32W Hybrid Audio Power Amplifier

This article cleverly combines the EL34 tube and the transistor (op amp), resulting in unexpected and high performance. This 32W power amplifier has a full power bandwidth of 5Hz to 55kHz, and its distortion is only 0.07% at 1kHz and 20W. It

is undeniable that semiconductor technology has made great progress in the past period of time, but many audiophiles still insist that tubes have better sound performance.

Although tubes require a separate filament heating power supply and a high voltage power supply, compared with solid-state devices, tubes still have many advantages that their semiconductor opponents do not have. Therefore, it is completely understandable that audiophiles still have a special liking for tube amplifiers. Why use tubes?

First of all, tubes are easy to excite. In the low frequency band, the impedance between the gate and the cathode of the tube is as high as 100Mfl, and it does not have a large parallel capacitance like VMOS. At the same time, the consistency of tubes is good, and the matching between tube samples in the same batch of products is much better than that of transistors. Therefore, using tubes to make class AB amplifier output stages may be far more linear than using equivalent solid-state devices. It can be concluded that the electron tube will never disappear in the near future, especially in the field of audio.

For a sense of exploration, we use a pair of EL34 electron tubes that have been widely used in the past, and drive them with solid-state circuits.

We think that using EL34 as the output electron tube is the best choice. The reasons are: first, the EL34 has a high plate dissipation of 25W. At the same time, it can be installed in a relatively cheap standard eight-pin locking tube socket. As for why solid-state drivers are used, you will gradually understand them in the process of reading this article. Now let's first study the basics of the output stage.

The simplest form of the electron tube output stage is a single-ended Class A three-stage tube amplifier, as shown in Figure 1a). Because the electron tube has a limited capacitive current and a considerable internal resistance, the plate drive voltage is applied directly to the speaker through an impedance matching transformer. This type of system works well, but its maximum theoretical efficiency is only 50%. Usually, due to the limitations of the plate characteristics, its actual efficiency is mostly around 25%. It seems that the single-ended three-stage output stage has become a thing of the past. However, audiophiles have brought it back to life. If you are financially comfortable and interested, you can spend a lot of money to buy a special triode tube amplifier, which costs 30,000 pounds in the UK.

Tube Output Stage

The general tube output stage is shown in Figure 1b. For simplicity, the tubes are all shown as triodes. The output is fed from the plate of the tube to the primary of the output transformer. The center tap of the primary winding is connected to the positive power supply.

When the same-phase and anti-phase input signals are applied to the grid of the tube, a push-pull effect is obtained. For solid-state devices, this type of operation depends on the bias current.

The push-pull stage usually has the advantages of offsetting even harmonics and increasing output power. In addition, the noise voltage on the plate can be eliminated and the ripple of the high-power supply can be suppressed.

Using EL34 in this type of circuit, plus a suitable high voltage, a power output of 20 to 50W can be obtained. However, the main design task of the tube output stage is the output transformer, especially to ensure a good frequency response, which depends on the careful design of the output transformer.

The difference between the actual transformer and the theoretical model. The difference is that the former needs to consider making the primary inductance large enough to obtain good low-frequency response. Similarly, at the high end of the frequency band, leakage inductance and winding capacitance limit the high-frequency response.

The circuit model of an actual transformer is shown in Figure 2, where Figure 2a is a low-frequency equivalent circuit. The components in Figure 2 are: r1 is the primary resistor, L1 is the primary leakage inductance, r2 is the secondary resistor, L2 is the equivalent secondary leakage inductance, R0 is the equivalent iron loss resistor, L0 is the primary inductance, C1 and C2 are the equivalent concentrated capacitances of the primary and secondary, Cw is the turn-to-turn capacitance, and RL is the secondary load. Here, the primary inductance and the plate impedance of the electron tube form a high-pass filter. Obviously, the larger the primary inductance, the better the low-frequency response. In Figure 2a, Rp is the plate resistance, Rw is the winding resistance, L0 is the primary inductance, and RL is the secondary load multiplied by the square of the turns ratio.

Figure 2b shows the high-frequency equivalent circuit of the transformer. In the high-frequency band, the primary inductance is large enough and has no effect on the frequency response, but the leakage inductance Lk and the winding capacitance C together form a second-order low-pass filter.

