Wideband UMZ with low distortion. UMZCH with complementary field-effect transistors. diagram, description UPS technical characteristics

Recently, designers of low-frequency power amplifiers are increasingly turning to tube circuitry, which makes it possible to achieve good sound with a relatively simple design. But you should not completely “write off” transistors, since under certain circumstances, a transistor UMZCH is still capable of working quite well, and often better than lamps... The author of this article had a chance to try a large number of UMZCHs. One of these most successful “bipolar” options is offered to the readers. The idea of good operation is based on the condition that both arms of the UMZCH are symmetrical. When both half-waves of the amplified signal undergo similar conversion processes, one can expect satisfactory operation of the UMZCH in a qualitative sense.

Even in the recent past, the introduction of deep environmental protection was considered an indispensable and sufficient condition for the good operation of any UMZCH. There was an opinion that it was impossible to create high-quality UMZCH without deep general environmental protection. In addition, the authors of the designs convincingly assured that, they say, there is no need to select transistors to work in pairs (arms), the OOS will compensate for everything and the spread of transistors in parameters does not affect the quality of sound reproduction!

The era of UMZCHs assembled on transistors of the same conductivity, for example, the popular KT808. assumed that the output transistors of the UMZCH were switched on unequally, when one transistor of the output stage was switched on according to the circuit with OE, and the second - with OK. Such asymmetrical inclusion did not contribute to high-quality signal amplification. With the arrival of KT818, KT819, KT816. KT817 and others, it would seem that the problem of UMZCH linearity has been solved. But the listed complementary pairs of transistors “in life” are too far from true complementarity.

We will not delve into the problems of non-complementarity of the above transistors, which are very widely used in various UMZCHs. It is only necessary to emphasize this fact. that under equal conditions (modes) of these transistors, it is quite difficult to ensure their complementary operation in push-pull amplification stages. This is well said in the book by N.E. Sukhov.

I do not at all deny the possibility of achieving good results when creating UMZCHs using complementary transistors. This requires a modern approach to the circuit design of such UMZCHs, with the obligatory careful selection of transistors for operation in pairs (switches). I also had the opportunity to design such UMZCHs, which are a kind of continuation of the high-quality UMZCH N.E. Sukhov, but about them - some other time. Regarding the symmetry of the UMZCH, as the main condition for its good operation, the following should be said. It turned out that the UMZCH, assembled according to a truly symmetrical circuit and certainly using transistors of the same type (with a mandatory selection of copies), has higher quality parameters. It is much easier to select transistors if they are from the same batch. Typically, copies of transistors from the same batch have fairly close parameters compared to “accidentally” purchased copies. From experience we can say that out of 20 pcs. transistors (standard quantity of one pack), you can almost always select two pairs of transistors for the UMZCH stereo complex. There were cases of more “successful catches” - four pairs out of 20 pieces. I’ll tell you about the selection of transistors a little later.

The schematic diagram of the UMZCH is shown in Fig. 1. As you can see from the diagram, it is quite simple. The symmetry of both arms of the amplifier is ensured by the symmetry of the transistors.

It is known that the differential stage has many advantages over conventional push-pull circuits. Without delving into theory, it should be emphasized that this circuit contains the correct “current” control of bipolar transistors. Transistors of the differential cascade have an increased output resistance (much higher than the traditional “swing” according to the OK circuit), so they can be considered as current generators (current sources). In this way, the current principle of controlling the output transistors of the UMZCH is implemented. It is very precisely said about the influence of resistance matching between transistor stages on the level of nonlinear distortion in: “It is known that the nonlinearity of the input characteristic of the transistor I b = f (U b e ) is most manifested when the amplifier stage operates from a voltage generator, i.e. The output resistance of the previous stage is less than the input resistance of the subsequent one.In this case, the output signal of the transistor - the collector or emitter current - is approximated by an exponential function of the base-emitter voltage U be, and the harmonic coefficient of the order of 1% is achieved at a value of this voltage equal to only 1 mV (!) . This explains the reasons for the occurrence of distortions in many transistor UMZCHs. It’s a pity that practically no one pays due attention to this fact. So what, transistors “die” in UMZCHs (like dinosaurs?!), as if there is no way out of the current circumstances except how to use tube circuits...

