: I only wanted to know the basic differences and general assumption. And like with the old analog vs digital, it is likely 'cheap D is better and cheap H' (here: $100 CD player beats $100 turntable). 

So if Class H is still analog, what is BENEFIT over A/B. Just power consumption/weight? I get the benefit of full Class D (big power, low consumption, low weight, cheap) and why it is used in applications where those advantages are very benefitial (car amps for example, portable electronics, etc). 

So if Class H is still analog, what is BENEFIT over A/B. Just power consumption/weight?

Yep, attempt to get to Class D efficiencies of manufacturing as well as power usage without the switching issues.

If the linear amp is kept in Class-A, then the efficiencies over a Class A are really outstanding. AFAIK, they make no claim that H is better sounding than A or A/B, but as good, with much higher efficiences.

The point made elsewhere, that it’s all in the implementation, is really important.

It may help to understand that due to FTC rules, most amplifiers have to be heavily overbuilt.  FTC rules require pre-heating at (I think) 30% of the rated power for an hour before testing for noise, distortion and output.  This is a situation never encountered in music.  If you have a 300 W amplifier, you don't play music at 100W continuously.  To meet the FTC requirements an amp must have excessively over-built heat sinks, which adds so much to manufacturing and transportation costs.  Anything that reduces the weight of an amplifier not only makes it cheaper to make, but greatly reduces the carbon footprint. 

Can an analog Class H still have (would make sense) a switched mode power supply (LOTS of weight saving right there)? 

Can an analog Class H still have (would make sense) a switched mode power supply (LOTS of weight saving right there)? 

Absolutely.  This seems to be what the NAD Hybrid amps are doing, but they have very very little information out in the public domain.

Yamaha's EEEngine amps also have this type of design. 

One engineering difficulty about Class H though is the voltage rails have to anticipate the signal peaks.  There's some interesting discussions online about Yamaha's implementation, distortion, and fixes.  Class G solves this problem by having multiple rails available all the time, but this adds to the possibility of distortion when those rails switch.