|Elliott Sound Products||Valves - Classes|
The different classes of valve amplifiers often confuse people, and not only beginners. They are all well documented, but there is still some confusion regarding "sub-classes", and almost nothing that discusses the changes that occur with a valve amp in overdrive.
Some of the information I've seen has to be considered doubtful, and advertising brochures and websites are big on hyperbole but rather low on technical detail or accuracy. The classes of operation are well established, and while some are applicable to both valve (tube) and transistor amps, others are not.
Figure 1A - Push-Pull Amplifier Schematic Used For Waveforms Shown Below
The push-pull schematic shown is useable for all classes described. This push-pull amp can be biased for Class-A, AB or B, but remember that any amp that draws grid current must always be driven from a very low impedance driver, preferably using a transformer to keep impedances as low as possible. While grid current (during linear operation) is discussed, it is not usually suitable for general purpose audio amplifiers because of the requirement for a very low impedance drive circuit. Basic voltage waveforms are shown in red and green, so you can see which valve is responsible for each half of the waveform.
Please don't take this literally though ... especially in Class-A, both valves are equally responsible for the whole waveform. As one turns on, the other turns off. This also applies for the other classes until the valve that is turning off ceases conduction altogether. Although triodes are shown, tetrodes or pentodes can be used in all classes of amplifier. The 10 ohm resistors in the cathode circuit are to facilitate setting the bias current. These should be 5W wirewound types, and should be included in all push-pull amplifiers. Where cathode bias is used with a common resistor they are still recommended, but can be downgraded to 1W.
Figure 1B - Cathode Biased Push-Pull Amplifier Schematic Used For Class-A & Class-AB
The generalised circuit for a cathode biased stage is shown in Figure 1B. This is a very common arrangement, and is a simple way to build Class-A and Class-AB amps. Because of the huge cathode current variations in a Class-B stage, this arrangement cannot be used. The bias is achieved by way of the voltage across Rk, and it is impossible to have a voltage equal to the valve's cutoff voltage developed across a resistor (of sensible resistance) if the valve is barely conducting.
The voltage across Rk is sacrificed from the main supply, so if B+ is 300V and the voltage across VBIAS is 20V, then the valve has 280V that's usable so some power is lost. The benefit of this arrangement is that is will adjust itself as needed as the valves age, and no adjustment is needed.
Figure 1C - Single-Ended Amplifier Schematic
The single-ended version in Figure 1C can only be used in Class-A. None of the other classes of operation are suitable when a single valve is used. This circuit has also been shown with cathode bias, because it is the most common. While fixed bias may be used for a single-ended stage, it's not common to see it done that way. The voltage and current waveforms are the same as those shown in Figure 2, but only for a single valve.
Class-A is defined as an operating condition where the amplifying device conducts for the full 360° range of the input signal - namely from zero, through to the maximum positive signal peak, through zero again to the full negative signal peak, and back to zero. This is a full cycle of the audio waveform. At no point during normal unclipped operation does plate current fall below the minimum value needed to maintain acceptably linear operation.
Virtually all valve preamp stages operate in Class-A - they use one or more single valve stages, each with a plate load consisting of either a resistor or another valve. When another valve is used, it's generally used as a current source, or as part of the so-called SRPP stage. Either way, Class-A dictates that the valve carries current for a complete cycle of any applied waveform - at no stage is the current reduced to zero during linear operation. (Class-AB operation is possible for preamps, but only for transformer coupled stages - this is very rare for audio applications.)
Single valve circuits can only function in Class-A, because they cannot act as an amplifier without the plate current through the plate load resistor. For preamps, it's generally advisable to ensure that the plate current at idle is at least 4 to 5 times the current that will be drawn by the next stage. This can be another voltage amplifier, a tone-stack, a transformer (either to drive an output stage or perhaps a reverb tank) or an external amplifier. Regardless of whether the stage is a voltage amplifier or a cathode follower, the basic rule is still followed, as this keeps distortion within reasonable limits. This is not a hard and fast rule though - it's quite common to see circuits where the external load is the same or less than the plate load resistance. This causes a loss of gain, and also reduces the maximum output swing.