But before you start winding the labor-intensive output transformer, you should still tinker with the symmetrical transistor circuit of the UMZCH. Looking ahead, I’ll also say that UMZCHs using field-effect transistors were also assembled using a similar circuit design; we’ll talk about this some other time.

Another feature of the circuit in Fig. 1 is the increased (compared to traditional UMZCH) number of power supplies. You should not be afraid of this, since the capacitances of the filter capacitors are simply divided into two channels equally. And the separation of power supplies in the UMZCH channels only improves the parameters of the stereo complex as a whole. The voltages of sources E1 and E2 are not stabilized, and a voltage stabilizer (40 volts) must be used as E3.

Speaking about the theoretical problems of push-pull circuits and transistor UMZCH in general, it is necessary to analyze one more cascade (or several such cascades) - a bass reflex. Long-term experiments confirm the fact of a significant deterioration in the quality of sound reproduction due to these cascades. Having assembled a completely symmetrical circuit, and even with painstakingly selected parts, you have to face the problem of bass reflex circuits. It was found that these cascades are capable of introducing very large distortions (the difference in the shape of a sine wave for half waves could be observed on the oscilloscope screen even without the use of any additional circuits). The above fully applies to simple circuits of tube versions of phase inverter amplifiers. You select the values in the circuit in order to obtain equality in the amplitudes of both half-waves (sine waves) of the antiphase signal using a high-quality digital voltmeter, and subjective examination requires (by ear!) turning the trimmer resistor sliders away from this “instrumental” method of adjusting the levels.

Peering at the shape of a sinusoid on the oscilloscope screen, you can see “interesting” distortions - at one output of the bass reflex they are wider (along the frequency axis), at the other they are “thinner”, i.e. The area of the sinusoid figure is different for direct and phase-inverted signals. The ear clearly detects this, and you have to “un-adjust” the setting. It is extremely undesirable to level the sinusoid in phase-inverted cascades with deep OOS. It is necessary to eliminate the causes of asymmetry in these cascades in other circuitry ways, otherwise the phase-inverted cascade can introduce very noticeable “transistor” distortions, the level of which will be comparable to the distortions of the output stage of the UMZCH (!). This is how it happens that the phase inverter is the main asymmetry unit for any push-pull UMZCH (be it transistor, tube or combined UMZCH circuits), if, of course, the amplifying elements in the arms are pre-selected with similar parameters, otherwise there is no point in expecting anything from such good sound circuits.

The easiest to implement phase inversion circuits that work well are tube options. Their simpler “analogues” are field-effect transistors, which (only!) with a competent circuit design approach are quite capable of competing with tube amplifiers. And if audiophiles are not afraid of using matching transformers in output stages, where this “hardware” still “sounds,” then transformers can be used with a clear conscience in previous stages. I mean phase-inverted cascades, where the current amplitude (namely, this component has a detrimental effect on the hardware) is small, and the voltage amplitude reaches a value of only a few volts.

There is no doubt that any transformer is a kind of step back in circuitry in the age of gigahertz Pentiums. But there are several “buts” that are very appropriate to remember from time to time. First, a well-made transition or matching transformer will never introduce as much nonlinear distortion as several "wrong" amplifier stages can introduce a wide variety of distortions. Secondly, a transformer phase inverter really allows you to achieve real symmetry of antiphase signals, the signals from its windings are truly close to each other both in shape and amplitude. In addition, it is passive , and its characteristics do not depend on the supply voltages. And if your UMZCH is really symmetrical (in this case, we mean its input impedances), then the asymmetry of the UMZCH will already be determined by a greater spread in the parameters of the radio components in the UMZCH arms than by the phase-inverted cascade. Therefore, it is not recommended to use in such a UMZCH there are radioelements with tolerances of more than 5% (the only exceptions are the circuits of the current generator feeding the differential cascade). You should be aware that if the parameters of the transistors in the UMZCH arms vary by more than 20%, the accuracy of the resistors already loses its relevance. Conversely, when well-selected transistors are used, it makes sense to use resistors with a 1% tolerance. Of course, they can be selected using a good digital ohmmeter.

