Adding an output stage to the amplifier circuit. Output stage. Bias circuit design

Transistors of the output stage of power amplifiers audio frequency(UMZCH) (in most cases this is a composite emitter follower, as in Fig. 1) heat up during operation, the emitter-base voltage of the transistors and the current of the cascade operating point change. The return to the optimal operating point, at which minimal distortion is generated, is carried out by a feedback circuit by changing the bias voltage U bias depending on the state of the temperature sensors installed on the radiator. The bias voltage must accurately track the temperature of the pn junction of two or more output transistors. Often this does not happen accurately enough, and even with a large delay, since the time constant of the circuit: pn junction - transistor body - radiator - temperature sensor can reach several tens of seconds! Thus, when amplifying a real signal, most of the time the output stage “searches” for the optimal operating point, which means it works with under- or over-bias and with increased transient distortion! In amateur designs, incorrect installation of the temperature sensor is common mistake and can even lead to overheating powerful transistors and their thermal breakdown.

In Douglas Self's famous book “Designing a UMZCH”, as many as 60 pages are devoted to the problem of optimal thermal compensation and choosing the location for installing thermal sensors, after reading which it becomes clear that the problem can only be reduced, but not solved.

Thermal sensors can be abandoned if the current is strictly stabilized at the operating point by introducing deep negative current feedback (NFE) into the output stage. Such Feedback, in addition to stabilizing the quiescent current, also allows you to implement the super-A (Non Switching) mode with non-switchable output transistors (and with minimal switching distortions). OOS also improves the linearity of the output stage and reduces the dependence of this linearity not only on the parameters of powerful output transistors (far from ideal ones), but even from the type of transistor used (field-effect or bipolar).

Let's consider the passage of a signal in a standard UMZCH (see Fig. 1). After the voltage amplifier, the signal enters the output emitter follower, made on complementary transistors and is actually divided into positive and negative half-waves, and each half-wave is amplified in current separately (and, unfortunately, unstable) by the output transistors. Now it becomes clear that to correct the situation, you need to solve the following problem: “thermostably” split the signal into two half-waves, then “thermostably” amplify them in the corresponding channels (adding a quiescent current), and then sum them at the output!

So, the scheme for solving the problem is shown in Fig. 2. The input signal is divided into positive and negative half-waves using a splitter on diodes VD1 and VD2, then the desired bias (quiescent) current I bias is added to each half-wave. Next, the sum of the currents Isign and Ibias is amplified by unipolar and thermally stable power amplifiers with deep current feedback (amplifiers X and Y). The output currents of the amplifiers are applied to the load, and the signal currents are summed, and the bias (quiescent) currents are subtracted, and the output signal is identical to the input.

It’s interesting that the idea of ​​separate amplification of signal half-waves was noticed by me, a young engineer, more than forty years ago (!) in a super article by Peter Blomley in Wireless World magazine, February-March, 1971, “A New Approach to Class B Amplifier Circuit Design.” (By the way - Fig. 1 is an exact copy of the drawing from this article!) Then, for many years, in articles and books (even in the book by Douglas Self) there were references to this idea like “there is an idea, but has not yet found commercial application ". It is also interesting that 19 years later, in the magazine Radio No. 12 for 1990, pp. 62-64, an article by Mr. G. Bragin appeared, where he intuitively came close to solving the problem of creating an UMZCH without thermal sensors, but the problem of stability of the input quiescent current remained unresolved and his scheme was forgotten... After 31 years - in 2002, even a patent by comrade appeared (for some reason). Zhbanova V.I. on this topic SU2189108 “Highly linear push-pull amplifier and a device for separating a signal into two half-waves,” but the friend did not fully understand the meaning of the idea and did not offer real circuits...

So, let’s consider a possible option for constructing a thermally stable half-wave amplifier with deep current feedback (for example, Amp X) in Fig. 3. This is actually a textbook-famous ITUN (voltage-controlled current amplifier) ​​circuit. The voltages at points A and B (relative to point C) are equal to each other and the more accurate the greater the gain of amplifier op-amp1, therefore, according to the theory of feedback amplifiers, all instabilities within the points covered by this connection (and these are the transmission coefficient and shifts voltages at the p-n junctions of transistors with their temperature instability), have little effect on the accuracy of the output current matching the input voltage! Thus, if we apply a positive half-wave voltage with a bias to the input of such an amplifier, at the output we will obtain an output current and a bias (quiescent) current independent of the temperature and parameters of the amplifying element - which is thermal stabilization of the operating point.

Let's consider possible options for creating a UMZCH using this basic ITUN amplifier - fig. 4,5,6.

As you can see, the outputs of the positive and negative half-wave amplifiers can be connected in parallel - as shown in Fig. 4, counter-connected - as in Fig. 5, or in series (for identical transistors) - as in Fig. 6. The output currents are summed across the load and reproduce the input signal. From the figures it becomes clear how to apply the input voltages Vsign+ and Vsign for half-wave amplifiers to their inputs. They need to be supplied from current generators Ibias and Isign and “tied” using resistors R1 and R3 in the circuit of Fig. 4: - to the load Rн, in the circuit of Fig. 5 - to power sources, and in the circuit of Fig. 6 - both there and here. For Fig. 6, the necessary inversion is carried out using a current mirror at T1 and T2. Note that in the circuits of Figs. 4 and 6, additional negative feedback occurs when the input current flows through the load resistance Rн.

Let's move on to the input voltage half-wave shapers - splitters. In the circuit shown in Fig. 4, the simplest versions of the splitter are applicable - on diodes or on emitter followers - they are shown in Figs. 7 and 8.

