Universal Audio WebZine
Volume 2, Number 7, August 2004
[UA Universe] [Ask the Doctors] [Digital Discourse] [Analog Obsession]
[Support Report] [The Channel] [Digi-Talk] [Analog Dialog] [Featured Promotion]
[Graphic-Rich WebZine]
[Back Issues] [UA Home]

Ask Professor Vacuum Tube: What's All this Class-A Crap, Anyway?

Who IS this guy?!? Where are my regular doctors?
Q: "What's all this class-A crap anyway? What does that really mean?"

A:Well, my goodness, I think you mean, "gee whiz Professor Vacuum Tube, would you please tell us about the various classes of amplifier operation". I'd be glad to. In fact your question brings back fond memories of the days when vacuum tubes roamed the earth as masters of their world. They were magnificent beasts, the triodes, tetrodes, pentodes (even the occasional nemnatode) and their roars echoed throughout the land. Alas, they are nearly extinct today, but small pockets are known to exist in what are now Russia, Eastern Europe, and the Far East. We believe it was the Great Silicon Meteor Showers of the 50's that brought them to the brink while the feisty little Bipolar and Field Effect Transistors were better suited to a changing environment. We haven't even unearthed other suspected species like the hexode, septode, and octode. But, I digress. Let's talk about amplifier classes.

Linear Operation, Cutoff, and Saturation
Consider a typical audio input signal as shown in figure 1. It's actually a snip of the last note of a Doug Kersaw fiddle riff, just before he slaps the bass player silly, but that's not important. We show 2 cycles of the input voltage to an amplifier and will use a single cycle of it to demonstrate the bias point, and conduction angles for various classes of amps.

"We believe it was the Great Silicon Meteor Showers of the 50's that brought (tubes) to the brink while the feisty little Bipolar and Field Effect Transistors were better suited to a changing environment."

For the first example (Class-A), we use a tube as the active device, but these discussions are valid for any active 3-terminal device, tubes, bipolars, or FETs alike. Later examples use a generic device symbol, D, for any of these types. In all cases, a controlling input current or voltage to the device modulates a larger current through and external resistor, Rp, tied to a supply voltage, V+, to generate an output voltage waveform.

Continuing in figure 1, the tube's conduction is controlled by the grid-cathode voltage. A zero or positive voltage with respect to the cathode will cause the maximum current to flow through Rp. This is called Saturation and the current flow is limited by Rp. More negatively going inputs cause conduction in the Linear Range and current through the load resistor is a linear function of the input signal (low distortion). At some point of negative grid-cathode voltage, the tube's plate current is a zero and the output is at V+; this is called Cutoff (clipped.)

Class A Operation
Class-A operation means the device conducts current over 100% of the input cycle. It fact, it always conducts, even for no input signal, and so the device always draws current from the supply, V+. For this tube circuit, Vbias offsets the input voltage so that grid voltage never goes positive. Thus, the tube always operates in the Linear region, never into Saturation or Cutoff, for a bounded input level. The shaded portion of output voltage waveform shows the full cycle conduction. The further the output stays away from Saturation or Cutoff (smaller level), the more Linear (less distortion) the output signal will be.

So here's what's good about Class-A. The output is always in the Linear operating space of the device. Because of the better linearity, there will be fewer harmonics generated and thus less distortion. The penalty is that the device is always conducting, drawing current from the supply, and dissipating heat. It is only about 20-50% efficient.

Class B Operation
Class-B operation is characterized by conduction over 50% of the input cycle and the device only conducts over half an output cycle. It's not terribly useful for audioby itself, but finds use in the next configuration, Class-AB. Its early use was in CW radio transmitters. Here the grid bias voltage only offsets the input voltage halfway into the linear range, and the output is in Saturation during the positive half of the input. The shaded portion of output voltage waveform shows the half cycle conduction.

The good thing about Class-B is its efficiency, which is about 75%. Because of this, there is less heat dissipation and supply current draw.

We see then that the bias voltage determines the percent of output conduction by offsetting the input voltage. Class-A means conduction of a device for 100% of the cycle, Class-AB refers to conduction between 100% and 50%, Class-B is conduction for 50%, and Class-C refers to conduction less than 50%.

Class AB Push-Pull
In Class-AB Push-Pull, 2 devices are used, and each operates 'almost' in Class-B. The bias is set so that the upper output device conducts on the positive half of the output and the lower one during the negative half. In fact, though, each device is biased just a little bit into the other half cycle to lower "crossover distortion". This occurs when the 2 devices are biased into true class-B (50%) and then one device finishes conducting before the other has begun. This is shown in the output waveform where red is the upper device conducting and blue is the lower one.

A bias adjust control is often found in power amps to tweak the crossover point for minimum distortion. Often its inside the amp for solid state models but older tube amps brought this control out to the rear panel for user adjust (remember the pre-CBS Fenders?). Many integrated op-amps are configured as symmetric class-AB push-pull amplifiers with high open loop gain. Typically lots of feedback is used to tame the non-linearities.

Class-AB Push-Pull is good because of its higher efficiency for each device and relatively low distortion. The crossover distortion never goes away but can be minimized by proper bias. Feedback can correct some of this, but at the expense of slew induced distortions. It also turns out that symmetric push-pull configurations have distortion spectra acceptable for use in audio. Maybe that's a subject for another article ... transfer characteristics, symmetry, higher order polynomials, 2nd and 3rd order harmonics...could be cool.

Class A Push-Pull
The Class-A Push-Pull configuration is similar to the AB version and has 2 devices each operating in class-A. The bias for each device is set so that they both conduct throughout the full input cycle. As in the single ended class-A design, there is always current drawn from the supply and heat dissipation is high. This is seen in the figure above where red shows the upper device always conducting, and blue shows the lower device doing the same. The output waveform is purple to indicate both devices contributing.

Its advantage is that each device is always in its Linear region so that distortion is lower than the AB Push-Pull version. By operating both devices in the linear region, there is no crossover distortion effect. Its drawback is the large supply current required and high heat dissipation. For audio preamp designs, the extremely low distortion overshadows this.

The Class-A and Class-A Push-Pull designs are the most linear and lowest distortion configurations. My friends at UA use these configurations in their products. The tube designs use Class-A gain stages with low to moderate feedback around cascade stages. The discrete preamps are typically Class-A Push-Pull types, again with low to moderate feedback topologies. That might make a good discussion too ...local and global feedback, slewing distortion, and high open loop gain ... wow.

So till next time, keep your heaters warm and your pins in the socket.

-Professor Vacuum Tube

Do you have a question for the Doctors?