One of the most successful circuit designs of a phase inverter is shown in Fig. 2. Seemingly too simple, it still requires close attention to itself, since it has several “secrets”. The first one is the right choice transistors according to parameters. Transistors VT1 and VT2 should not have significant leakages between the electrodes (meaning gate-source junctions). In addition, transistors must have similar parameters, especially with regard to the initial drain current - specimens with I initial current are most suitable here. 30-70 mA. The supply voltages must be stabilized, although the stabilization coefficient of the power supply does not play a significant role, moreover, the negative voltage can be taken from the UMZCH stabilizer. To ensure that electrolytic capacitors introduce less distortion, they are shunted with non-electrolytic capacitors - type K73-17.

Let's take a little closer look at the manufacturing features of the main unit in this circuit - the phase-split (phase-inverted) transformer. Both the leakage inductance and the range of effectively reproduced frequencies, not to mention the level of various distortions, depend on the accuracy of its manufacture. So, the two main secrets of the technological process of manufacturing this transformer are as follows. The first is the need to abandon simple winding of windings. I give two options for winding this transformer that I used. The first is shown in Fig. 3, the second - in Fig. 4. The essence of this winding method is as follows. Each of the windings (I, II or III) consists of several windings containing strictly the same number of turns. Any error in the number of turns must be avoided, i.e. differences in turns between windings. Therefore, it was decided to wind the transformer using a long-proven method. According to Fig. 3, six wires are used (for example, PELSHO-0.25). The required length of the winding wire is calculated in advance (not always and not every radio amateur will have six coils of wire of the same diameter on hand), put the six wires together and wind all the windings at the same time. Next, you just need to find the taps of the required windings and connect them in pairs and in series. According to Fig. 4, nine conductors were used for this option. And yet, it is necessary to wind in such a way that the wires of one turn do not diverge in different directions far and wide from one another, but stick together in the common roll. Winding with separate wires is unacceptable, the transformer will literally “ring” in the entire range of audio frequencies, the leakage inductance will increase, and the distortion of the UMZCH will also increase due to the asymmetry of the signals at the transformer outputs.

Yes, and it is very easy to make a mistake with certain methods of winding symmetrical windings. And an error of several turns makes itself felt by the asymmetry of antiphase signals. If we continue frankly, a bass reflex transformer was manufactured (in a single type, copy) with... 15 cores. There was an experiment that was included in the collection of great-sounding UMZCH designs. Once again I would like to say that it is not transformers that are to blame for the poor performance of some circuits, but their designers. All over the world, the production of tube UMZCHs has expanded greatly; the vast majority of them contain isolation transformers (or rather, matching ones), without which the tube stage (a typical push-pull output stage circuit contains 2-4 tubes) is simply impossible to match with low-impedance speaker systems. There are, of course, also instances of “super tube” UMZCHs that do not have output transformers. Their place was taken by either powerful complementary pairs of field-effect transistors or... a battery of powerful tube triodes connected in parallel. But this topic is beyond the scope of this article. In our case, everything is much simpler. Transistor VT1 (Fig. 2) of the MOS type, connected in a circuit with a common drain (source follower) operates on a current generator (current source) made on transistor VT2. You should not use powerful field-effect transistors like KP904; they have increased input and pass-through capacitances, which cannot but affect the operation of this cascade.

Another stumbling block, a serious problem in creating a wideband transformer, awaits the designer when choosing a magnetic core. Here it is appropriate to add something to what can be found in the literature available to radio amateurs. Various design options for both radio amateurs and professionals suggest the use of different materials for the magnetic cores of transformers, which would not cause hassle both when purchasing them and when using them. The essence of the methods is this.

If your UMZCH will operate at frequencies above 1 kHz, then you can safely use ferrite cores. But preference should be given to specimens of magnetic cores with the highest magnetic permeability; cores from horizontal TV transformers work very well. Designers should be warned against using cores that have already been in operation for a long time. It is known that ferrite products lose their parameters with “age,” including the initial magnetic permeability; “unique” old age kills them no less than, for example, the magnets of long-term loudspeakers, which for some reason almost everyone is silent about.

Next about the cores - if UMZCH is used as a bass option, then you can safely use traditional W-shaped plate versions of magnetic cores. It must be emphasized that shielding of all such transformers was almost everywhere a necessity and a requirement. What can you do, you have to pay for everything. Usually it was sufficient to make a “cocoon” from ordinary roofing sheet 0.5 mm thick.

