Amplifier on complementary field-effect transistors. Complementary FET Hi-Fi Amplifier

Hi-Fi AMPLIFIER ON COMPLETE FIELD TRANSISTORS

E.PIRET.

The amplifier circuit is shown in Fig. 1. Through the RC-chain of the low-pass filter, the signal enters the complementary input stage (T1, T2, T3, T4). If desired, you can increase the capacitance of the blocking capacitor C1, but it makes sense to do this only in the case of a very low cutoff frequency of the sound-emitting system.

A 100 Ohm linearizing resistor R11 is included in the emitter circuit of the input stage, and a total negative feedback of about 30 dB is connected to the emitters. "Inside" the stage, between the collector of the "lower" transistor (T2) and the emitter of the "upper" (TZ), a second ("internal") feedback loop of about 18 dB operates. This means that with the exception of transistors T1, T2, both loops have the same effect on all other stages.

Fig. 1

Through the emitter follower (the main role of which is the shift of the constant voltage level), the signal from the input stage is fed to the voltage amplifier (T7, T8). In the emitters of the transistors, linearizing resistors are again installed here. The collector current of these transistors flows through the circuits that regulate the quiescent current of the power amplifier's field effect transistors.

Let's stop for a moment! The temperature coefficient Kt of field-effect transistors (i.e., gate voltage / drain current ratio) is close to zero. For small currents it is small and negative, for large currents it is small and positive. Sign reversal occurs for high-power transistors at a current of about 100 mA. The final amplifier operates at a quiescent current of 100 mA. Field-effect transistors "swing" through transistor emitter followers, in which, as is known, Km is positive. Therefore, it is necessary to use a pre-biased circuit that compensates for the temperature dependence.

The temperature dependence of the emitter followers is compensated by diodes D3 and D4.

The quiescent current of the field-effect transistors of the final amplifier is set by the potentiometer P at a level of about 100 mA.

Resistors (R29, R30) are installed in the gate circuits of field-effect transistors to prevent self-excitation. The circuit, consisting of diodes and zener diodes (D5 ... D8), prevents the emergence of a gate-source voltage, which is dangerous for field effect transistors.

In the source circuit of the field-effect transistors, there are resistors (R31 and R32) with a nominal value of 0.47 ohms. Of these, R32 is marked with an asterisk - in the prototype, its value was equal to zero. This resistor smooths out possible differences in the slope of the field-effect transistors. As a rule, the inclusion of R32 does not have a catastrophic effect on the amplification, one can expect an increase in distortion by an amount of the order of 20 ... 30%.

As usual, the RCL link at the output of the amplifier protects it from self-excitation at extremely high load reactive impedance.

Resistance Rx in the emitter circuit T1 at the input of the amplifier is used to accurately balance the amplifier. If R13 and R14 are of the same size (6.8 kΩ), and Rx is short-circuited, then the output bias is quite satisfactory. But if it is necessary to improve it, then R13 is reduced to 6.2 kOhm, and instead of Rx, a 1 kOhm potentiometer is temporarily connected. After about 30 minutes of "warming up" the amplifier, this potentiometer sets the output voltage level to zero. The resistance of the potentiometer is measured and a resistor with the value closest to the measured value is soldered as Rx. As a rule, when replacing D1 or D2, it becomes necessary to replace Rx.

Capacitor C9 performs frequency correction of the amplifier. It has a double effect: it performs, on the one hand, a "lagging" correction at a capacitive load of the collectors T7 and T8 and, on the other hand, "advanced", being connected not to ground, but to R21.

Resistor R34 prevents the occurrence of two different ground loops when two or more UMZCH are powered from one power supply. The input ground connects to the metal case or chassis and to the preamplifier, while the other grounds, which are essentially zero current return wires, are individually connected to the power supply zero point.

Mounting. The amplifier is assembled on a double-sided printed circuit board, the drawing of which is shown in Figure 2-3.

There is a solid grounding foil on the part side. Countersink at the points of "entry" of the outputs of the parts into the board prevents short circuits. The pins of the parts connected to the ground are soldered directly (without holes) to the grounding foil. In the assembly drawing, these points are marked in black.