Both the leakage inductance and the winding capacitance depend on the structure of the transformer. In order to reduce the impact of these two factors on the frequency response, the transformer winding is usually wound in sections. It is obvious from the equivalent circuit that in order to obtain a good high frequency response, the leakage inductance must be minimized.

When the plate resistance is given and the required primary inductance is calculated, it can be seen that the lower the plate resistance, the more drastically the required primary inductance decreases. In fact, if the output impedance can be made zero, the required primary inductance can also be zero. Similarly, it can be shown that the distortion introduced by the transformer depends largely on the plate resistance. If zero impedance drive can be achieved, the distortion is also reduced to zero.

Therefore, the triode output stage is preferred in a device because the triode electron tube has a lower plate impedance than the pentode. Therefore, for a given low frequency response, the primary inductance required for the triode output transformer can also be lower. Most practical designs use deep negative feedback to reduce the effective plate resistance.

Usually, the feedback is taken from the output winding of the transformer, that is, the secondary winding is included in the feedback loop. However, since the output transformer has a reactive element, the amount of feedback that can be introduced in this way is usually strictly limited to avoid causing parasitic oscillations.

The best way to solve this problem is to use a cathode output stage, as shown in Figure 3. This circuit has similar functions to the emitter follower familiar to everyone in solid-state circuits. Its voltage gain is always less than 1, but its output impedance is much smaller than that of the usual cathode-grounded triode amplifier. And the distortion is usually an order of magnitude smaller.

Due to the above limitations, cathode followers are more used in laboratory applications, because driving such circuits requires almost double the signal amplitude within the range allowed by the high voltage. However, before developing the practical circuit described below, push-pull cathode followers were tried, driven by an interstage transformer. However, there is another way to produce the same effect as a cathode follower, which has all the advantages of a normal electron tube output stage and few of the side effects. This circuit is a combination of a transconductance amplifier and a transresistance amplifier, as shown in Figure 4.

It is difficult to understand why this fancy circuit has not been used more widely. This circuit achieves very good performance with a small number of components. Figure 4a shows a transimpedance amplifier that works like an ordinary virtual ground amplifier.

If the open-loop gain is high, the closed-loop performance is determined by the ratio of R1 to R2. If R1 is replaced by a constant current source, Figure 4b is obtained, and the amplifier "sees" 100% negative feedback at its inverting terminal, and its voltage gain is zero.

Replacing the constant current source with a transconductance amplifier, the output of the amplifier is IK1, and the distortion produced by the transimpedance stage is very small, because the feedback factor p (signal feedback ratio) is almost 1. Since the transconductance amplifier can also be made to unity gain, a circuit with excellent performance is obtained.

In the present circuit, the transimpedance amplifier is formed by a tube. The transistor in the feedback loop of the TL072 op amp is the basis of the transimpedance amplifier. The circuit gives an output impedance greater than 10MI over the entire audio frequency band. The required voltage gain can be obtained by varying the transconductance ratio R2, and the voltage gain of both the transimpedance amplifier and the transimpedance amplifier is 1. The well-balanced push-pull output from the driving circuit also needs to drive the push-pull output stage. For this purpose, the inverting input of the op amp can be easily driven through a resistor and a DC blocking capacitor to obtain the drive for the push-pull output stage.

Figure 4 summarizes the entire design concept. Figure 4a is a transimpedance amplifier of a virtual circuit in normal operation, and Figure 4b is a circuit in which the resistor R2 is replaced by a constant current source, with 100% negative feedback on its inverting input and zero voltage gain. Figure 4c is a circuit in which the constant current source is replaced by a transconductance amplifier, and the distortion is very small because the feedback coefficient p is approximately 1. Figure 4d converts Figure 4C into a hybrid circuit with an electron tube.

With the basis of the circuit of Figure 4, we can now discuss the complete circuit of the hybrid amplifier shown in Figure 5. The input signal is fed to the non-inverting input of A1 through R1, thereby setting the input impedance. The op amp A1 and the transistor Trl together form a transconductance amplifier as described above. The feedback is taken from the emitter resistor R3 and passed through R2 to the inverting input of A1. Resistors R12 and R13 are connected to the power supply Ve and provide bias for Trl and Tr2 to set the quiescent current of this stage.