When using an input voltage amplifier with a current output according to the diagram in Fig. 7, for the “correct” operation of the diodes, a blocking voltage of at least 250 mV is required. If this is not done, the currents through the diodes VD1-VD3 and VD2-VD4 will be equal to half the direct current of the output transistors of the voltage amplifier VT1 and VT2, but we do not need this. The blocking voltage is obtained using stable voltage dividers Vbias - R4-R5 (R7-R6). So that this offset does not “interfere” with the operation of op-amp1 (op-amp2), it is necessary to subtract exactly the same voltage using a divider R8-R10 (R9-R11) at its negative input. Next, we note that when a signal is applied to such a splitter on a negative wave, the diode VD2 closes and in order to maintain the minimum current of the idle channel Imin (see current diagrams in Fig. 9), we apply the necessary bias to the positive input of op-amp1 through resistor R2. In the other channel, the minimum current is provided by resistor R3. To obtain the stable and adjustable bias needed to select the operating point of the splitter, we pass the stable output current of the voltage amplifier through trimming resistor R1. This bias, together with the minimum currents Imin, sets the adjustable quiescent current Iok for the output transistors (simultaneously for both arms). In order not to select the divider resistors R8-R10 (R9-R11) with great precision and to take into account the non-zero shift voltage at the input of amplifiers OU1 and OU2, it is recommended to make resistors R1, R2 and R3 trimmers and use them to set the currents Imin and Iok.

The transistor splitter shown in Fig. 8 works similarly, only its input resistance, as an emitter follower, is much higher, so it does not load the output stages of the voltage amplifier and provides them with greater gain.

It should be said that in the simplest emitter follower, as in Fig. 8, at a large signal amplitude, the right transistors VT5 and VT6 heat up much more than the left ones VT3 and VT4, since they operate at a high supply voltage at the collector. Therefore, there will be no thermally stable transfer of bias voltage from resistor R1 to resistors R4 and R5, and here it is better to use a repeater based on a Wilson current mirror, which does not have this effect.

Let's consider a real amplifier circuit (according to the structure of Fig. 7) - Fig. 10.

The voltage amplifier is built according to the classical scheme: a differential stage (VT1, VT2), a cascade with OB (VT6), loaded on a current mirror (VT4, VT5), then cascades OK (VT9) and OB (VT8). It provides high gain and stable 4 mA output stage current. Then, through a splitter on diodes VD4-VD7, half-waves of the signal are supplied to the differential stages on VT10-VT13 (VT14-VT17) and then to the emitter followers on VT18-VT20-VT22-VT23 (VT19-VT21-VT24-VT25). Power for the differential stages is taken from the voltage boost circuit on R32-C6-VD8 - plus 9V and R33-C7-VD9 - minus 9V. Feedback to the bases VT13 (VT15) is supplied from the emitter resistors of powerful output transistors VT22-VT25, so the circuit monitors the equality of the voltages on them (taking into account the bias for blocking the diodes) and on input resistors R17/R18 and R19/R20 (according to the diagram in Fig. 7). For amplifiers with a power of more than 80 W, it is recommended to use at least two powerful transistors in each output arm, so in this circuit the feedback monitors a certain average current of transistors VT22, VT23 (VT24, VT25) using summing resistors R42 and R38 (R43 and R39).

The circuit is insensitive to supply voltage ripples. It works perfectly from unstabilized voltage sources from +/- 20 V to +/- 60 V. The output voltage amplitude is only 3.5 V less than the supply voltage. For example, with a +/- 45 V power supply, the output power is 80/140 watts at load 8/4 Ohm. The slew rate of the output voltage is 70 V/µS, the full signal bandwidth is 300 kHz, the output impedance is about 0.005 Ohm at frequencies up to 50 kHz, non-linear distortion even at 20 kHz is less than 0.003% at full power! The output transistors can easily withstand output voltage full swing frequency 200 kHz! The circuit operates in super-A class (current diagrams are shown in Fig. 9) with a minimum current and quiescent current of 30 mA and 120 mA, respectively (for each output transistor). These currents are maintained with an accuracy of better than 10% for any temperature of the output transistors. The currents should be adjusted without a signal and without a load in the following sequence: first, we short-circuit the resistor R14, which determines the quiescent current, and with resistors R22 and R23 we set a voltage of 10 mV on the emitter resistors R42 and R43 (0.33 Ohm) in both arms - these are the minimum currents 30 mA. Then, using resistor R14, we set the voltage to 40 mV on the same resistors. This corresponds to a quiescent current of 120 mA. Then, with a load and a signal, on the same resistors we check the correct shape of the currents, corresponding to the diagrams in Fig. 9. At short circuit load, the output current is limited by LEDs HL3 and HL4 with a voltage of 1.7 V at 8 amperes.

By the way, this circuit has an interesting “bonus” - a circuit consisting of a 220 μF capacitor and a resistor Roc, ranging from 390 Ohms to 4.7 kOhms, installed between the emitters VT5 and VT9 (the connection is shown in the diagram with a dashed line) transforms the output resistance of the amplifier into the negative!!! At Roс = 390 Ohms, the output resistance is minus 0.35 Ohms, at 620 Ohms - minus 0.22 Ohms, at 1.1 kOhms - minus 0.11 Ohms, and this resistance is constant throughout the entire audio band while maintaining the other parameters of the amplifier! Good opportunity Acoustics lovers can experiment with expensive acoustic wires or resistance compensators for “bad” wires.