Two terminal field-effect transistors are installed on aluminum corners, which are connected to the radiator, creating a thermal bridge, and both are attached to the board. They must be insulated from the corners and the board. The resistor in the emitter circuit "hangs in the air", since it is mounted on the wall. Resistors R29 and R30 for shortening the leads are soldered from the side of the tracks of the board. The heatsinks must not form a false ground with the "zero" foil, so the "zero" foil is interrupted by a deep scratch running parallel to the heatsinks. For normal cooling of field-effect transistors, a cooling surface of about 400 cm 2 is sufficient. Transistors T9 and T10 are attached to the "zero" foil through a thin mica plate. A short circuit can very easily occur here, so the installation must be carefully checked with an ohmmeter.

Coil L1 with a diameter of 10 mm consists of approximately 15 tightly wound turns of wire with a diameter of 0.5 mm (without a core). Resistor R33 is located along the L1 axis, and its leads are soldered together with the coil leads, and then attached to the board.

The three wires going to the power supply are twisted together. The two wires leading to the speaker are also twisted into a separate bundle (regardless of the previous ones). Since large currents flow here, their magnetic fields can significantly increase distortion - mainly at high frequencies.

Twisting the wires together causes the magnetic fields of currents flowing in opposite directions to cancel each other out.

The zero point of the power supply and the speaker lead are not connected to the chassis, and the wires leading to them are not stacked with other wires.

Power Supply. The power supply circuit is the simplest (Fig. 4). A transformer tapped from the middle of the secondary winding feeds a full-wave rectifier consisting of two groups of 2 diodes each. Ripple smoothing is carried out by capacitors with a capacity of at least 4700 μF (40 V). Such a unit can supply power to two power amplifiers.

The upper limit of the voltage of the secondary winding of the transformer is determined by the type of used transistors T7, T8. In the case of using a pair ВС 546/556, the supply voltage (in the absence of a signal) should not exceed 30 ... 32 V. These transistors "do not tolerate higher voltages". With a supply voltage of ± 30 V, a 220 / 2x22.5 V or 230 / 2x24 V transformer can be used. An amplifier with a supply voltage of ± 30 V can deliver a power of about 24 W (at 8 ohms) to the load.

The field effect transistors used in the power amplifier are very expensive. For the price of one such transistor, you can get the rest of the set of parts. The question involuntarily arises as to whether the excess costs are compensated by the expected improvement in quality. The answer to this question depends on many circumstances, because:

We are talking about subjectively perceived distortions, so the sound sensations from different people will be different;

The perception of distortion depends on the music being played. When playing purely "author's" electronic music, it makes no sense to talk about distortions, because it is impossible to know whether or not these distortions were in the original material;

Reproduction of music coming from CD is problematic. According to the "critical ears" and the author, this music has a specific color. Reproduction from a good analogue record or directly from a concert gives excellent quality.

Translated by A. Belsky.

Here's a simple 100 watt HI-FI MOSFET amplifier. The main feature of this amplifier is its simplicity of design and assembly. It should be noted that many hi-end amplifiers have very simple but good quality designs. Less detail means fewer problems.

The circuit is pretty simple. MPSA56 - differential input. These transistors were chosen for their sound quality as a result of years of experience. Complementary pair of output transistors 2SK1058 and 2SJ162. For better speaker protection, it is recommended to add a power-on delay.

Power transformer for 8 ohm output load 35-0-35 volts and at least 3 amps. Rectifier and filter on 2 capacitors 4700 uF 63 V. This supply circuit is for one channel.

The amplifier is intended for home use, but with its peak power of 300 watts, it will quite cope with a small concert hall.

Note: This circuit does not have AC DC protection. This HI-FI amplifier is not suitable for beginners, as special measuring tools and skills are required for proper tuning.

The harmonic spectrum of this transistor amplifier is selected in such a way that it sounds like a good old pentode single-cycle.