The output current from the collector of Trl is fed into R7, and resistor R7 forms a shunt between the gate and plate of the electron tube V1. Capacitor C1 isolates the gate of the electron tube from the DC level on Trl. R6 provides a ground path for the gate. For AC, R7 and R6 form a parallel load for Trl. Due to the gain of the electron tube, this impedance is greatly reduced to about one-ninth of the original value.

The bias of the output stage is provided by R10, C3 is an AC bypass capacitor, and the screen grid is biased by R14 and R15.

The left and right halves of this circuit are identical. The phase splitting effect is achieved by coupling the inverting input terminals of A1 and A2 together by resistor R11 and DC blocking capacitor C4, which results in two signals of opposite phase and equal amplitude appearing on the emitters of Trl and Tr2 to drive the output stage.

The output voltage from V1 and V2 is applied to the primary coil of Tl, and the high voltage is applied to the electron tube through the center tap of the primary of T1. The audio output signal is taken from the secondary coil of Tl and applied to the speaker. Resistor R16 ensures that the output stage does not get out of control when the appropriate load is not applied.

Because there is a deep negative feedback in the circuit, there is no need to introduce excessive feedback through the output transformer. However, when doing experiments, the feedback can be taken from the output side of the output transformer and led to the non-inverting input of A2. If the experiment is carried out in this way, the value of R11 should be reduced to increase the open-loop gain.


High-voltage power supply

The power supply of this circuit is a commonly used general form. The high-voltage secondary of T2 is 280V, which is full-wave rectified by BRl and then smoothed by the parallel combination of C5 and C6. The combination of C5 and C6 suppresses the ripple, and a huge amount of energy is stored on it - about 68 joules. This helps to keep the power supply stable even when the load condition is very severe.
The power supply for the op amp circuit is taken from the secondary of the filament of T2. For a stereo amplifier, at least 6V and 3A are required for each channel. A 6-0-6V, 50VA transformer is sufficient.

The secondary is connected in series, and D1 and D2 double the voltage to provide 2 DC voltages, which are smoothed and filtered by C7 and C8. The filaments are connected in series and added to the 12V power supply, as shown in the circuit diagram. Because this amplifier is completely in a balanced working mode, the ripple is effectively suppressed, so the power supply design is simplified.

Production

The production of this design is not complicated, and the prototype uses a general chassis and bottom plate. For the wiring of the filament, 5A speaker cable can be used. The filament wiring should be as close to the bottom plate as possible, but it should not be twisted in pairs as in low-level electron tube circuits.

A fatal high voltage is applied to capacitors C5, C6 and all high-voltage lines, and the primary of the power transformer is supplied with AC power. As soon as the power is turned on, the amplifier composed of EL34 and the like enters the working state, and the filament should be heated before the power is turned off again. As long as the tube is connected and conducting, the decoupling capacitor will discharge quickly after power is removed. If the tube is not plugged into the socket, the high voltage will be maintained for a long time, perhaps hours or days.

This amplifier does not need adjustment. As long as the wiring is correct, this circuit will work properly from the beginning.

Conclusion

Was the effort we put in here worth it? Yes, it was. This prototype gives a continuous power of 32W per channel, and the full power bandwidth is 5Hz to 55KHz, -3dB. The distortion measured at 1KHz and 20W output is 0.07%, and the output impedance is only 0.6Ω-well below the standard value of tube amplifiers.

Not only that, this amplifier can competently drive harsh loads and can withstand the output terminals being shorted without damage.

Components list

Resistors: Unless otherwise specified, all are 1%, 2.5W metal film resistors

R1 56K
R2/5 10K
R3/4 1K8
R6/9 60K
R12/13 68K
R7/8 220K
R10 470K, 3W wirewound
R11 6K8
R14/15 470K, 1W
R16 1K, 1W

Capacitors

Cl,2 100nF, 1000V WKG polypropylene capacitor
C3 100μF, 100V
C4 220pF, 25V
C5,6 470μF, 400V
C7,8 1000μF, 63V

Active devices

A1/A2 TL072
V1/V2 EL34
Trl,2 2SC2547E
D1,2 1N4001
BRl W08

transformer

T1 Output transformer 20:1 turns ratio, center tap. Primary inductance>8H, leakage inductance<10mH.

T2 Power transformer, primary 220V, secondary 280V, 700mA, secondary 6-0-6V, 4A