It is also interesting that the circuit differs from the standard “classical” circuit with thermal compensation only in a small middle part - from diodes VD4-VD7 to transistors VT18-VT19, which means it is possible to easily modify many ready-made amplifiers by discarding thermal sensors and increasing reliability and sound quality. Fig. 11 shows another version of the modification amplifier circuit, which is simpler and more economical, but provides the same amplifier parameters as the previous version.

The circuit in Fig. 10 also works with field-effect output transistors, only to recharge the large gate capacitance of powerful transistors such as IRFP240 - IRFP9240, a more powerful emitter follower is required than VT18, VT19. The modified scheme is presented in Fig. 12 and is called universal, since with field effect transistors has exactly the same parameters as with bipolar ones, only the rate of rise of the output voltage is slightly lower (50 V/µS), which, however, is quite sufficient for UMZCH “with high fidelity”.

The circuit of a fully symmetrical and capable of operating as an operational amplifier with supply voltages +/- (6 - 60) V and output currents of 10 amperes is shown in Fig. 13.

The use of a splitter in this circuit on emitter followers with a high input resistance, as shown in Fig. 8, made it possible to use at the input the simplest differential amplifiers on complementary transistors with high gain (h21 more than 500) (VT1-VT5 and VT2-VT6) and circuits with a common base on VT7 and VT8. The input currents with such inclusion of differential cascades are determined by the difference in the base currents of complementary transistors and at close values ​​of h21 amount to hundreds of nanoamps, which makes it possible to do without capacitors in the OOS circuit or without input capacitance, and use the circuit as a powerful DC amplifier!

Here, after a splitter on transistors VT9...VT14, both half-waves of the signal are fed to thermally stable amplifiers on VT15 - VT22. For such amplifiers, it is important to maintain equality of currents through transistors VT16, VT20 and VT17, VT21, since they compare the input voltages of the amplifier (at resistors R19 and R20) and the output voltages at the emitter resistors of powerful output transistors. Perfect option Of course, here we can use pairs of matched transistors with close h21 and Vbe, such as KR159NT1 and KTS3103, however, the long-term use of BC546C/BC556C transistors in such circuits has shown their full suitability (it is advisable to take transistors from the same batch and place them side by side on the board or glue them together). Practice has shown that such a circuit maintains a minimum voltage, for example, 10 mV (Imin = 30 mA) and a quiescent voltage of 40 mV (Iquiescent = 120 mA) with an accuracy of 10% at any temperature of powerful transistors! Stable current generators on VT15,VT19 and VT18,VT22, by adjusting resistors R23 and R29, allow you to set the minimum current for high-power transistors. Recommended currents for “Super A” mode are 30/120 mA. The output stage described above has a slew rate of 120 V/µS.

The entire amplifier is capable of operating from 8 Hz to 450 kHz at full power. The slew rate of the output voltage is 80 V/µS. Meander perfect shape 60 V swing up to 200 kHz! Output impedance less than 10 mOhm throughout the entire audio range! Harmonic distortion less than 0.01% even at 20 kHz and full power! There is no overshoot when limiting the signal! The amplitude of the output voltage is 3.5/4 V less than the supply voltage for an 8/4 Ohm load, therefore, with a supply voltage, for example, +/- 45/40 V (without and with load), the output power is the same as the circuits in Fig. 10 , also 80/140 watts for 8/4 ohm load.

Fans of using operational amplifiers may be interested in the circuit in Fig. 14, which is also built according to the structure of Fig. 7.

The quiescent current and minimum currents of the output transistors are set using trimming resistors R13 and R18, R21, respectively. The scheme is the same for field and bipolar transistors! (connection of bipolar transistors is shown in Fig. 15). All parameters are limited by the properties of the operational amplifiers used. For OPA2134: supply voltage +/- (20-50) V, bandwidth 8Hz-200 kHz, slew rate - 40 V/µS with input filter 1kOhm/300pF. The output voltage amplitude is 2.5/3.5 V less than the supply voltage - for an 8/4 Ohm load and for any transistors! Zero at the output is maintained perfectly. Output impedance less than 5 mOhm! The circuit can also be used as an operational amplifier (and as an inverting amplifier too). Unfortunately, nonlinear distortion increases sharply with increasing frequency up to 0.05% at 20 kHz and surges appear when the signal is limited, so we recommend it for high quality amplifiers I would not. Experiments with more advanced microcircuits that radio amateurs want to conduct may lead to positive results.

To implement circuits with the structure shown in Fig. 5 (also proposed by Peter Blomley in 1971), current-controlled splitters are required. Instead of the diode-biased splitter circuit that the author used, let's consider the splitter circuit using current mirrors, the circuit and operating graphs of which are shown in Fig. 16-17.

In such a circuit, in addition to splitting, you can immediately set the necessary minimum and quiescent currents! Let's turn on two Wilson current mirrors on complementary and matched pairs of transistors, ground the emitters of the left transistors VT1 and VT2, and apply the input current Isign to the right emitters VT4 and VT5. Output currents Iout1 and Iout2 flow in collectors VT3 and VT6. They are the sum of currents: 1/2 Isign +Iok1+Imin1 and 1/2 Isign +Iok2+Imin2. Current diagrams are shown in Fig. 17. The current-controlled splitter shown in Figure 16-1 has excellent frequency response, high linearity over a wide range of input currents, sufficient symmetry, and the high output impedance required in the circuits of Figures 4-6! A splitter with voltage control (Fig. 16-2) is characterized by a higher input resistance, worse frequency response and linearity, however, with the introduction of OOS from the amplifier output to the emitter resistor Re, all characteristics become acceptable. Possible schemes for constructing ITUN (for a positive half-wave amplifier) ​​are shown in Fig. 18 - 21.