Over the past 10-15 years, scolding the sound of transistor amplifiers and extolling the advantages of tube amplifiers has become almost the responsibility of audio critics. I think that the specific sound of the former is associated with a purely formal approach to their construction. Nowadays any audiophile with a little bit of listening experience knows that parameters like "0.002% THD at 100 watts of power" really say little about the musicality of the device. What does it depend on? Let's try to figure it out.

Hardly anyone will dispute the fact that the tube triode is the most linear element that has been invented by man in the last hundred years. Transistors, both bipolar and field-effect, are very far from it. But is it all so hopeless?

about the author

Jean Tsikhiseli. A somewhat unexpected combination of first and last name seems to symbolize the eclecticism of the genres of this designer. The assortment of the Time Wind laboratory, headed by Jean, includes a variety of projects: amplifiers based on triodes, pentodes in one- and two-stroke switching, and even, let's not be afraid of the word, on transistors. Belongs to the category of nuggets, for which it is a simple matter to make a capacitor on your own or wind an output trans. A constant participant of the Russian Hi-End exhibitions, he is modest in everyday life, does not impose his opinion on anyone. Moreover, it is worth listening.

It turns out not. It is known that there are three types of transistor amplifier stages: common emitter, common collector and common base. The first type is the most widespread, but, unfortunately, it has such distortions that there is no need to speak of any linearity. The common-collector stage, or emitter follower, is much better, but its gain is less than unity. It is usually used as a matching device when it is necessary to obtain a large input impedance and a small output impedance, in particular, to match a loudspeaker with a voltage amplifier. The optimum is a cascade with a common base - it has less distortion and a wider bandwidth (which is why it is often used in RF circuits), and the gain is quite decent. As a result, we are left with only cascades with a common collector and a common base as bricks for building an amplifier. Move on.

Those who are familiar with industrial amplifier circuits have probably noticed that the number of transistors there can reach hundreds of pieces per channel. Passing through each pn junction, the signal degrades, so the conclusion suggests itself: to build a really high-quality amplifier, you need to use the minimum possible number of them, and I think hardly anyone would disagree with this. Now let's talk about feedback. Perhaps everyone knows that it is better to do without it, but the nature of transistor amplifiers is such that this is hardly possible. The only thing we can do is to make the OS depth the minimum necessary.

Now, briefly about the modes of operation of transistors. Even with a cursory analysis of their output characteristics, it is easy to see that only in class A they have the greatest linearity. But in nature, you have to pay for everything, and here's an example: the output stage on a complementary pair of bipolar transistors, included in class A, due to overheating fails after a few seconds. To make such a scheme workable, you need to put 10 instead of one pair, and this already contradicts the requirement to use the minimum possible number of active elements. In most cases, there is no gain here, and the most reasonable thing is to put the output stage in the "forced AB" mode, and such a scheme will be durable and reliable. But all the other cascades must work in "pure" class A. But that's not all. Each specific type of bipolar or field device has an optimal collector (drain) current at which it has maximum linearity, and it must be used in this mode. All of these requirements are necessary, but far from sufficient to achieve our only goal - good sound.

Another and very important condition is the correct selection of the element base, namely transistors, diodes, capacitors, resistors, wires and solder.

After several months of testing and blind listening, it was found that the following types of elements are most suitable for the described circuit: BSIT (Bipolar Static Induction Transistor) - for the input stage, current generator and voltage amplifier; field-effect transistors as a source follower, bipolar - in a level shifter and a current generator, a pre-output push-pull stage and an output push-pull emitter follower.

Rice. 1. Schematic diagram of the amplifier.

Now about the passive components. The volume control should be taken with a high-quality and reliable ALPS, fixed carbon resistors, C1-4, and wire-wound in the emitter circuits of the output transistors. Paper capacitors at the input and feedback circuits, K42-11, MBM, etc. They may seem too bulky, but I do not recommend using other types due to the noticeable deterioration in sound. If you cannot buy branded electrolytes, then it is better to use K50-24 from domestic ones.