For power output stages, Class-A is also used, and it can also be single-ended (using a triode, tetrode or pentode), or perhaps two of the same type in parallel. As can be expected, any single-ended output stage has advantages and disadvantages. The primary advantage is simplicity - or to be more precise - apparent simplicity. A single valve is needed, and can often be used at low power without needing any additional amplification. If a triode is used, a prior gain stage is generally necessary because of the power valve's relatively low gain.
The main disadvantage is that the output transformer's windings have DC flowing at all times, so there is a net static flux in the transformer that dramatically reduces transformer performance. This is generally catered for by using a much larger transformer than might otherwise be needed, and providing an air-gap in the core to reduce the effects of saturation. Rather high distortion (especially intermodulation distortion) and poor damping factor are additional disincentives.
Push-pull amplifiers can also operate in Class-A, and the general conditions are the same as for a single valve - the plate current never falls to zero, and preferably never falls below the minimum current that is needed to ensure linear operation. Push-pull amplifiers have the advantage of cancelling the DC current in the transformer windings, so there is no static field in the core. The transformer can therefore be smaller for the same or greater output power.
It is sometimes mistakenly believed that all Class-A amplifiers (single-ended and push-pull) use cathode bias - a resistor in the cathode circuit that raises the cathode potential above zero, and provides the required grid bias. This resistor is usually bypassed with a capacitor, but there are some circuits that are designed to operate either with separate cathode resistors and/or without the capacitor. Cathode bias has the advantage of not requiring adjustment in service, but well matched valves are essential for good performance.
While cathode bias is indeed common with Class-A output stages, fixed bias from a separate negative voltage source can also be used. This does not mean that the circuit is not Class-A, contrary to what you may read elsewhere. The only requirement for Class-A is that the valve (or valves) conduct for the complete signal waveform, and are never allowed to be reduced to zero. Fixed bias is just as valid as cathode bias for a Class-A stage, whether single-ended or push-pull.
Although cathode bias is the most common for Class-A amps and in theory the resistor requires no bypass capacitor, it is almost always necessary to include it unless the output valves are perfectly matched in every way. Since this is virtually impossible to ensure for the life of the valves, the bypass capacitor ensures that both valves get equal bias current and drive signal. It is rare to see it omitted.
When operating an output stage in Class-A, it is common that the plate current is such that maximum rated plate dissipation is the normal condition at idle. While the peak dissipation may exceed the rated maximum when operating, the average value of plate dissipation either remains the same, or (and more commonly) is reduced slightly. Operation above the rated plate dissipation is not recommended under any circumstances. The load impedance should be chosen such that the valves still conduct (perhaps 5-10% of the quiescent current) at full power, so the inherent non-linearity at very low current is minimised.
While Class-A2; operation is possible (meaning that control grid current is drawn to get full undistorted) power, it is very uncommon. Class-A amps are favoured for their fidelity, and it is extremely difficult to maintain fidelity while driving grid current. This would also complicate the drive circuit. In essence, this mode of operation would qualify as a bad idea.
Figure 2 - Class-A Voltage and Current Waveforms
Figure 2 shows the basic scheme of single-ended and push-pull Class-A amplifiers. Both use fixed bias, and you can see that the plate current never falls to zero. The bias current must be determined for the lowest likely load impedance. If the bias current is too low, a single-ended stage will clip, and a push-pull circuit will be converted to Class-AB.
Note that Class-A is not restricted to triodes. A vast number of consumer level products were built that used small pentodes in single-ended Class-A output stages - mantel radios, "radiograms" (combined radio and record player), stand-alone tape recorders and many other products used this scheme because it was comparatively cheap, and worked well enough for the intended purpose.
Class-AB is defined as a condition where each output device conducts for less than 360° of the input cycle, but greater than 180°. In other words, the output devices conduct for somewhat more than a half-cycle of the input waveform.
Class-AB is the most common for push-pull amplifiers, and cannot be used with single-ended stages. The advantage over Class-A is that the stage is more efficient, and can deliver far more power from a pair of valves of the same type as might be used for Class-A. There are a few sub-classes with Class-AB - these are ...
Most Class-AB amplifiers are Class-AB1, and do not draw grid current to achieve full power, allowing simple drive circuits which may be capacitor coupled. Class-AB2 is far less common, and is generally reserved for extremely high power amplifiers as might be used as AM transmitter modulators. This type of operation requires a very low impedance driver stage to avoid excessive distortion, and it will typically use a driver transformer.
