For fun I've been working with some vintage tube amps that were supposed to be class B. These amps, made by Electro Voice, date from the mid 1950s. They made three of them that made 20, 30 and 50 Watts.
During this exercise I got involved in a conversation about how SETs don't have crossover distortion while push-pull amps do (which, for the record, is not correct). In that conversation was a great deal of talk about class A, AB and class B operation.
But after working with the EV amps and also ruminating on that conversation I realized that class B amps do not in fact exist. Once I sorted that out, I then realized that there's a fair bit of misinformation out there (mostly caused by people not really thinking this thru), so I thought it might be interesting to set the record straight, and in the process put some audiophile myths to bed.
To understand this bit of geekiness, and why it might matter to you, an overview of the various classes of amplifier operation is likely in order. They are ordered mostly by the timeline in which they appeared. On that account let's start with class A.
In a class A amplifier, the output tube or transistor does not turn off (the technical phrase is 'go into cutoff'). The output device (or devices, since class A can be single-ended or push-pull) conducts through the entire waveform that is being amplified. Generally this class of operation is the least efficient on that account; a fair amount of heat is generated by the output device since its conducting (and so dropping a Voltage across it, thus dissipating some power) all the time. The reason for class A is to minimize distortion in the output section of the amp.
In a class B amp the output devices, plural since this is in theory a push-pull embodiment, conduct for exactly half of the audio waveform, which is to say 180 degrees as opposed to the 360 degrees of class A. In the past class B has not been considered suitable for hifi, since there is distortion that is generated at the zero crossing point. But you'll recall I said such amps don't exist. More on that just in a bit...
Class AB is a combination of A and B, where the output devices conduct less than 360 degrees but more than 180 degrees. So in a push pull amp, this means there's a lower power region where the push and pull aspects of the amplifier are both conducting at the same time (the 'A' region). The A region prevents the amp from having any crossover distortion.
There are variants of these classes when tubes are involved. Class A can be A1 (no grid current during the entire waveform amplified), class A2 (some grid current during part of the amplified waveform) and class A3 (recently patented by Jack Elliano of Electra Print; where the tube is operated at a higher plate Voltage and grid current occurs less than 180 degrees of the waveform). Class A2 and A3 allow for more power from a given tube or tubes and can be single-ended or push-pull.
So there's class AB1 and AB2 variants having to do with grid current as well. If gird current is present, this means the driver circuit has to be a bit more robust to allow it to be linear during the grid current portion of the amplified waveform.
Finally, and where this discussion gets interesting, there is class C. The textbooks tell you that class C is only used for Radio applications. In a class C circuit, the output device conducts less than half (less than 180 degrees) of the waveform. In a radio circuit, the rest of the waveform is made up by resonance in a tuned circuit.
But class C can be push-pull too and that's where things get interesting. As mentioned, class C is when the device is conducting less than 180 degrees.
My contention is that a good number of so-called class AB amps that have crossover distortion issues are actually push-pull class C amps. This is because the output devices are cutting off before the waveform to be amplified actually gets to zero; that means less than 180 degrees, less than half of the waveform to be amplified. So class C by definition.
If the output device conducts more than 180 degrees the circuit is considered class AB. So you can see that actual class B is a very special case. My contention is due to the requirement of exactly 180 degree operation of the output devices, such amps don't actually exist. They are either very lightly biased class AB or they are push-pull class C. In the case of the EV amps I mentioned earlier, they are actually AB.
Why does this matter? One of the biggest arguments I've heard in favor of single-ended amplifiers is they inherently have no crossover distortion as opposed to other classes of operation. But this is incorrect; class D amps (which are not part of this conversation, yet, anyway...) are immune to this, but any push-pull A or AB amp is too. A class AB amp won't have crossover distortion for the simple reason that at the zero crossing point they are operating class A.
The problem is there are push-pull class C amps out there that are purported as AB when they aren't. This has caused push-pull amps to have a bad rap with the single-ended crowd, one not deserved on that account.
My contention is that an actual class B amp isn't possible, unless perhaps there is some kind of computer control that could keep the output devices operating precisely 180 degrees opposed. So far the amps I've seen that purport to be class B simply are not. One example is the so-called 'blameless amplifier' designed by Douglas Self. This is a class C design but its described as class B. But if you understand the classes of operation then you know that can't be possible.
I know this flies in the face of a lot of what we are taught in school, what we read and so on. So the question for anyone reading this is, is it wrong? And if so, why? I'm open to conversation, but simply using Appeals to Authority and the like are not going to wash if we want to get to the bottom of this. So please don't post things like 'class C push-pull doesn't exist because so and so says its class B' or some such. Show your work.