The input stage on VT1, VT2 is a single-ended differential amplifier with local current feedback, loaded onto a current generator on VT3. From the output of the differential stage, the signal is fed to the gate of the field-effect transistor VT4, which is turned on by the source follower. From the VT4 source, the signal goes through the VT5 KT9115A level shifter to the VT6 voltage amplifier. That, in turn, is loaded on the VT7 current generator and two series-connected push-pull emitter followers on VT8, VT9, VT10 and VT11. Series connected diodes VD7 - VD10 set the quiescent current of the output stage (approximately 0.2 A). By adding one more or more (the fifth diode is shown in the diagram with a dotted line), you can increase the quiescent current to 0.8 A and, thus, transfer the stage to class A. By selecting the resistor R7, set the zero potential +/- 10 mV at the amplifier output. It is not recommended to use trimmers here, so it is better to choose the desired value by soldering another, larger or smaller value in parallel with the 470 Ohm resistor.

Pairs of transistors VT2 and VT2, VT8 and VT9, VT10 and VT11 should be selected with the same gain value with an accuracy of at least 1%. A special device is used to protect acoustic systems from constant voltage at the amplifier output (Fig. 2).

Rice. 2

For reliable operation of the protection circuit, capacitors C1, C2 are better to use oxide-semiconductor tantalum series K53.

Now a few words about the power supply (fig. 3, page 14). It uses a 200 - 250 VA toroidal transformer with a shield winding that must be grounded. In order for the active resistances of the secondary windings to be the same, it is better to wind them in two wires and connect the midpoint to the chassis with a thick short wire. As rectifier diodes, KD2994A with a Schottky barrier, which have a high speed, are used. Electrolytic capacitors of the K50-24 type, and shunt capacitors - paper MBM, BMT. If you want to equip the amplifier with a protection device, to power it you will need an additional winding for a voltage of 18 V and a current of about 300 mA, as well as a simple rectifier with a smoothing filter.

Rice. 3

When installing the amplifier, you should pay attention to the quality of the connecting wires and solder. Installation must be carried out with a copper wire with a cross section of about 2 sq. mm, speaker cables costing 30 - 40 rubles are very well suited for this purpose. per meter. From solders I can recommend POS-61, it is inexpensive and you can buy it on any radio market. It is better to make printed circuit boards of 2 mm thick foil fiberglass and rigidly fasten to the bottom of the case using metal bushings. All transistors, except VT1, VT2, VT3, are attached through insulating gaskets to the bottom of the case, made of an aluminum plate 10 mm thick, which is also a heat sink.

The layout of the "earthen" buses also has a great influence on the sound. Signal and high current ground should be connected to the chassis at the same point, near the input connectors. The housing should be made of non-magnetic material. Manufactured in 1995 in the Time Wind laboratory using the circuit described above, the amplifier demonstrated sound quality comparable to that of a good tube pentode push-pull. Thanks to carefully selected distortion spectral content, the amplifier delivers rich mids, transparent highs and tangible bass.

The scheme has another obvious advantage - good repeatability and easy setup, since it was intended for small-scale production in an industrial environment.

Table 1. Amplifier Parts
Resistances
R1	1k	1/4 w	carbon
R2, R9	15k	1/4 w	carbon
R3	8k2	1/4 w	carbon
R4, R5	13	1/4 w	carbon
R6	24k	1/4 w	carbon
R7	150	1/4 w	carbon
R8	200	1/4 w	carbon
R10, R11	750	1/4 w	carbon
R12	5k6	1 w	carbon
R13	48	1/2 w	carbon
R14	24	1/2 w	carbon
R15, R16	100	2 w	carbon
R17	18	2 w	carbon
R19, ​​R20	0,47	5 w	wire
R21	10	2 w	carbon
Capacitors
C1	2.2 uF		MBM, K42-11 (paper)
C2	1000 pF		CSR, SGM (mica)
C3	3.9 pF		ceramics
C4	22 uF		MBM, K42-11 (paper)
C5	0.1 μF x 160 V		MBM, K42-11 (paper)
C6, C9	1 μF x 160 V		MBM, K42-11 (paper)
C7 - C11	2200 μF x 63 V		K50-24
Semiconductors
VD1 - VD10	KD522B
VT1 - VT3	KP959A		BSIT
VT4	KP902A		CMOS
VT5, VT7	KT9115A		bipolar
VT6	KP956A		BSIT
VT8	KT850A		bipolar
VT9	KT851A		bipolar

