This circuit shows a complete class AB output stage and driver stage.
The driver stage works in the same manner as the class A output stage described previously. However it is only working at a couple of milliamps, so efficiency is not an issue here. The biasing arrangement should be familiar by now.
Some sets, including the circuit above, have an extra stage of amplification between the volume control and the driver stage. Such sets will be able to receive weaker stations better and probably produce higher volume or better quality on stronger signals. This would take the total to seven transistors, whereas sets without this stage have six transistors. These are the standard line-ups, and most early MW/LW sets will have either six or seven transistors depending on the model and original selling price.
The extra amplification stage (VT1 and related components) works as described in the basic common-emitter amplifier section earlier.
In this design, negative feedback is coupled round from the output, back to the input of the driver stage via R8, to reduce distortion and improve quality.
Because the class AB output stage current consumption varies with the signal being amplified, the current drawn from the supply (battery) varies in sympathy. This will cause the supply voltage to fluctuate. We need to prevent these rapid fluctuations from finding their way back to the driver and earlier stages, otherwise they could cause distortion and/or instability.
In the circuit above we have a 100uF capacitor (C7) directly across the power supply. This would be mounted close to the output stage and would smooth the fluctuations. The supply to the driver stage biasing and earlier stages in the set is further filtered or decoupled by a 1k-ohm resistor (R13) and another 100uF capacitor (C5). If this capacitor fails the set will probably make a horrible noise known as "motor-boating" because it resembles the sound of an outboard motor. If the 100uF capacitor directly across the power supply were to fail the same effect would probably occur but only as the battery starts to run down.
Class AB Output (One Transformer)
The main problem with the two-transformer circuit above is the cost and weight of the output transformer. This is a particular problem with higher quality sets, because good quality output transformers tended to be larger and more expensive than lower quality types. If a set is built on a PCB you really don't want a big heavy lump of metal mounted on it. The driver transformer only has to deal with signals, not power, so it will be smaller and less of a quality or weight problem.
This circuit shows the basic arrangement of a single transformer class AB output stage. It may also be called a direct-coupled output stage since the speaker is directly connected to the transistors.
The two transistors are connected in series across the supply and individually biased. The transistor bases are driven by individual windings on the driver transformer. The central point in the circuit (between the two transistors) will be at half the supply voltage with no signal, and will vary above and below with the signal.
If two batteries are used in series, as shown in solid line, the central point between them will also be at half supply voltage, so the other side of the speaker (RL) may be connected to here. More often however a single battery will be used, so the speaker will be returned to one side of the supply and an electrolytic capacitor will be used for DC blocking and AC signal coupling (shown dotted). If this capacitor fails the output will become weak and/or distorted.
The main problem with this circuit is the need for a higher impedance speaker - typically between 25 and 75 ohms. These are more difficult and expensive to produce than standard 4 or 8 ohm units, which negates some of the cost saving from eliminating the output transformer.
This circuit shows a practical implementation of this arrangement. This is a Mullard reference circuit and is typical of the circuit and component values that will be encountered in many sets. The bias resistors marked * are specified at 5% tolerance, other resistors are 10%. Negative feedback is applied around the whole circuit from the output to the emitter of the driver transistor by the 2.7k resistor at the bottom.
The various bias and temperature stabilisation schemes discussed for the two-transformer circuit above could be used for this circuit too. In practice most sets rely on emitter resistors for stabilisation, while a few use thermistors.
The one-transformer circuit is a little more
difficult to design and optimise successfully, and requires better matched transistors.
It will be found that many far-eastern manufacturers continued with the two-transformer
circuit (using tiny, low quality transformers), probably due to its simplicity.
British and European manufacturers tended to use the one-transformer circuit
in their more basic models and retained the two-transformer circuit in their
higher quality models. This reinforces my personal view that the one-transformer
arrangement does not sound as good as the two-transformer circuit.
Complementary Symmetry Class AB Output (No Transformers)
While only PNP transistors were available it was not economically possible to design an output stage using no transformers. Once NPN transistors that would work at the currents involved in output stages became available, it became possible to design a simple and neat output stage with no transformers, using one NPN transistor and one PNP transistor.
This is often referred to as a complementary-symmetry output stage because the design is symmetrical and the transistors are a complimentary pair (two transistors of opposite polarity but otherwise similar characteristics working together).
RL is the speaker and is returned to the centre point of the two batteries, as with the single-transformer circuit. The two transistors are configured in common-collector mode, with the speaker as the load resistor. They are biased to the just-slightly-on point as before. On positive half cycles the NPN transistor will conduct more and drive the speaker, while on negative half-cycles the PNP transistor will work the same way.
Transistors in common-collector mode have high current gain but less than unity voltage gain. This circuit is therefore a current amplifier and driver only. The voltage swing on the output will be slightly less than the voltage on the input, so the stage must be driven by a voltage amplifier circuit, to give the full voltage swing required.
The circuit below shows a practical complementary-symmetry
amplifier. In this circuit a single battery supply is used, so the speaker is
connected via a DC blocking capacitor. For now, assume the 560 ohm resistor
on the base of the 2G381 PNP output transistor is connected to the -9V supply
instead of the speaker.
If the 2G381 driver transistor (centre of diagram) is biased such that its collector is at about -4.5V, the base of the 2G339 NPN output transistor will be at the same voltage. The 560 ohm and 47 ohm resistors provide the bias to the output transistors.The values are selected so that the 47 ohm resistor drops just enough voltage to bias both output transistors so that they are slightly conducting.
