diyAudio (
-   Tubes / Valves (
-   -   [Theory] Amplifier efficiency (

Giaime 18th August 2006 01:16 PM

[Theory] Amplifier efficiency
Hello all!

I've got some questions about theoretical maximum efficiency of audio amplifiers.

Starting from the basics: efficiency = Output / Dissipation, or better, the ratio between the available output power and the "lost" power in thermal effect in a quiescent state.

For example, a single ended class A tube that's dissipating 20W and has 5W output, has 25% efficiency. Let's keep the math simple forgetting about the heaters.

So, there's no difference about SE or PP when in class A, because in the latter, output power doubles and quiescent dissipation doubles as well.

But... what's the maximum efficiency for class A? And what about class A2? The texts says 25% for class A at the oneset of clipping, but what about pushing tubes in positive grid bias region?

And what about class B and AB? 78%? And in AB2???

So, does operation with grid current matters? I will be grateful if some expert enlightens me.

Thank you very much in advance :)

Ps, this all started when I heard an EL34 PSE in triode mode, that declared 20W output power. If the math is correct, that's just impossible.

EC8010 18th August 2006 02:41 PM

The maximum theoretical efficiency for Class A is 50%. That assumes a transformer load but makes no distinction about device - transistor or valve. Similarly, the 78.5% efficiency you quoted for Class B assumes a transformer load and is irrespective of device. Practical maximum efficiency of a Class A1 triode is generally about 20%. A2 would allow you to approach 50% at the expense of dissipating lots of power in the necessarily beefy driver stage. 20W out of a single EL34 is impossible, but you mentioned PSE, so how many EL34 were used?

Giaime 18th August 2006 07:42 PM

Hello EC8010,

thank you very much for your reply. The amp I was referring to had 2 EL34 per channel. I estimated an output of 12W, without exceeding any tube maximum rating.

I have another question: what happens when the load is resistive? For example (a *rough* approssimation), let's consider an OTL output stage, or better, the classic transistor output stage. I think that the efficiency goes down, am I right?

Anyway thank you very much.

tubetvr 18th August 2006 09:16 PM

The definition of efficiency is the power dissipated in a load divided with the input DC power. In all practical cases you always assume that the load is pure resistive or that the power is dissipated in the resistive part of the load. I think it is more likely or probable that the output power will be max with a pure resistive load and could be lower with a reactive load.

An EL 34 have a max anode dissipation of 27.5W according to Philips datasheet so 2 have 55W so the output power could theoretically be 50% of that but more likely it will be 20 to 35% of that, (35% is the highest efficiency figure I have seen in a textbook for a practical push-pull class A amplifier)

35% of 55W is 19.25W so it is very close to the 20W you mentioned, then you can of course always exceed the anode dissipation slightly.

Important to note is that the operating point giving max output power is not the point of best linearity.

Regards Hans

EC8010 20th August 2006 07:09 PM


Originally posted by Giaime
I have another question: what happens when the load is resistive? For example (a *rough* approximation), let's consider an OTL output stage, or better, the classic transistor output stage. I think that the efficiency goes down, am I right?
Hello Giame,

as tubetvr says, we always assume a resistive load, but I think what you mean is, "What happens when we don't couple that load using a transformer?" And yes, you're absolutely right, the efficiency drops even further, so Class A transistor amplifier has even poorer output stage efficiency than a Class A valve amplifier using an output transformer (ignoring the heaters). And as for valve OTLs...

Giaime 20th August 2006 07:12 PM

I asked that because getting 25W pure class A without any form of feedback from 4 EL509 in OTL, seemed a bit unrealistic to me.

EC8010 20th August 2006 07:23 PM

The killer with OTLs is the load resistance. They often quote power with a 16 Ohm load. 25W Class A from 4 x EL509 sounds a bit unlikely to me. On the rare occasion that I've contemplated an OTL, extracting any power at all has always been a design struggle.

All times are GMT. The time now is 05:08 PM.

vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2017 DragonByte Technologies Ltd.
Copyright 1999-2017 diyAudio

Content Relevant URLs by vBSEO 3.3.2