Literature:
1. P. Horowitz, W. Hill. "The art of circuitry", Moscow, "Mir", 1993
2. N.V. Password, S.A. Kaidalov. "Photosensitive devices and their application". Publishing house "Radio and Svyaz", 1991

When creating high-power amplifiers in the output stage, it is necessary to use parallel connection of specially selected and matched groups of transistors, which significantly complicates and increases the cost of manufacturing the amplifier. It is much easier and cheaper to use leaders in gain in this cascade and powerfully sti - insulated gate bipolar transistors(IGBT), since the questions of selection and installation of transistor groups disappear. But it is believed that such transistors can only work in switching modes. In addition, there are practically no complementary pairs among them.

Currently, there is a strong opinion that only stages with a symmetrical output on complementary transistors are able to provide high parameters of the UMZCH. This is due to the fact that almost all of them repeat the topology developed by Lin at the firm RCA back in 1956, - ​​an input differential stage, a second stage of voltage amplification and an output symmetrical push-pull stage - a current amplifier.But this structure is far from optimal if one of the arms of the output stage is built according to the Shikpai scheme, as is the case when designing UMZCH with powerful transistors of the same conductivity.

The main problem of an amplifier with an output stage based on transistors of the same conductivity is the potential instability caused by the fact that one of the arms of the output stage is covered by local negative feedback. As a result, the phase-frequency characteristics of the arms are significantly different. And this generates ringing and parasitic generation in the output stage and requires additional correction, balancing such an output stage, which reduces the overall cutoff frequency of the UMZCH and ultimately leads to an increase in distortion. Although such circuits do not cause enthusiasm among designers, nevertheless, transistors of the same conductivity are widely used in the output stages of powerful UMZCH microcircuits due to the low cost of production. Of course, among bipolar transistors, there are a lot of complementary pairs, and difficulties arise only with the selection of pairs of complementary transistors of the group IGBT, the attractiveness of using which is obvious. This hinders the use of such transistors, with their undeniable advantages over bipolar and field-effect transistors.There are power stage bridge circuits that do not require complementary pairs of transistors. But they are quite complex, and it is difficult to use effective feedback in them.,as a result, bridge circuits have not become widespread, except in car radios, where they are used due to the limited supply voltage.

Consider separately a single-ended push-pull output stage on IGBT (Fig. 1), when the upper transistor is switched on according to the scheme with a common collector and the lower transistor is switched on according to the scheme with a common emitter.

The dependence of the output voltage on the control current for the upper transistor will be: U H = l e (1 + R 3 * S) * R n, and for the lower transistor - U H = l e * R 3 * S * R H. It can be seen that these dependences of the output voltage are very close, and with an equal value of the slope and a large resistance of the resistors in the gate circuit(R 1, R 2) the output stage is practically symmetrical. But symmetry and linearity are different properties. And the remarkable property of this circuit is that the difference in the transistor slope can be compensated by the selection of resistors. This symmetry is unattainable for complementary field-effect transistors. The difference in the slope of complementary pairs of field-effect transistors reaches 300%, approximately the same difference in their input capacitance.

Of course, symmetry is only high at low frequencies, which is what audio frequencies represent. The challenge is to design a circuit that preserves symmetry over the widest possible frequency range. And here the Lin topology is no longer optimal.

But back to the diagram in Fig. 1. The disadvantage of the stage is that each arm requires its own signal generator, and as a result, difficulties arise in ensuring the thermal stability of the quiescent current of the stage. Much more convenient is the cascade excitation circuit in Fig. 2. Its attractiveness is that now two signal sources are not required, and control of such a cascade is much easier. Moreover, here the change in the resistance of the signal source R changes the current from the current source to the resistors in the gate circuit of the transistors, and the change in resistance R r leads to an antiphase change in the voltage at the gates of the transistors. When increasing Rr the upper transistor is unlocked and the lower one is locked, with a decrease Rr the upper transistor is locked and the lower one is unlocked. The total value of the currents across the gate resistors, at any value Rr, remains unchanged and is determined by the current source. ,,,That is, it converts the input signal into a control symmetrical antiphase current, but dozens of times different control voltage, for the upper and lower arms of the single-ended output stage, which is necessary to control the single-ended output stage. This is how the push-pull mode of operation of a powerful single-ended output stage is realized..The initial current of the output transistors and the thermal stabilization of the quiescent current are achieved by changing the current of one current source, since when the current of the current source decreases, both transistors are locked.