A real circuit with a current splitter according to Fig. 19 is shown in Fig. 22

The output voltage is only 1-1.5 V less than the supply voltage! Output voltage slew rate 100 V/µS at 600 kHz bandwidth (without input filter R1-C2). Output impedance less than 5 mOhm. Signal delay 300 nS. The amplifier can withstand the full output voltage for sine and square waves with a frequency of 150 kHz!, and also does not burn during a load short circuit and when only one supply voltage is supplied! The quiescent and minimum currents of 30/120 mA are inherent in the splitter itself (resistors R18, R19, R22, R23), but to implement them, you need to set the zero (initial) currents of the ITUN output amplifiers using resistors R25 and R34. At such currents, Kg is less than 0.01% even at a frequency of 20 kHz and a power of 80 W / 8 Ohm.

A simple and reliable circuit with a current splitter and an operational amplifier according to Fig. 21 is shown in Fig. 23

Here, the super-A mode with currents of 30/100 mA and zero at the output are set automatically! Bandwidth 8 Hz - 520 kHz, output voltage slew rate of at least 40 V/µS. For the LM4562 op-amp, the distortion is the same as in the previous circuit, but the amplitude of the output voltage is 4 V less than the supply voltage (for an 8 Ohm load).

Figure 24 shows the use of a current splitter in asymmetrical circuits (according to Figure 6-2).

Here the super-A mode is adjusted to 30/130 mA by resistors R29 and R34. The parameters are identical to the diagram in Fig. 22.

Fig. 25 shows a successful option for including a splitter in the local feedback of the output stage, which made it possible to create an almost ideal output stage (in the absence of settings for the super-A mode) with a high input impedance and excellent frequency and amplitude characteristics. For the entire amplifier: operating supply voltage from 20 to 50 V, output voltage amplitude is 2.5 - 3 V less than the supply voltage, output voltage slew rate 80 V/μS with a bandwidth of 400 kHz, super-A mode with quiescent and minimum currents of 110 /30 mA, respectively, limiting the signal without emissions, playing a square wave with a frequency of 150 kHz and an amplitude of +/- 25 V, harmonics less than 0.003% even at 20 kHz.

All amplifiers described above can be reconfigured, if desired, to operate in modes A, B or AB, and with stabilization of the selected mode. I note that minimal distortion in UMZCH can of course be obtained in mode A, but measurements show that the difference between modes A and super-A appears only at frequencies of 18-20 kHz and only a few thousandths of a percent, which no “absolute” can hear. hearing! Douglas Self, in the fifth edition of the book "Circuit Design of Power Amplifiers. Handbook. (2009)" writes that back in 1975 he investigated the super-A mode (the author called it the Peter Blomley method), but found some "artifacts at the crossover point" and evaluated it "as having no commercial prospects", which seems unfair to me. The amplifier circuits proposed in this article work great and prove that modern amplifiers should only be made in the super-A class without the headache of “where to install thermal sensors and the dynamics of their operation.” And in Douglas Self’s “Handbook” you need to add the chapter “Output stages with deep OOS and good thermal stability of modes”! (I think so)

In conclusion I would like to say that UMZCH circuits with mode stabilization, they are quite reliable and can be made from inexpensive components, while the parameters of the amplifiers will satisfy the most demanding lover of high-quality sound. Some schemes (as in Fig. 23) are so simple that they can be recommended even for novice radio amateurs. Many schemes could certainly be improved! A new (or rather, well-forgotten old) super-A class is waiting for its researchers! The circuit design of the amplifiers also allows microcircuit design in the form of an original UMZCH microcircuit or a powerful operational amplifier, but that’s a completely different story (and hardly Russian)...

Alexander Gladky

Chapter 3 examined the principles of constructing power amplifier circuits operating in modes A, B or AB. It has been shown that the most favorable mode for output power amplification stages is class AB mode. Schematic diagram a push-pull power amplifier based on the same type of bipolar transistors, operating in class AB mode, is shown in Fig. 4.26. A small voltage bias is applied to the bases of the transistors using resistors.

Instead of a resistor, you can use a forward-biased diode, which creates a bias voltage based on the transistor to ensure class AB mode.

The diode also carries out thermal compensation of the operating point of rest, since when the temperature changes, the voltage at the emitter junction of the transistors and the voltage drop across the open diode change in the same direction. To obtain a greater thermal stabilization effect, the diode and transistors should be selected.

Calculation of output power, efficiency and nonlinear distortions in the class AB power amplification stage can be carried out with a sufficient degree of accuracy using formulas (3.14), (3.16), (3.19) derived for class B mode in § 3.2.

The transformers used in the considered circuits do not allow reducing the size and weight of power amplifiers and worsen their amplitude-frequency characteristics. Manufacturing transformers is expensive manual labor, scarce materials, and as circuit elements, transformers have low reliability. Therefore, transformerless push-pull power amplifiers, built on a pair of transistors of different types of electrical conductivity, are now widespread (Fig. 4.27, a).

The circuit consists of two single-ended emitter followers (arms) operating alternately during one half-cycle of the input signal. The arms are powered separately, from two opposite-polar DC voltage sources, united by a common bus, which is usually grounded. Due to the different types of electrical conductivity of the transistors, the cascade does not require paraphase input voltages.

Negative feedback reduces nonlinear distortions, as well as the influence of shoulder asymmetry. However, in circuits using emitter followers, the output voltage cannot exceed the input voltage, i.e., essentially only current amplification occurs. The cascade (Fig. 4.27, a) works as follows.

