An amplifier is said to work in class A when its power devices that deliver current to the speaker never switch off, that is, when the current flowing through them never goes to zero, whatever the module and the load impedance. applied.
This statement has quite banal implications: for example, all tube amplifiers with output transformers work in class A(1). This is because if one of the output tubes were to turn off, the output transformer, which is conceptually a large inductance, would not like it – the principle of continuity of current would be lost – and would react with deleterious effects for the speaker.
Similarly, stating that the current must never go to zero regardless of the applied load means that the quiescent current – i.e. in the absence of signal – in the final devices is infinite, to manage loads with a small module or phase rotations close to 90°.
Living in a real world, designers typically decide arbitrarily on a lower limit to the magnitude of the speaker load impedance to match the amplifier they are designing, often considering only purely resistive loads, and stating that for all speakers in the range of “allowed” impedances the amplifier operates in class A.
Let's take for example the original Musical Fidelity A1 from the 1980s: it was declared to work in class A, but its operation was actually such only for 8 ohm resistive loads, and in reality it switched to class AB even before the maximum power, even on 8 ohms. With any speaker with an impedance module lower than 8 ohms and/or even non-excessive phase rotations, the A1 actually worked in class AB. Perhaps with a pair of LS3/5A with 12 ohm crossover we would have had class A operation, but certainly not with speakers from other brands with a minimum module of 4-5 ohm and phase rotations of +/- 30°.
We could list other models of amplifiers produced by various companies, some very successful and produced and sold in thousands of pieces, all declared "pure class A", but for the majority of them the description "high class AB" should be used more correctly. polarization". Editor's note | One of these, for example, was the Krell KSA-80, which the designer Dan D'Agostino conservatively declared for 80 watts in class A but in reality "switched" up to 120 watts in class AB, and hence its reputation of “hard&pure”.
The question is: why would a class A amplifier sound better than a class AB one? The reasons are various, but two perhaps more than the others:
- The transfer function of power devices – the output current versus input current for transistors or the output current versus input voltage for MOSFETs – is strongly nonlinear for small values of bias current, i.e. a class A amplifier runs its output devices at high current, where the transfer function is more linear and therefore produces less distortion.
- In class AB push-pull amplifiers each of the two power devices turns off alternately: for example, the NPN transistor turns off when the signal becomes negative and the PNP transistor turns off when the signal becomes positive. This generates the so-called "crossover distortion" which, in addition to being felt and therefore requiring more feedback, also tends to create instability in the circuit. The small polarization of class AB effectively reduces this problem, but introduces others.
To summarize and simplify, a class A amplifier can sound better with less feedback and is generally more stable on reactive loads than a class AB one of similar circuit structure.
How to distinguish an amplifier operating in high bias class AB from one operating in pure class A(2)? First of all, we should know for which load impedance values we are carrying out this test and the declared maximum power data on various resistive load modules. Then we should measure the bias current of the power devices. In consideration of the circuit topology of the amplifier, this current should be equal to or slightly greater than the maximum current that the most difficult permissible load will require.
Let's take a practical example and consider a 20 Wrms push-pull amplifier into 8 ohms capable of operating in class A up to 4 ohm resistive. This amplifier must be able to apply a maximum peak voltage of 17.9 volts to the load, both on 8 ohms and 4 ohms, corresponding to 20 Wrms on 8 ohms and 40 Wrms on 4 ohms. The maximum current flowing into a 4 ohms resistive load when this voltage is applied is 17.9volts/4ohm = 4.47amps. We round up to 4.5 amps. Since the amplifier is push pull, each of the two power devices it is equipped with – NPN and PNP transistors, or N-channel and P-channel MOSFETs – will contribute half the current. It follows that, in order to state that the amplifier operates in class A, each of its power devices must have a bias current of at least 2.25 amperes. If the amplifier reaches the rated power with a bias current lower than the minimum one just calculated, then it is in effect a high bias class AB amplifier.