The construction of an output stage based on transistors of the same conductivity structure according to the proposed scheme is quite attractive by its simplicity, especially with a high output power of the amplifier (more than 100 W), when IGBT - transistors have a number of advantages over bipolar and field-effect transistors. In addition, according to the company's developers PLINIUS the sound with amplifiers on p-p-p structure transistors is better than on p-p-p transistors, and in expensive models they prefer an asymmetric output stage. This is explained by the fact that the transistors of the preferred structure are more linear and have better frequency properties, as well as a higher gain.

For effective use IGBT, as well as field-effect transistors of the same conductivity, I propose a new structure of UMZCH - an input cascode amplifier, then a composite stage on transistors of different conductivity with a current source and a zener diode, and, finally, a push-pull asymmetric output stage with transistors of the same structure. This structure with voltaddition th and auxiliary circuits is shown in fig. 3. The new structure creates the shortest signal path to the lower transistor, which has T the worst frequency properties and, despite its simplicity, has a large overall gain.

Consider the circuit in Fig. 3 in more detail. Input signal through a resistor R 1 , which determines the input impedance of the amplifier, goes to the base of the transistor VT 1. The inclusion of this transistor in the cascode allows the use of a low-voltage, high-frequency, low-noise transistor at the input and neutralizes the Miller effect, as well as reduces the influence of the common-mode voltage. Transistor VT 2 should withstand the required voltage, i.e..be relatively high voltage. Using a "broken cascode" instead of the usual one protects the transistors VT 1 and VT 2 from breakdown, since when the input signal is overloaded, the current increases VT 1 and VT 2 limited by resistor R 3.

Using a differential input amplifier instead of a cascode one will reduce the slope of the input stage by half and increase the noise of the input stage by 2 dB, and this, ultimately, will lead to an increase in distortion. It will also be necessary to select a pair of input transistors.

From the output of the cascode amplifier, the signal goes to the compound stage on transistors VT 3 VT 4, which perform the function Rr. These transistors are included in the OB-OE structure with a combination of emitters, which is optimal for the selection and use of transistors. Voltage and power gains of transistors VT 3 and VT 4 vary greatly, this requires application as VT 3 medium-power high-voltage transistor, the frequency properties of which, as a rule, are much worse than low-power low-voltage transistors. Therefore, turning it on in the OB mode is more effective than in the OE mode. Voltage gain for VT 4 not as big as for VT 3. Therefore, turning it into OE mode will not worsen the overall frequency response too much.

Choosing a suitable cheap high voltage transistor p-p-p structure for VT 3 does not cause problems, and the transistor VT 4 - low-voltage low-power pn-p structure of high-frequency transistors of wide application.

Field effect transistors as VT 1 ... VT 4 it is impractical to use, since they have a lower slope than bipolar T transistors, which will be equivalent to reducing the gain of the stages and the linearity of the amplifier as a whole.

In order to increase the maximum voltage amplitude for half-periods of positive polarity, a volt-additive is introduced in the form of a circuit R 6, C1. Although, instead of a voltage boost, an additional power supply can be applied, which will expand the range of the amplifier to the low frequency region. Zener diode VD 1 compensates for residual voltage drop across transistors VT 3, VT 4 in half periods of negative polarity with and thus reduces the saturation voltage at the minus supply.

The use of parallel feedback, instead of the more common serial feedback, makes the amplifier less sensitive (in terms of linearity) to changes in the impedance of the signal source. So, with its increase, the nonlinear distortion of the amplifier does not increase as it happens when using serial feedback.