During this exercise I got involved in a conversation about how SETs don't have crossover distortion while push-pull amps do (which, for the record, is not correct). In that conversation was a great deal of talk about class A, AB and class B operation.
But after working with the EV amps and also ruminating on that conversation I realized that class B amps do not in fact exist. Once I sorted that out, I then realized that there's a fair bit of misinformation out there (mostly caused by people not really thinking this thru), so I thought it might be interesting to set the record straight, and in the process put some audiophile myths to bed.
To understand this bit of geekiness, and why it might matter to you, an overview of the various classes of amplifier operation is likely in order. They are ordered mostly by the timeline in which they appeared. On that account let's start with class A.
In a class A amplifier, the output tube or transistor does not turn off (the technical phrase is 'go into cutoff'). The output device (or devices, since class A can be single-ended or push-pull) conducts through the entire waveform that is being amplified. Generally this class of operation is the least efficient on that account; a fair amount of heat is generated by the output device since its conducting (and so dropping a Voltage across it, thus dissipating some power) all the time. The reason for class A is to minimize distortion in the output section of the amp.
In a class B amp the output devices, plural since this is in theory a push-pull embodiment, conduct for exactly half of the audio waveform, which is to say 180 degrees as opposed to the 360 degrees of class A. In the past class B has not been considered suitable for hifi, since there is distortion that is generated at the zero crossing point. But you'll recall I said such amps don't exist. More on that just in a bit...
Class AB is a combination of A and B, where the output devices conduct less than 360 degrees but more than 180 degrees. So in a push pull amp, this means there's a lower power region where the push and pull aspects of the amplifier are both conducting at the same time (the 'A' region). The A region prevents the amp from having any crossover distortion.
There are variants of these classes when tubes are involved. Class A can be A1 (no grid current during the entire waveform amplified), class A2 (some grid current during part of the amplified waveform) and class A3 (recently patented by Jack Elliano of Electra Print; where the tube is operated at a higher plate Voltage and grid current occurs less than 180 degrees of the waveform). Class A2 and A3 allow for more power from a given tube or tubes and can be single-ended or push-pull.
So there's class AB1 and AB2 variants having to do with grid current as well. If gird current is present, this means the driver circuit has to be a bit more robust to allow it to be linear during the grid current portion of the amplified waveform.
Finally, and where this discussion gets interesting, there is class C. The textbooks tell you that class C is only used for Radio applications. In a class C circuit, the output device conducts less than half (less than 180 degrees) of the waveform. In a radio circuit, the rest of the waveform is made up by resonance in a tuned circuit.
But class C can be push-pull too and that's where things get interesting. As mentioned, class C is when the device is conducting less than 180 degrees.
My contention is that a good number of so-called class AB amps that have crossover distortion issues are actually push-pull class C amps. This is because the output devices are cutting off before the waveform to be amplified actually gets to zero; that means less than 180 degrees, less than half of the waveform to be amplified. So class C by definition.
If the output device conducts more than 180 degrees the circuit is considered class AB. So you can see that actual class B is a very special case. My contention is due to the requirement of exactly 180 degree operation of the output devices, such amps don't actually exist. They are either very lightly biased class AB or they are push-pull class C. In the case of the EV amps I mentioned earlier, they are actually AB.
Why does this matter? One of the biggest arguments I've heard in favor of single-ended amplifiers is they inherently have no crossover distortion as opposed to other classes of operation. But this is incorrect; class D amps (which are not part of this conversation, yet, anyway...) are immune to this, but any push-pull A or AB amp is too. A class AB amp won't have crossover distortion for the simple reason that at the zero crossing point they are operating class A.
The problem is there are push-pull class C amps out there that are purported as AB when they aren't. This has caused push-pull amps to have a bad rap with the single-ended crowd, one not deserved on that account.
My contention is that an actual class B amp isn't possible, unless perhaps there is some kind of computer control that could keep the output devices operating precisely 180 degrees opposed. So far the amps I've seen that purport to be class B simply are not. One example is the so-called 'blameless amplifier' designed by Douglas Self. This is a class C design but its described as class B. But if you understand the classes of operation then you know that can't be possible.
I know this flies in the face of a lot of what we are taught in school, what we read and so on. So the question for anyone reading this is, is it wrong? And if so, why? I'm open to conversation, but simply using Appeals to Authority and the like are not going to wash if we want to get to the bottom of this. So please don't post things like 'class C push-pull doesn't exist because so and so says its class B' or some such. Show your work.