And if we wanted this 20 Wrms amplifier to operate in class A up to 1 ohm load with 45° phase, what should be the bias current of its final devices? Sparing the reader, the calculations, the value is: 18 amps. This is a rather significant value. Let's not forget that this value determines a power dissipation in heat, in the absence of a musical signal, of at least 360 watts for each of the power devices, for a total of 720 watts. This is to make sure that the amplifier always works in class A with any speaker on the market. And we are talking about a single channel of a 20 Wrms push-pull amplifier into 8 ohms: in stereo the total dissipation amounts to 1440 watts. For 20 Wrms per channel into 8 ohms! Can you imagine what heat sink should be used? Not to mention the power supply that must deliver all that power continuously. This explains why all the class A amplifiers on the market are only suitable for loads of 8 or 6 ohms and reduced phase rotations.
How to understand from the nameplate data, without measuring the quiescent current of the power amplifiers, whether an amplifier is class A and for which loads? It is necessary to check the maximum absorbed power data. The efficiency of a class A push-pull amplifier is ideally 50%, in practice 40% or a little less. Taking the 40% figure as valid, we take the maximum power declared at 8 ohms on a channel and double this value if the amplifier is stereo. We divide the value obtained by the efficiency: if the maximum absorbed power is lower than the result obtained, the amplifier is in class AB with high polarization, if it is higher, then the amplifier operates in class A at least on a load of 8 resistive ohms. If the absorbed power is more than double the value we calculated, then it will work in class A up to maximum power even on 4 ohm resistive loads and more or less on all reactive loads with an 8 ohm module. Practical example: if a 45 Wrms per channel push-pull stereo amplifier into 8 ohms absorbs no more than 200 watts, then it is a class AB, because (45W+45W)/0.4=225W(3).
So far we have always considered amplifiers in push-pull configuration. Different configurations lead to slightly different calculations which however do not change the gist of the discussion.
Specifically for the M2Tech Larson, we have an amplifier in a single-ended configuration. Unlike push-pull, in the Larson only one of the output devices modulates the current applied to the load. The other simply imposes the right polarization current on the first. This circuit choice brings the Larson together with single-ended monotriode tube amplifiers in terms of sound performance and clipping behavior. Such a circuit has a maximum efficiency of 25%, in reality it rarely exceeds 20%. This means that to deliver 40 Wrms into 4 ohms it must draw at least 200 watts at rest. In reality, in the Larson the polarization device also participates in a small part in the modulation of the current on the load, so the total efficiency of the final stage is 23% and therefore "only" 185 watts of absorption are enough to guarantee class operation A on 4 ohm resistive. This measure, in addition to reducing the absorption at rest and therefore also the need for heat dissipation, contributes to slightly reducing the harmonic distortion without altering the spectrum, which sees the second harmonic predominate, followed by the third, while the higher harmonics they decay monotonically, just as with a single-ended monotriode tube. In this way the applied feedback rate can be lower than with a pure single-ended.
Being a single-ended amplifier, the Larson can only operate in class A, i.e. it cannot switch to class AB. Once it reaches its maximum current limit, it simply goes into clipping, saturating in a very "gentle" and gradual way, just like a triode. Therefore, it can be defined as a pure class A amplifier, even if, just as happens with tubes, if the load module drops below 4 ohms, instead of increasing the maximum power delivered, it will reduce it: for example, the maximum power delivered into 3 ohms it will be equal to approximately 39 watts. In fact, the rating data is rather conservative and in the laboratory up to 32 Wrms have been measured at 8 ohms and 60 Wrms at 4 ohms.
(1) In fact, in the past some valid designers, such as McIntosh, have studied particular configurations of output transformers in which some windings applied a certain feedback on the cathodes of the power tubes to prevent them from turning off completely, while still obtaining low-voltage operation. all effects in class AB, to increase the maximum power obtainable from a pair of power tubes without stressing them too much with anode currents and excessive plate dissipation. Today it is preferred to increase the number of output tubes, using more pairs in parallel.
(2) For brevity we are not considering here those so-called "non-switching" circuits and their variants which, while actually making the final stage work in class AB, ensure that both transistors always remain on, even one of the two with a small residual bias current when the other is called upon to supply the current to the load.
(3) For simplicity we are considering power amplifiers: with integrated amplifiers the efficiency value to be considered is even lower, because part of the absorbed current is used for the pre-amplification stages and for that of the services, such as relays, displays, LEDs, etc.
For further info: to M2Tech website