The driver transistor acts as a common-emitter amplifier as usual, with the 560 ohm plus 47 ohm resistors as its collector load. The output transistors provide the current amplification to drive the speaker but, as mentioned before, do not amplify the voltage.
The 15K and 10K resistors across the output set the DC bias for the driver transistor. By taking the bias from here, some negative feedback can be introduced to reduce distortion.
If we imagine the 560 ohm resistor is still connected to -9V, there is a problem with this circuit. On the negative peaks, when the driver transistor is nearly off, there is very little current flowing through the 560 ohm resistor into the base of the 2G381. Ohms law tells us that I = V/R, and since the voltage across the resistor at this time is very low, the current must also be low.
The output transistors, being current amplifiers
need sufficient base current into their bases to amplify on their emitters.
Also, if the voltage across the 560 ohm resistor varies then the current through
it and the 47 ohm resistor will also vary, which will mess up the biasing of
the driver transistors.
The speaker is coupled through a DC blocking capacitor. The average voltage on the left side of this capacitor will be -4.5V and the average voltage on the right side will be -9V. There is therefore a charge of 4.5V across the capacitor. As the output moves towards -9V on negative peaks, the voltage on the right side of the capacitor will follow and will move towards -13.5V (the capacitor value is sufficiently large that it does not charge or discharge significantly due to the audio). By connecting the 560 ohm resistor to this point instead of the -9V supply, the voltage at the top of the resistor will vary in sympathy with the voltage at the bottom. The voltage across the resistor is therefore constant, giving a constant current through it, allowing the 2G381 to have sufficient base drive and the biasing to remain stable.
This circuit arrangement is often called a "bootstrap circuit" because it is comparable to lifting oneself up by ones bootstraps - an impossibility! It is possible in this circuit however, and works using the charge stored on the capacitor.
In this circuit, the bootstrap circuit uses the speaker DC blocking capacitor. The drawback of this is that the speaker needs to be returned to the -9V supply line rather than the more usual 0V line. For a portable transistor set with no external speaker connections this is fine. However if the set had an external speaker or earphone output socket, it would be preferable to use the 0V line as the common. Also, if the speaker is disconnected and a higher impedance load connected (by connecting a tape recorder to the earphone socket for example) the bootstrap circuit would no longer work and the output would be distorted.
This circuit shows a complete complementary-symmetry output stage, driver and preamplifier from a Hacker set. This is about as complex as it gets in a portable transistor radio. It is perhaps not the clearest diagram on this page, but it does include several interesting points. Note that it is the opposite way up to the previous diagram, with the positive rail at the top.
In this circuit, the speaker DC blocking capacitor is C10. This only connects to the speaker and the earphone socket. The bootstrap circuit uses a separate bootstrap capacitor, C9, which works in the same way as described before. T2 is the driver transistor, which works as before.
As with the two-transformer circuit discussed earlier, there are several methods of stabilising the biasing of the output stage. Indeed, exactly the same methods may be used. Our previous example used a single 47 ohm resistor. In many sets either a thermistor or diodes will be used. Sometimes there will be a combination of the two. There may also be a small variable resistor to allow the bias current to be adjusted. In the Hacker circuit above, a transistor (T3) is used with a variable resistor (RV2) to adjust the bias current.
To set this up, the "test link" would be temporarily disconnected and replaced with an ammeter. RV2 would then be adjusted to set the required bias current, as detailed in the service data. In the absence of service data, start at 5mA and if distortion is noticeable at low volume levels increase it to perhaps 8mA. Don't set it above 10mA without guidance from the service data.
T1 is the preamplifier transistor. Note that the collector is directly connected to the base of T2 and is the only means of biasing T2 and the output stage. The whole circuit is therefore DC coupled, since changes to the biasing of T1 affect the biasing of T2, which in turn affects T4 and T5. Another variable resistor, RV1, is used to set up the biasing of the whole circuit. In practice this would be adjusted to give half the supply voltage at the output. Stabilising is achieved by connecting the output back to the emitter of T1. Audio negative feedback is also implemented through this route. This particular circuit has fairly complex tone adjustment and correction circuits in the negative feedback circuit.
With everything DC coupled, a fault anywhere will cause voltages all over the circuit to be wrong. This sort of circuit is therefore quite difficult to fault-find because it is difficult to work out whether an incorrect voltage is the fault or just a symptom. Likely suspects are the electrolytic capacitors, which can become leaky or even short-circuited. Measure the voltage across each electrolytic, and suspect any that have zero or virtually zero volts across them. Once you are happy that the electrolytics are OK, look at the transistors themselves. We will look at transistor testing and fault-finding in more detail later. If the set has variable resistors to set up the biasing, look to see of anyone has been fiddling with them. If the bias current is set too high the output transistors could be destroyed. Connecting a lower impedance speaker or short-circuiting the speaker wiring can also result in the demise of the output transistors.
Some manufacturers had reliability problems with
this sort of amplifier circuit. Notable examples are ITT and Philips. Generally
it seemed to be early Far-Eastern made sets using silicon transistors in this
circuit arrangement that had the most problems, whereas British and European
models tended to be more reliable. It is likely that very few of the problem
sets will still exist because most probably died when they were only a couple
of years old, and would have been thrown away.