This circuit is inspired by an amplifier published in the ’60s that produced 8 watts a channel into 8 Ω and was based on AD161 and AD162 germanium (not ‘geranium’) power transistors. These, at last, made it possible to build complementary-symmetry power stages with performance similar to that obtained with the standard at the time: a class AB ‘push-pull’ using with two EL84 (6BQ5) pentodes. Modest as it is, the power of the ‘Mini Sixties’ is still more than enough to drive high-quality speakers and provide comfortable listening for a signal from a computer or MP3 player. It goes without saying that for a stereo project, you’ll need to build two channels. The input signal is applied to the base of T1, which is biased via the divider formed by R1, R2, and R3, decoupled by C2. T1’s emitter receives the negative feedback signal tapped off the output by R6. As T1’s collector current is determined by the difference between the input and negative feedback signals, this transistor forms an error amplifier.
Series network R5 and R6 determine the voltage gain of the ‘Mini Sixties’ in the audio band. In the configuration shown here, the gain is 11 (1+R6/R5). Selecting a value of 22 Ω for R5 (and 470 µF for C3) enables you to increase the gain to 22 if this proves necessary. The values for R5 and C3 have been chosen to obtain a low-frequency cut-off of about 15 Hz. The amplifier’s voltage gain stage is formed by transistor T2, with resistor R12 as its load. The latter is connected to the loudspeaker output and not to the supply rail, in such a way that the voltage across it virtually doesn’t vary at all: this is the ‘bootstrap’ effect. The current through it then stays constant and is enough to drive the power transistor, even when the output voltage nears its maximum.
The disadvantage is that this current also passes through the load, resulting in a small DC voltage across the terminals of the load (26 mV @ 33 mA). Resistor R13 avoids T2 finding itself open collector when there is no load connected to the amplifier, in such a way that the quiescent voltage at the junction of R8||R9 and R10||R11 maintains its value which is half the supply voltage. Emitter resistor R7 linearizes the voltage gain stage and capacitor C4 establishes the dominant pole, which ensures the stability of the amplifier. The power stage is formed by T3 and T4 wired as a very classic complementary-symmetry ‘push-pull’ stage. Diode D1 and D2 stabilize the power stage quiescent current, which will need to be set to 20 mA by adjusting preset P2. A multi-turn type is highly recommended for P2. The quiescent current is measured using a voltmeter between the emitters of T3 and T4: the voltage measured in mV corresponds to the current in mA. If necessary, the quiescent current setting may need to be tweaked once the amplifier has reached its normal operating temperature.
The power transistor will need to be fitted to a heatsink with a thermal resistance of less than 4 °C/W, using insulating spacers and heatsink compound. It will also be necessary to make sure that D1 and D2 are in good thermal contact with T3 and T4.
The amplifier does not use a symmetrical power supply, which is why the load is connected via capacitor C7. Since the amplifier is not protected against short-circuits on the load, a 1 A slo-blow fuse lets us limit the damage in the event of a problem. The 28 V power supply is taken care of by an LM317 regulator, whose current limiter offers an additional degree of protection. The regulator will also need to be mounted on a heatsink with a thermal resistance of less than 2 °C/W. Where applicable, you may also need to provide insulation. The supply transformer TR1 must be capable of supplying 24 V @ 1–1.5 A. The fuse F2 should have the value recommended by the transformer manufacturer.
The voltages and currents shown on the circuit were measured on our prototype. We measured the distortion as 0.14 % (1 kHz, 1 W) — not all that bad for an experimental project using just four transistors.