In the absence of an input signal, the point has zero potential. On the basis of each transistor, due to the divider, a constant pressure bias equal to the voltage drop across the corresponding diode and ensuring operation of the cascade in class AB mode.

If we neglect the bias current of the base of the transistor and assume that a current flows through each diode

With a positive half-wave of the input voltage amplitude, the diodes remain open. Voltage is supplied to the bases of the transistors. In this case, the transistor is turned off, and the base current of the transistor increases by the amount

The current through the diode becomes equal

where is the current through resistor R at a positive voltage.

The current will become equal to zero, i.e. the diode will close, at the maximum value, which can be determined from formula (4.84), putting in it. After transformations we get

Thus, to expand the dynamic range of the input signal, it is necessary to reduce the resistance of the resistor R in the bias circuit. However, as R decreases, the input resistance of the emitter follower, which makes up the arm of the cascade, is shunted.

With a negative half-wave of the input voltage, the transistor is turned off and the transistor current increases.

The processes of converting the input signal in the power amplification stage for positive and negative half-waves proceed in principle the same way. Therefore, formulas (4.83) and (4.84) ​​For both half-waves of the input signal are identical and differ only in the indices corresponding to the open transistor.

The graphical calculation of a transformerless cascade is made using the output characteristics of the transistors and does not differ from the graphical calculation of a cascade using. In this case, the role of resistance in a transformerless cascade is played by resistance.

To determine the input resistance, input power and nonlinear distortions of a transformerless cascade, you should use dynamic input characteristics, when constructing them, voltage should be plotted on the abscissa axis rather than voltage.

The presence of two power supplies in the circuit shown in Fig. 4.27, but may cause certain inconveniences when using the circuit. To replace two power sources with one, a separating capacitor of sufficiently large capacity is connected in series with the load (Fig.). For direct current, the transistors of the circuit are connected in series. Therefore, with identical parameters of the transistors, the constant voltage on the separate capacitor constitutes and is the “power source” for the transistor.

The collector-emitter voltage of the transistor is equal to .

To eliminate distortion of the output signal due to the capacitor, it is necessary that the voltage remains constant during the negative half-cycle (transistor open) of the input sinusoidal signal with a frequency corresponding to lowest frequency bandwidth. Then the change in voltage across the load will be determined by the change in voltage at the emitter of the open transistor.

The capacitance of the capacitor is selected using the relation

where is the output resistance of the emitter follower of one of the amplifier arms.

The method for calculating the cascade does not differ from the method for calculating the considered power amplification stages, i.e., it is performed using the static characteristics of the transistor of one arm. It should be taken into account that the resting operating point corresponds to the supply voltage level of the transistor of one arm.

The disadvantage of transformerless cascades shown in Fig. 4.27, is the large difference in parameters for different types of electrical conductivities. To eliminate this drawback, the industry produces “pairs” of transistors with the same parameters, but different types electrical conductivity, so-called complementary transistors, the range of which corresponds to different levels amplifier output power, for example.

To increase the load power of power amplifiers based on emitter followers, composite transistors are used. The schematic diagram of such a power amplifier is shown in Fig. 4.28. In the circuit (Fig. 4.28), instead of resistors R that determine the current of the bias diodes, direct current sources I are used, which make it possible to expand the dynamic range of the input signal.

Indeed, replacing in the formula with and equating , we get

In addition, direct current sources, having high internal resistance, do not bypass the high input resistance of the emitter followers to composite transistors, which is also a significant advantage of the current source over conventional resistors.

As a source of direct current, you can use a transistor connected according to a common base circuit, the input circuit of which ensures a constant emitter current, i.e. Then, with various changes in the collector voltage, the operating point will move only along one branch of the family of output characteristics (Fig. 4.29) and the collector current will remain almost constant.

More precisely, the change in the collector current with a change in the collector voltage of the transistor and a constant emitter current is determined by the value of the differential resistance of the collector junction

which in the OB scheme is large and amounts to several (compare with in the OE scheme).

In the diagram of Fig. 4.30 DC sources are made using transistors. Current flows through each transistor

where is the voltage drop across the resistor or the stabilization voltage of the zener diode, which, obviously, must exceed the voltage at the emitter junction of the transistor.

In addition to zener diodes, in the bias circuits of the transistor, you can use a red LED, the voltage drop across which in the open state is 1.8 V, or two rectifier diodes connected in series.

The transistor emitter current is selected from the condition

where is the amplitude of the transistor base current.

The current in the divider is chosen equal to the collector current of the transistor. Then the resistances are found from the formula

The task of the output stages is to provide the specified power to the load. The voltage gain is a secondary parameter for the output stages; - for them, the most important are the efficiency and the coefficient of nonlinear distortion when providing a given power.

Output stages typically consume the bulk of an amplifier's power, so high efficiency is essential. This is especially important for integrated circuits, in which the power dissipated by the crystal is limited. As for the nonlinear distortion factor, it is of no small importance for output stages, since in such stages the amplified signals are maximum.

The efficiency is defined as the ratio of the output power of the cascade to the power taken from the power source Ucc: Efficiency = u„1„,/2u„1,р, where U„, !„ are the amplitudes of the output current and voltage; Icp is the average value of the current consumed by the cascade.

The coefficient of nonlinear distortion characterizes the difference between the shape of the output signal and the shape of the input signal, which is due to the nonlinearity of the transfer characteristic of the cascade. Nonlinear distortions are characterized by the appearance of new harmonics in the output signal that are absent in the input signal. The characteristic of nonlinear distortions is the ratio of the total power of higher harmonics, starting from the second, to the power of the first harmonic (at the input signal frequency).