A remarkable property of the proposed structure is the "natural" limitation of the maximum output current.The fact is that the voltage across the resistors R 5, R 7 can take maximum only twice the value of the original, and by choosing the resistance of the emitter resistors R 8, R 9 you can limit the maximum current of transistors by calculating it using the formula: Imax = (2 U start - Umax) / R e,

where U start - gate-emitter voltage of transistors VT 5, VT 6, at which a given initial current flows through the transistors; Umax - gate-emitter voltage of transistors VT 5, VT 6 when the maximum current flows through them; R e - resistance of resistors R 8, R 9.

Due to the fact that the maximum voltage across the resistors R 5, R 7 does not exceed twice the initial value (for example: if U ze start 5.7 V, then U ze max = 11.4 V), it makes no sense to install gate overvoltage protection.And since the currents of all amplifier devices are limited, there is no need for additional cascade protection circuits, which greatly simplifies the amplifier.

In practice, the gate-emitter voltage of transistors when the maximum current flows through them is not known in advance, therefore, by experimental selection R the choice of I max.

As you can easily see, R 8 and R 9 perform not only a restrictive, but also a linearizing function for VT 5 and VT 6, creating a local CBO in itself and x nonlinear elements.

A variant of the practical scheme for the implementation of a powerful UMZCH is shown in Fig. 4.

As can be seen from the given parameters of the technical characteristics, the described amplifier is not inferior in quality to the best amplifiers with a symmetrical structure, and such a high output power is realized on only eight transistors! Not a bad result with the cost of components of the order of 10 USD, taking into account the fact that selection and selection of transistor groups is not needed. In general, the scheme is one of the best in terms of cost / quality ratio.

The most detailed features of the UMZCH operation can be described in the full scheme (Fig. 4) as follows.Input signal, via C1 circuit, R 1, setting the lower cutoff frequency and input resistance, is fed to the base of the transistor VT 1. As input to NS brane microwave transistor KT368A (for a quick exit from saturation after an overload when the output signal is limited). A feedback signal is sent to the base of the same transistor through the C2 circuit, R 3.

Chain SZ, R 2, R 4, R 7 designed to set zero bias voltage at the amplifier output.Since the trimmer resistor R 7 over time, it can change the resistance, instead of it it is better to install a constant resistor selected when tuning. Diodes VD 2 and HL 1 set the offset to the base of the transistor VT 2 and at the same time carry out temperature compensation of zerovoltage at the amplifier output due to the same thermal coefficients of the transistor VT 1 and diode VD 2 (it also sets the bias voltage along the circuit R 2, R 4, R 7).

Capacitor C4 corrects the input stage. From the collector VT 1 signal through VT 2 goes to the base of the emitter follower on the transistor VT 3. Its task is to increase the input resistance and thereby increase the overall gain, as well as accelerate the blocking of the transistor. VT 5 and neutralization of the Miller effect. Zener diode VD 3 increases the supply voltage for VT 3 and thus accelerates the blocking of transistors VT 4, VT 5, increasing the speed of the leading edge.

From the emitter VT 3 the signal goes to the base of the transistor VT 5. Chain L 1, R 13 performs correction of the composite stage on transistors VT 4 and VT 5. From the collector of the transistor VT 5 the signal goes to the gate of the output transistor of the lower arm.From the collector of the transistor VT 4 a similar but antiphase current signal is fed through a zener diode VD 7 to the gate of the upper arm output transistor.

Chain R 11, C7 in VT 4 base implements inclusive correction of the output stage, increasing stabilityamplifier in limiting mode. Chains C 10, R 22 and L 2, R 24 increase the stability of the amplifier when changing the load resistance and with its capacitive nature.

Diode VD 8 halves the thermal power dissipated across the resistor R 20, due to the fact that only the charging current of the capacitor C8 flows through it. The quiescent current of the output stage, equal to 0.2 A, is set with a tuned resistor R 17.

For thermal stabilization of the quiescent current UMZCH diodes VD 5 and VD 6 installed on the heat sink next to the output transistors. Transistors VT 4, VT 6 supply with small plate heat sinks, since the heat power dissipated by them reaches 0.8 W. Light-emitting diode HL 2 used to set the bias of the current source on the transistor VT 6 and at the same time to indicate that the amplifier is on.

