??? WHY WHY WHY ??? Class A vs Everything Else

The choice between Class A operation vs Class AB comes down to a simple decision between:

(A) Inherent low distortion and absence of device switching over the entire waveform, and also the least efficient mode of operation.

(AB) Significantly higher efficiency, greater power output (typically 3X or more) and an implied requirement for local or global feedback to linearize the switching transition in the output devices.

So if you’re building a transistor or tube power amp, the only difference from a design perspective is the rail or B+ voltage and the operating point for the output devices. Class A operation involves lower voltages but a high standing current ... in effect, the amplifier draws the same amount of power from the AC line, whether it is idling or at clipping. So it runs hot all the time. They are room heaters, thanks to low efficiency.

Class AB has the interesting property that AC power draw varies with signal level; this has implications for power supply design, since current flow through the supply is program modulated, and not in a simple way. This is a roundabout way of saying Class AB amplifiers need better power supplies, and full regulation isn’t a bad idea. Otherwise, 100/120 Hz rectification buzz in the audio signal will be program modulated, which is extremely undesirable.

There is an additional complication which applies to solid-state but not tube amplifiers. Tubes don’t need heat sinks, since device characteristics are not affected by operating temperature, and they are designed to radiate heat on their own without assistance.

Transistors can be destroyed by high device temperature, or have their useful lives shortened. The failure mode is complete destruction, which happens without warning, and might expose the loudspeaker to full rail voltage, which will destroy it.

In practice, transistor amps operating in Class A need substantial heat-sinking and effective temperature monitoring to prevent runaway thermal faults. These additional circuits must be highly reliable, since runaway thermal faults will destroy the output section before the user can get to the power switch.

In tube amps, Class A versus Class AB is a simple decision between power output versus linearity. Not much to it. In transistor amps, in addition to a significant power derating, Class A implies good thermal management if the amp is to be reliable.

In practice, transistor amps that operate in true Class A are thermally limited. How big a heat sink will the consumer accept, and are fans acceptable?

(Note: there are various sliding-bias schemes, which have been around since the Seventies, that claim to be Class A. They’re not. Non-switching is not the same as Class A.)

Ralph, I hate to disagree, but a pure Class B amp (like the Quad 405), switches from the upper set of transistors to the lower set with no region where both are operating. Class AB has an intermediate region where both are operating.

In a typical Class AB amp, whether tube or transistor, there are three operating regions: the plus direction, which activates the upper output transistors (or tubes), the middle zero-crossing region, which activates all devices at once, and a minus region, where only the lower devices are operating.

The size of the middle, zero-crossing region is at the discretion of the designer. If this middle region is so large that the B regions are never activated, it becomes a Class A amplifier (by default).

The Quad 405 had no A region, and relied on the feedforward system to supply current and voltage for +/- 0.7V region where all output transistors were turned off. As a result, it ran quite cold, but if you had a good enough distortion analyzer, you could see the switching region along with a spray of harmonics.

Class C is reserved for radio frequency transmitters ONLY. This has massive distortion since not all of the waveform is amplified ... there’s holes in it. It doesn’t matter in RF applications because tuned circuits filter out all of the harmonics, and Class C is more efficient than Class B or AB.

Class D is a switching amplifier, akin to a switching power supply. Pulse-width modulation converts the incoming analog signal into PWM, which is applied to very high speed switching transistors. A lowpass filter at the output removes most, but not all, of the ultrasonic grunge. It is normal to apply substantial feedback to linearize the PWM modulator and correct for small timing errors in the switching devices. (In PWM, timing errors translate into distortion when lowpass filtered.)

Here's a succinct explanation on Wikipedia:

Amplifier modes of operation

Keep in mind that nearly all audio amplification is done in Class A: Class AB or Class D stages are only used in the final stage of amplification to drive inefficient loudspeakers. (Speakers with a 1% conversion efficiency are actually "efficient" as most loudspeakers go. 0.3% to 0.5% are more typical figures.)

Let’s break down a typical amplification chain:

* DAC conversion element, whether R2R, FPGA, or single-chip conversion -> lowpass filter (class A) -> buffer stage (class A)

* Linestage: single gain or buffer stage (class A)

* Power amplifier, which are typically 3-stage, whether transistor or tube: input stage (class A) -> driver (class A) -> output stage (class AB) -> loudspeaker

Note the only element of this signal chain that is Class AB is the final stage that directly drives the loudspeaker. Everything else, whether tube or transistor, is Class A. In a studio environment, with far more complex electronics, the same applies: Class AB is only used for final drive for loudspeakers, and not anywhere else, no matter how complex.

The reason is simple: everything except for speaker drive is very low power, and operating at low currents (microamps to milliamps). So it doesn’t get very hot unless you have a big studio console, or a rack full of effects boxes. So efficiency doesn’t matter. When efficiency is unimportant, Class A is the invariable choice in professional and consumer equipment.

But ... if you need a 100 to 200 watts to wake up a loudspeaker, efficiency does matter. A lot. My home theater system has a 2010 vintage Marantz MM8003 power amplifier, with a THX Ultra certification for 140 watts/channel. That doesn’t mean it sounds good or anything, merely that it cranks out 140 watts from 20Hz to 20kHz at specified distortion ... that it works as claimed. So far so good.

Suppose this was a Class A stereo amplifier, instead of Class AB. As mentioned in the Wikipedia link posted above, 25% is the best efficiency we can hope for in a Class A transistor amps. So our 280 watt stereo amp must dissipate 280 * 4 = 1,120 watts, all the time, from the heat sink. Uh oh. I can tell you that is a gigantic heat sink, about the size of a medium-size HDTV. Also a pretty good room heater, that’s plenty warm any time the amp is on. Hope you have good air conditioning, because you’ll need it. Or ... the amp can be normal consumer size, or rack-mountable, but those fans are going to be moving a lot of air. Think gamer PCs, with top-of-the-line video cards, noise levels.

Many manufacturers claim sliding-bias Class AB is Class A. It isn’t. That’s marketing talking. True Class A is inherently inefficient and gets hot, whether tube or transistor. If genuine Class A is desired, versus inflated marketing claims, get used to 20 to 30 watts/channel with an amp that still gets plenty warm.

The bad actor is the speaker, not the power amplifier. A miserable 1% conversion efficiency comes out as 92 dB/meter/watt, or in other words, 100 watts of electrical energy is converted into one watt of acoustical energy (which is very loud). It gets worse: the other 99 watts do nothing but heat the voice coil.

Which isn’t good, because the voice coil is very small, with an area of only a few square inches, and is deep inside the magnet structure. For heat to escape, the voice coil must then heat the magnet, and the magnet, in turn, heats the air in the enclosure. Of all those expensive watts that come from our treasured amplifier, 99% are wasted heating the voice coil. And most speakers on the market are even less efficient, typically 0.3% to 0.5%.

In loudspeakers, there is a direct tradeoff between size and efficiency. Bookshelf speakers will always be inefficient. Sometimes called Hoffman’s Iron Law, it is also expressed as an equation in the Theile/Small formulas that define bass performance. No way around it.

So you have a big amp, or big speakers, or Class AB or Class D. You choose.