The permissible value of the nonlinear distortion factor is determined by the specific requirements for a particular equipment. For example, when reproducing sound in equipment of average quality, distortion of 2...3% is allowed; in high-class measuring devices and amplifiers, its values ​​are significantly less.

As noted above, there are several types of output stage operating modes.

Class A is characterized by minimal nonlinear distortion and low efficiency. Class B is characterized by the fact that the operating point in rest mode is located on the boundary of the quasi-linear section, which corresponds to the off state of the transistor. Obviously, in this case only the positive half-waves of the input signal are amplified. Therefore, the output voltage turns out to be significantly non-sinusoidal, i.e. contains a large number of harmonics. Analysis shows that the coefficient of nonlinear distortion in class B, regardless of the signal amplitude, is about 70%, which is unacceptable in most cases. Class B mode is implemented in a so-called push-pull circuit, consisting essentially of two amplifiers, one of which amplifies the positive half-wave of the signal, and the other - the negative. Under load, these half-waves add up and form a full sine wave.

In Fig. 7.13, and the simplest one is shown push-pull circuit class B, made on complementary transistors (transistors of different conductivity). Load Rn is included in the emitter circuit of transistors operating in voltage follower mode. In rest mode, both transistors are locked, since the voltages at the emitter junctions are zero. During the positive half-wave of the input signal Ui, transistor VT1 opens, and during the negative half-wave, transistor VT2 opens. The power gain is close to the ratio of the emitter and base currents, i.e. equals B+1.

Despite the obvious simplicity of the circuit in Fig. 7.13, and it is characterized by relatively large nonlinear distortions, which is associated with the presence of the so-called “heel” on the input current-voltage characteristic of bipolar transistors. Obviously, such distortions will be especially significant for small input signals with an amplitude comparable to the base-emitter voltage at the operating point. To eliminate this drawback, separate circuits are used to supply bias to the bases of transistors (Fig. 7.13, b), which ensures class AB mode.

When constructing an output stage using transistors of the same type, the circuit in Fig. is used. 7.13, c. In it, transistor VT2 is open during both half-cycles. In rest mode, the transistor current is selected so that the collector potential VT2 is equal to zero. In this case, the diode VD and transistor VT1 are locked; there is no current in the load. During the positive half-wave of the input signal, the collector potential VT2 decreases, at the same time the diode VD opens and current begins to flow through the load. Transistor VT1 remains closed, since the forward voltage E on the diode creates a reverse bias at the emitter junction. During the negative half-wave, the potential of the collector VT2 rises, transistor VT1 is unlocked and the current caused by transistor VT1 flows through the load. In this case, the diode is locked, since the forward voltage E at the emitter junction creates a reverse bias on the diode.

In order for the diode VD (with a positive half-wave) or transistor VT1 (with a negative half-wave) to open, the collector potential VT2 must change by ±E (base-emitter voltage in static mode) compared to the rest potential. Consequently, the minimum amplitude of the input signal to which the cascade in question reacts is E/K, where K is the gain of the cascade on transistor VT2. To study the cascade in Fig. 7.13, in the diagram in Fig. is used. 7.14.

Rice. 7.14. Circuit for studying the output stage

Rice. 7.15. Single-supply push-pull output stage

Other schemes for implementing output stages are possible, including those with single-polar power supply. One of them is shown in Fig. 7.15. Its peculiarity is that the capacitor Ck, connected in series with the load Rn, after charging it to a voltage E equal to the voltage at the emitters of the transistors in static mode, operates during one of the half-cycles as a power source.

In powerful output stages based on emitter followers, a short circuit at the output, as a rule, leads to failure of the transistors due to excess collector current acceptable value. To protect against short circuits, small current-limiting resistances (several ohms) are included in the emitter circuits of powerful output transistors or additional transistors are introduced that open only at high load currents and, by shunting the input circuit, limit the output current to a safe level. One of the possible protection schemes using additional transistors is shown in Fig. 7.16.

Rice. 7.16. Output stage with short-circuit protection

The protection circuit works as follows. When there is a short circuit in the load, the current through the resistance Ro increases and creates a voltage drop, which opens transistors VT5, VT6 during the corresponding half-cycles. Finding themselves in saturation mode, they bypass the input circuit of a powerful amplifier stage. As a result, the input voltage is limited by resistance Ri and the currents of transistors VT3, VT4 do not exceed the values ​​at which they operate in nominal mode. Such protection has high performance and ensures reliable operation of powerful amplification stages. When introducing it, it is necessary to have an additional resistor Ri, the resistance of which is selected based on the minimum permissible value of the load resistance of the pre-amplifier to which the output stage is connected.

Test task

1. By selecting resistance RI in the circuit in Fig. 7.14 set the static mode recorded by the devices at R2=Rn=100 Ohm. Determine the stage gain and the maximum input signal at which it is transmitted to the output without distortion (determined visually).

2. Draw up a diagram for studying the output stage (Fig. 7.15) and carry out its simulation.

The main purpose of the output stage is to transfer the maximum and required power to the load, close to the maximum for of this type transistor, with the lowest power consumption from the power source and permissible levels distortions.

Therefore, the output stage is a power stage. The main indicators of such a cascade are:

· Given to load power,

· Nonlinear distortion level and permissible transmission frequency.

Nonlinear distortion and efficiency depend on the initial resting point of the transistor. Therefore, the operating mode is important when choosing an output stage. With large signals, nonlinear distortions can arise due to both the nonlinearity of the input and output characteristics of the transistors.