While the above are the "official" designations, almost without exception, any Class-AB amplifier that is driven into distortion will become Class-AB2. Typically, full power is achieved at a point where the control grid voltage on the power valves reaches (or is close to) zero volts. Once the plate voltage reaches saturation (its minimum possible voltage without grid current), the driver stage will continue to produce higher than needed grid voltage, and grid current cannot be avoided. Once the drive signal exceeds zero volts (becomes more positive), grid current flows and cannot be avoided. Because the drive signal is capacitor coupled (and is not low impedance), this causes severe distortion of the drive signal. However, this is (more or less) immaterial because the output is distorted anyway.
There is also an unexpected result - the coupling capacitor is charged by the grid current, and this increases the effective negative bias voltage for a short time until the excess charge is dissipated through the circuit resistances. If you'd like to know more about this, the topic is covered in much greater detail in the Analysis article.
Figure 3 - Class-AB Voltage and Current Waveforms
Figure 3 shows the voltage and current waveforms in a Class-AB1 amplifier. While there can be a vast difference between circuits, the general principle is unhanged. Some Class-AB amps will provide up to perhaps half the total voltage swing (a quarter of the maximum power) in Class-A, while others may make the change to Class-B at much lower power levels.
Class-AB amplifiers may use cathode bias or fixed bias. With cathode bias, the voltage across the bias resistor(s) must remain stable throughout the cycle, so the bias point is generally set by choosing a cathode resistance that ensures that the total current is roughly equal - the average AC current component should be close to the same as the quiescent DC current to prevent bias shift during operation. A (generally quite large) capacitor is almost always necessary to maintain a constant bias voltage. Some valve manufacturers insist that separate cathode resistors be used, but most cathode biased circuits use a common resistor bypassed with a capacitor.
The definition of 'true' Class-B is that each output device conducts for exactly 180° of the input cycle. There is (in theory) no period when both valves are conducting. This means that both valves are biased to cutoff (zero plate current) with no signal. In reality, there is no way that this can be achieved with valves without gross distortion at the zero signal point - referred to as crossover distortion. This is where the signal crosses over from being positive or negative (at the speaker terminals). The low-current linearity of output valves is generally very poor, so even Class-B amplifiers will typically operate with a small quiescent (bias) current.
Class-B is actually very uncommon for audio valve amplifiers because of the issues with crossover (aka 'notch' distortion) that is inevitable with valves biased into (or close to) cutoff. All audio Class-B amplifiers are actually Class-AB, but (as noted above) biased for a very low quiescent current. Crossover distortion can be minimised by the application of feedback around the output stage, but this can (and does) cause the harmonics caused by crossover distortion to be shifted to higher frequencies. The result is just as objectionable as a transistor amp with crossover distortion.
Almost all amps that are referred to as being Class-B are actually Class-AB, but with the lowest possible bias current that ensures at least acceptable low-level distortion figures. Some amplifiers intended solely for general purpose public address use (often through efficient but poor fidelity horn loudspeakers) were very close to being Class-B, because the primary goal was high volume reproduction of the speech band only (typically 300Hz to 3kHz). These amplifiers didn't need fidelity - they needed to be efficient and loud. Typical applications would include PA systems for horse racing events, stadium announcements, etc. The sub-classes are the same as for Class-AB ...
For high power PA usage, Class-B2 was reasonably common, because you can get the maximum possible power from a pair (or several pairs) of output valves this way. For this class of service, fidelity simply was not an issue, but high efficiency and power output were both high priorities.
Figure 4 - Class-B Voltage and Current Waveforms
The voltage and current waveforms are shown above. The crossover distortion is deliberately exaggerated so you can see the result. Normally (and unlike a transistor amp with crossover distortion), the kink around the zero crossing point is not especially pronounced, but it is still audible. Distortion at very low levels can be intolerably high.
True Class-B is/was uncommon (and this still applies with transistor amps). Transistor power amps can get a lot closer to real Class-B than valves, but the same general comments exist. Crossover distortion is usually higher than desirable, although it should not be audible when used for the intended purpose - at relatively high volume and through speakers with a limited bandwidth. It's worth noting that both crossover and clipping distortion contain the same harmonic structures and in the same ratios. The difference is phase, and the fact that crossover distortion is worst at very low levels, while clipping distortion only occurs at very high levels.