The output transistors must be installed on a radiator with an area of ​​at least 3000 cm 2. The use of a fan will sharply reduce its size, which will significantly reduce the dimensions and weight of the amplifier.

When you turn on the amplifier for the first time to protect the output transistors, the resistors R 19 and R 23 it is recommended to replace with higher resistance (up to 3…10 Ohm), and only after checking the voltage at the gates can the corresponding circuit be set to 0.10m and set the quiescent current. Moreover, for IRG 4PC 30W voltage U ze = 5.7 V.

As can be seen from the complete circuit (Fig. 4), the amplifier uses a rather complex frequency response correction (four capacitors and a choke, not counting the resistors). This is a small price to pay for the asymmetric structure to behave as well as symmetrically. th (with complementary devices) and obtain high stability of the amplifier in the restricted area. We can say that the first problem of achieving low distortions after choosing a structural circuit is the problem of choosing an amplifier frequency response correction, which creates the necessary amplifier stability margin with a large change in output currents and voltages, and at the same time provides a minimum phase delay in the operating frequency range. In most cases, it is the correction that becomes decisive, negating the advantages of many schemes..

The developer is always faced with a dilemma - whether to increase the depth of the overall feedback to improve the linearity of the amplifier or decrease its depth in order to increase the stability margin, which is necessary if the speaker impedancehas a complex character. And if the amplifiers sound differently, then to a large extent this is due to the stability margin, which is very noticeable at high levels. That is why UMZCH with "simple" circuits often show better results than those with a complex (often on microcircuits) structure. And each new cascade must be introduced after rigorous testing of the effectiveness of the new NS x elements. Moreover, an increase in the OOS depth in most cases does not give the desired result, but only worsens the stability margin. And here the correct assessment of the criteria of linearity and dynamic stability of the amplifier comes to the fore, which in turn depend on competent correction. Moreover, competent correction should minimize the phase delay in the operating frequency range without degrading the overall stability. A good correction is often much more effective than a new cascade.

Of course, the chosen method of zero balance at the amplifier output is far from the best, and it is attractive only for its simplicity. At the output of the UMZCH, a "floating" bias of up to several tens of millivolts can occur, but it does not noticeably affect either the sound or the operating point of the output transistors. To reduce the "zero" drift, it is useful to introduce a tracking unit on a precision microcircuit, even if this complicates the amplifier.

Applied transistors IRG 4PC 30W inexpensive, but they have a noticeable nonlinearity in the initial section and a large input capacitance. If you check the whole series of episodes IGBT, offered by manufacturers, you can surely find devices with greater linearity and lower input capacitance. The author did not have the opportunity to carry out such work. With the proposed transistors, linearity can be improved by two times by increasing the quiescent current to 0.5 A, but this will require an increase in the radiator area.

In conclusion, I want to note that if there is no need for a large amplifier power, then it is quite possible to use at the output instead of IGBT transistors field-effect transistors with an insulated gate and channel n -type, the linearity of which is noticeably higher. The amplifier will receive a higher linearity, while it is only necessary to select resistors for a different supply voltage and zener diodes for a different gate voltage. A reduced power analogue of the UMZCH on field-effect transistors, corresponding to the scheme shown here, has been successfully operated by the author for six years, delivering a lot of pleasant minutes when listening to various kinds of music programs at home.

Literature

1. Danilov A.A. Precision low frequency amplifiers M Hotline - Telecom, 2004.

2. Kozyrev In Amplifiers " Krell KAV -4- xi "," Audio Analogue Maestro "," Plinius 9200 ". - Audio Store, 2003, No. 6, p. 71.72.

3. Shpak S.V. Patent RU No. 2316891 dated 04/10/2006.

4 Douglas Self about a previously unnoticed source of distortion of transistor UMZCH with general OOS - Radiohobby, 2003, No. 3, p. 10.11.

5. Vitushkin A., Telesnin V. Stability of the amplifier and natural sounding. - Radio 1980, No. 7, p. 36.37.

Sergey Shpak Kazan Tatarstan

P.S. The site has already raised the topic of editorial errors, here is another example of such an error: -

[email protected]