Based on the above reasoning, it is possible, based on the output characteristics of the transistors in Fig. 4.1. show what amplifiers can be three classes

1. A - the rest point is chosen such that when moving along the load line it does not go into any nonlinear zone.

2. B - the rest point is in the extreme right position on the characteristic I b = 0. Thus, such an amplifier amplifies only one half-wave of the input signal. As a rule, such amplifiers operate in a push-pull circuit.

3. AB - Intermediate class, it allows you to reduce nonlinearity, but does not eliminate it completely.

With stringent requirements for nonlinear distortion, the output stages operate in class A. High efficiency. can be obtained in classes B, AB.

As a rule, the output impedance of power amplifiers is high and the load resistance is low, and therefore transformer coupling is used in the cascades, which makes it possible to obtain high values ​​of undistorted power.

When the load is switched on by a transformer, the DC component of the output current does not flow through the load, which reduces the power consumption and increases the efficiency.

Fig.2.34. Power amplifier output characteristics.

Class A Amplifiers

Such amplifiers are designed to produce a certain power at the load. Such amplifiers use a transformer connection with the load (Fig. 2.35). When the load is switched on by a transformer, the DC component of the output current does not flow through the load, which reduces the power consumption and increases the efficiency.

Quiet mode (U input = 0). Due to the displacement, currents appear: I bp. , I kp = βI bp +(1+β)I kbo

If the transformer is ideal, the resistance of the primary winding of the transformer to direct current is zero and U kp = E k

When Uin >0, an increment appears ΔI b, ΔI k = βΔI b. The load is:

R n / =R n ω 1 2 /ω 2 2 . As was said earlier, the necessary parameter is efficiency.

Where

Fig.2.35 Single-ended power amplifier.

Fig.2.36. Dependences η = f (ξ), Р к =f(ξ), for class A amplifier

Based on the obtained curves, the following conclusions can be drawn:

1. Efficiency - max value is obtained with a significant input signal.

2. The power consumed from the source does not depend on the magnitude of the input signal.

3. The maximum value of power loss is obtained in rest mode.

Such amplifiers are designed to transmit bipolar signals. However, they have a number of disadvantages:

· Low efficiency., especially with a small input signal,

· R o does not depend from the input signal and in rest mode is wasted,

· Availability of a transformer determines the unfavorable nature of the frequency characteristics,

· Impossibilityь transmission of unipolar signals

Class B amplifiers

In such amplifiers the load is connected directly to the collector circuit

Figure 2.37. In rest mode, when uin = 0, no bias is applied to the base of the transistor and

I kp = 0, P n = 0, i.e. There is no heating of the transistor in rest mode. When a positive input signal is applied to the base, the collector current increases, and a voltage drop appears across the collector resistance. With a negative signal, the output voltage is zero, i.e. Such an amplifier can amplify signals of the same polarity. This eliminates the use of a transformer for communication with the load.

Fig.2.37. Single-ended Class B amplifier

Let's determine the efficiency. cascade for the case of the specified signal. We determine the power supplied to the load taking into account the fact that in this case the effective value is U out = U out

/R n =

The power consumed from the source depends on the average current flowing through the load
ξE k 2 /R n

We get the efficiency ή=ξ

From the considered curves in Fig. 2.37. the following conclusions can be drawn:

· Efficiency class B cascade is higher than in the diagram in Fig. 2.36, especially for small and medium signals u input.

· The power consumed from the source E k is minimal in rest mode and increases with increasing uin.

· Power losses are maximum at average values ​​of ξ, but much less than the maximum power losses in scheme 2.36. At small ξ, P k is small, since the currents through the transistor are small, at large ξ, P k is also small, since the voltage drop across the load is large, and the voltage drop across the transistor u k = E k – u out. few.

All of the above allows us to conclude that class B amplifiers have advantages over class A cascades. The impossibility of amplifying bipolar signals has been overcome in a push-pull power amplifier.

Class B push-pull power amplifier stage

One of possible options such an amplifier is shown in Fig. 2.38

In rest mode, both transistors are locked. When a positive input signal is applied uin. the current i k1 p-p-p transistor V 1 increases. The circuit works the same way as the cascade in Fig. 4.4 transistor V 2 is locked

Figure 2.38. Class B push-pull power amplifier.

At a voltage of negative polarity, transistor V 1 is locked, current i k2 p-p-p transistor V 2 flowing through the load increases. Thus, the transistors come into operation alternately depending on the polarity of the amplified signal. A voltage u k = E k + u out is applied to the locked transistor. , which in the limit at large ξ tends to 2E k, which must be taken into account when choosing a transistor

In multistage amplifiers, the last (output or final) stage is the stage for amplifying the power delivered to the payload. In this case, the output power of the PA cascade must be sufficient to drive the load connected to the entire output circuit. The output stage of the PA must maximize the power of the amplified signal with an acceptable nonlinear distortion factor and higher efficiency.

There are single-cycle or push-pull PA output stages, which can be assembled using powerful amplification tubes, or transistors, or gas-discharge thyratrons.

Single-ended power amplification stages. Such PAs operating in class A mode make it possible to select into the load output power useful signal from fractions of a watt to 3 ÷ 5 W with an electrical efficiency of up to 10 ÷ 30% and minimum permissible levels of nonlinear distortion in a given frequency band.

In this case, the optimal value of the load resistance connected directly to the output circuit of the powerful stage is selected based on the relations Ra = R n= (2÷ 4) * Ri - for triode circuits and Rн = Ra ≈ (0.1 ÷ 0.5) * Ri - for PA stages on. high-power pentode or beam tetrode. Moreover, the circuits of such PA cascades and the methods of their graphical-analytical calculation are similar to the previously given circuits of voltage amplifier cascades (see Fig. 5, 7 and 8). Such simple PA stages make it possible to amplify the signal in terms of power with minimal nonlinear distortion in a wide frequency range.

