# Class AB formula

#### cobretti

I want to ask about simple class AB 2x EL34 power amp operation. Lets say I have 400VDC plate voltage and both EL34 are biassed to 40mA. So each tube is dissipating 16W.

Does it mean that amp will be operating in class "A" up to 22W roughly?

#### DF96

That depends on the details of the valve curves. If you assumed that everything is linear and unconstrained then 32W quiescent dissipation Class A means up to 16W output power. Real life is not so simple. You will need to see how much power is being developed by the valve turning on, at the point where the other valve is turning off.

Does it matter?

#### lcsaszar

Under similar conditions I couldn't manage to get more than 12W.

#### PRR

Paid Member
....each tube is dissipating 16W. Does it mean that amp will be operating in class "A" up to 22W roughly?

Ignoring specifics: a speech/music class A amp is at best 50% efficient. 2 * 16W is 32W, we can only expect 16W in A. In real life, less because tubes won't pull to zero, but more because tube Gm is curves and the up-swing is bigger than the downswing. Which means we always rise "toward B", but we don't think of that way.

However your 2xEL34 amp is probably loaded for big power in B, which is not optimum load for A. lcsaszar's result of 12W in A is more likely than the abstract 16W.

And yes, does it matter?? Many AB designs are exceptionally clean. The curvature of tubes does not give the harsh crossover of simple BJT amps (or unpopular zero-bias B tube amps).

#### cobretti

Ignoring specifics: a speech/music class A amp is at best 50% efficient. 2 * 16W is 32W, we can only expect 16W in A. In real life, less because tubes won't pull to zero, but more because tube Gm is curves and the up-swing is bigger than the downswing. Which means we always rise "toward B", but we don't think of that way.

However your 2xEL34 amp is probably loaded for big power in B, which is not optimum load for A. lcsaszar's result of 12W in A is more likely than the abstract 16W.
And yes, does it matter?? Many AB designs are exceptionally clean. The curvature of tubes does not give the harsh crossover of simple BJT amps (or unpopular zero-bias B tube amps).
No it doesn't matter, at least to me. I just wanted to know the power behavior of
50W JCM800 for example. So it would be around 12W. Playing low volume, the amp is probably in class "A", that's all.

#### Printer2

And yes, does it matter?? Many AB designs are exceptionally clean. The curvature of tubes does not give the harsh crossover of simple BJT amps (or unpopular zero-bias B tube amps).

I just meant if you are making 12W as compared to 16W. I have said in other forums our amps play clean mainly in Class A unless we get into full output.

#### Gnobuddy

The curvature of tubes does not give the harsh crossover of simple BJT amps (or unpopular zero-bias B tube amps).
I remember seeing a design for a pure class B, BJT power amp in an ancient issue of Wireless World.

IIRC the design concept was that it was hard to get two BJTs to produce a well-behaved transconductance through the crossover region, so the designer decided to have the output transistors turn hard on and hard off, by biasing them to a small DC quiescent current, and driving them from a current source via a pair of small-signal switching diodes. The claim was that the current transfer function of the diodes produced a clean sharp break between on and off, unlike the transconductance curve of the output BJTs.

Although the two output transistors had a small DC bias current, the circuit was designed so that each of them conducted over exactly 180 degrees of the input signal, so this was indeed pure class B, and not AB. There was no overlap region with both transistors contributing to the output signal.

There are a number of oddities with this design approach (such as the fact that the current source output has to slew across two diode drops worth of voltage to switch from one diode to the other). Clearly this particular approach to solid-state power amplifier design was an evolutionary dead-end which never went anywhere.

All the same, this was just the sort of thing that made old Wireless World back-issues fun to read. Electronics wasn't yet an entirely mature field of engineering then, as it is now. So there were still new ideas being tried out from time to time.

Nowadays it's pretty much just "Throw a microcontroller in a box and write some C++ code". Most of the time, the code is fairly boiler-plate, and not particularly interesting, either; read this push-button, blip that pin logic high, write this ASCII character to the LCD display, yada, yada, yada.

-Gnobuddy