Class-B (and Class-AB amplifiers that are biased to a low plate current) must use fixed bias. Because the current varies so much between zero or low power and maximum power, a cathode resistor would cause very wide variations in the bias voltage, which in turn will change the bias voltage along with the signal. A capacitor doesn't help, because it will charge due to the much higher average current, causing the valves to be completely cut off at low signal levels until it discharges. This leads to gross distortion. Since the reasons for using low-bias Class-AB and Class-B are to obtain higher than average efficiency, it would be rather silly to promptly reduce efficiency by using cathode resistors ... even if they did work.
Although not relevant to audio, it's worth pointing out that a common application with RF amplifiers is Class-C. This is defined as a condition where output device current flows for (considerably) less than 180° of the signal waveform. This is an extension of Class-B, but the output device only applies a brief pulse of current to a tuned circuit load. This class of amplifier is common with FM (Frequency Modulation) or CW (Continuous or Carrier Wave - usually modulated) power stages, but cannot be used for AM final amplifiers that must handle the modulated signal.
Because it is so far outside audio applications it will not be discussed in any more detail.
It's worth looking at what happens when a valve draws grid current due to overdrive. Once the grid is more positive than the cathode, it becomes another anode, so current will flow between the positive grid and (now) negative cathode. Figure 5 shows the diode that is formed within the valve itself, and what happens to the grid drive waveform if it is from a typical high impedance source. To prevent distortion of the drive current, the source (driver stage) must have extremely low impedance and resistance.
Figure 5 - Grid Drive Waveform Distortion Caused By Grid Current
Although this shows the effect with a single-ended stage, exactly the same thing happens with push-pull, and it happens to both drive signals. Note that this applies to an overdrive condition, and it happens with triodes, tetrodes and pentodes, regardless of class of operation. Where an amplifier is specifically intended to draw grid current, the drive section must be completely redesigned. The output stage will generally use a lower impedance transformer than would normally be the case, and more power can be obtained from a lower supply voltage.
In the diagram above, the normal saturation voltage of the valve can be lowered as shown by the grey lower tip of the waveform, but only if the drive circuit can supply an undistorted grid signal despite the relatively large grid current. This is why specialised low impedance drive circuits are needed for Class-AB2 operation. The drive circuit may need to supply several milliamps in a high powered amplifier, with no waveform distortion.
Hopefully, the above will remove any confusion. The definitions provided are 'text book' - these have been established for a long time, and there is no reason or excuse for anyone to think that they might have other meanings. It is important to understand that the type of bias used has no bearing on the class of operation, with the single exception that Class-B and low-bias Class-AB cannot be achieved using cathode bias.
An amp that has an adjustable fixed negative bias supply can (valve dissipation and bias adjustment range permitting) be biased anywhere from Class-B right through to Class-A. However, few amplifiers that are designed to be Class-AB will be able to be operated in Class-A unless the power supply voltage is reduced to prevent grossly excessive plate dissipation. It is also common that the required output transformer ratio will be different from that fitted.
It is important to understand that all definitions are based on maximum undistorted output power. All amp classes will draw grid current when driven into distortion, and once an amp is used consistently this way (such as guitar amplifiers), the operating class is more or less meaningless. A clipping Class-A amplifier is unlikely to sound any different from a clipping Class-AB amp - while the player may well imagine a difference, it's probable that there is no audible difference at all. Of course, it's possible that some players can hear a real difference, but this is likely to be more to do with playing style and subtle nuances that most will miss.
For hi-fi applications, there is no doubt that (push-pull) Class-A will give superior sound quality in almost all cases. I suspect that an ultra-linear Class-A output stage could provide extremely good performance - certainly better than you'll get with triodes. This is not a common topology, but there is no reason at all that it could not out-perform almost any other in terms of low distortion and output impedance. Considering the number of Class-AB ultra-linear stages that already give outstanding results, Class-A would give it that little bit extra.
|Copyright Notice. This article, including but not limited to all text and diagrams, is the intellectual property of Rod Elliott, and is Copyright © 2009. Reproduction or re-publication by any means whatsoever, whether electronic, mechanical or electro- mechanical, is strictly prohibited under International Copyright laws. The author (Rod Elliott) grants the reader the right to use this information for personal use only, and further allows that one (1) copy may be made for reference. Commercial use is prohibited without express written authorisation from Rod Elliott.|