A significant disadvantage of such transformerless PA circuits is that not only the useful alternating component of the anode current passes through the load, but also its direct component, significantly reducing the cascade efficiency and requiring more high voltage power supply E A. In addition, to maximize the use of useful output power that can be transferred by a transformerless final stage to an external load, it is necessary to maintain the equality of the optimal value of the output resistance of the output circuit of the PA cascade with the value of the external load resistance R n, included directly in this circuit, that is, R out= R n.

However, in practice, in most cases the load resistance R n may be less than the above-mentioned optimal value of anode resistance R a. This is explained by the fact that the winding of an electrodynamic loudspeaker, an electromagnetic relay, an electric motor, an electric contactor, a step finder, a recorder, a sound recording and sound reproducing head, a two-wire subscriber or feeder line, etc., which have small resistance (units, tens, hundreds of Ohms, units of kOhms).

Therefore, if R n< R вых к-да , то внешняя нагрузка включается в выходную цепь каскада УМ при помощи выходного трансформатора, согласующего величину Rн с оптимальной величиной выходного сопротивления каскада R вых к-да . При этом сопротивление внешней нагрузки, включенной во вторичную обмотку трансформатора, перерсчитывается в приведенное сопротивление его первичной обмотки, включенной в выходную цепь каскада, по следующей формуле:

where is the transformation ratio

More precisely, the optimal value of the equivalent resistance of the PA cascade can be determined graphical method, using the most appropriate load line on the family of anode characteristics (Fig. 14) of the selected high-power amplification lamp, that is, the segments about and oa in the appropriate units of measurement:

Thus, according to the variable component of the anode current, the optimal value of the reduced resistance of the anode load Rout k-da can reach from units to tens and hundreds of kilo-ohms.

Using the same graph, using the ABC triangle you can determine the useful power in the load

The efficiency of powerful transformer cascades of the PA is higher than that of transformerless ones, since the quiescent current I a0 flows only through the low active resistance of the primary winding, bypassing Rн. Wherein

where Po = I a0 * E a - total power in class A mode consumed from the power source.

It should be borne in mind that single-cycle PA transformer cascades have a narrower frequency band, larger dimensions, weight, and higher cost, which reflects their disadvantages.

In Fig. Figure 15 shows typical circuits of single-ended transformer stages of a PA based on a powerful triode (a) and a beam tetrode (b), operating in class A mode with automatic shifting of the operating point.

In these circuits, the purpose of each element of the PA cascade is similar to the previously discussed circuits of voltage amplifier cascades with an anode load (Fig. 6 and 8).

As can be seen from the graphs in Fig. 16, to obtain the optimal value of useful output power

it is necessary to apply an input voltage with amplitude |±U to the input of the PA cascade max| ≈ |-U c 0 |, taken from the pre-amplifier stage or from the input signal sensor. In this case, the load line should pass almost tangentially to the permissible power curve P and additional, without crossing it.

Since in class A mode the operating point is in the middle of the straight section of the input dynamic characteristic of the cascade, this ensures the condition of operation with minimal nonlinear signal distortion.

Triode PA cascades have less nonlinear distortion than PA cascades based on pentodes or beam tetrodes.

However, in most cases, the electrical efficiency of the PA cascade in class A mode practically exceeds 10 ÷ 15% for triode circuits and 15 ÷ 30% for high-power pentode and beam tetrode circuits.

It must be borne in mind that in PA cascades with a transformer output, with a low active resistance of its primary winding (r 1tr = tens ÷ hundreds of Ohms) the anode voltage in rest mode is only slightly less than the power source voltage E a, that is

For triode circuits,

For circuits using pentodes or beam tetrodes with an additional screen grid circuit.

Therefore, the DC load line on the family of static anode characteristics (Fig. 16) goes very steeply, at a large angle

In the dynamic mode of operation, when a sinusoidal (harmonic) input signal is supplied to the input of the transformer stage of the PA at the optimal value of the reduced load R eq, the highest voltage E a the maximum between the output electrodes increases almost twice (and sometimes more) compared to U a0. This phenomenon is explained by the fact that when the output current decreases, the back EMF of the inductance of the primary winding of the transformer is added to the value of E a, which delays the process of decreasing the anode current. Therefore, in the dynamic operating mode of such a PA cascade, the load line for the alternating component of the anode current is determined by the value of R eq and E a max > Ea and, passing through the non-working point through which the DC load line passes, has a significantly smaller angle of inclination (Fig. 16)

When calculating the maximum output power of the transformer cascade of the PA, taking into account the efficiency of the transformer, the required value of the output power of the cascade is determined from the given value of the required useful power in the load Pnet, namely:

Then select an amplification tube whose permissible power dissipated by the anode, P and extra ≥ 6P out yes for triode and a P and extra ≥ 4P out to -yes for.pentode or beam tetrode. In this case, the voltage on the anode in quiescent mode is taken equal to Uа0 = (0.7 ÷ 0.8) * Ua add, and the quiescent current value is taken equal to

The useful power released in the load will be equal to P useful = η tr* P output k-da = 0.5 η tr* I ma*U ma =0,5 η tr* I 2 ma R eq.

From here you can determine the transformation coefficient

Voltage gain of the PA cascade

To take into account useful power losses in the output transformer, take the value of its efficiency within the limits specified in table. 1.

V. Mayorov, S. Mayorov - Amplifier devices based on lamps, transistors and microcircuits