Toroidal cores also work well at low frequencies. By the way, their use simplifies the destruction of all kinds of interference from network transformers. Here the “reversibility” of the advantages of the toroidal core is preserved - in the network version it is distinguished by a small external radiation field, but in the input (signal) circuits it is insensitive to external fields. As for the broadband option (20 - 20,000 Hz), the most correct would be to use two different types of cores placed side by side in one window of the frame for winding the transformer windings. This eliminates blockage both at high frequencies (ferrite core works here) and at low frequencies (transformer steel works here). Additional improvement in sound reproduction in the region of 1-15 kHz is achieved by coating the steel core plates with varnish, as is done in tube UMZCHs. Moreover, each plate “works individually” as part of the core, which reduces all kinds of losses due to eddy currents. Nitrovarnish dries quickly; a thin layer is applied by simply dipping the plate into a container with varnish.

This technology for manufacturing a transformer in a bass reflex may seem too painstaking to many, but take my word for it - “the game is worth the candle,” because “what goes around comes around.” And as for the complexity, “low-tech”, we can say the following - in one day off it was possible to manufacture two such transformers without haste, and even solder their windings in the required order, which cannot be said about output transformers for tube UMZCHs.

Now a few words about the number of turns. The theory requires an increase in the inductance of the primary winding (I), with its increase the range of reproduced frequencies expands towards lower frequencies. In all designs, winding the windings before filling the frame was quite sufficient; the wire diameter was used 0.1 for 15 cores, 0.15 for 9 cores and 0.2 for the 6-core version. In the latter case, the existing PELSHO 0.25 was also used.

For the same. For those who cannot stand transformers, there is also a transformerless option - Fig. 5. This is the simplest one. but a completely sound version of the bass reflex cascade circuit, which was used not only in symmetrical UMZCH circuits, but also in powerful bridge UMZCHs. Simplicity is often deceiving, so I will limit myself to criticism of such schemes, but I dare say that it is quite difficult to symmetry the areas of sinusoids; it is often necessary to introduce additional bias and balancing circuits, and the quality of sound reproduction leaves much to be desired. Despite the phase, amplitude and frequency distortions introduced by transformers, they make it possible to achieve an almost linear frequency response in the audio frequency range, i.e. over the entire range of 20 Hz - 20,000 Hz. From 16 kHz and above, the capacitance of the windings can be affected, but the additionally increased cross-sectional area of the magnetic core allows us to partially avoid this problem. The rule is simple, similar to network transformers: by increasing the cross-sectional area of the magnetic circuit of the transformer core, for example, by two times. feel free to reduce the number of turns of the windings by half, etc.

Expand the range of effectively reproduced frequencies downwards, i.e. below 20 Hz, you can do it in the following way. Field-effect transistors (VT1, VT2 - Fig. 2) are used with large values of I initial. and increase the capacitance of capacitor C4 to 4700 uF. Electrolytic capacitors operate much cleaner if a direct polarizing voltage of several volts is applied to them. It is very convenient in this case to do the following. Install in the upper (according to the diagram) transistor VT1 an instance with an initial drain current greater than that of transistor VT2. You can do it even more “efficiently” by using a balancing resistor for transistor VT2; a fragment of a circuit with such a resistor is shown in Fig. 6. Initially, the slider of the tuning resistor R2" is in the lower (according to the diagram) position, moving its slider upward causes an increase in the drain current of transistor VT2, the potential on the positive plate of capacitor C4 becomes more negative. The reverse process occurs when the resistor R2 moves in the opposite direction. In this way, you can adjust the cascade according to the most suitable modes, especially when there are no transistors (VT1 and VT2) with close values of I initial. , but you have to install what you have at hand...

I dwelled in some detail on this seemingly very simple scheme. It is simple, but not primitive. It also has undeniable advantages over “all-passing” galvanically connected amplifier-phase inverter circuits. The first such advantage is the suppression of infra-low-frequency interference (for example, in electronic control units), the second is the “cut-off” of ultrasonic interference such as powerful radio stations, various ultrasonic installations, etc. And one more positive property of such a scheme should be especially emphasized. We are talking about the absence of any problems when connecting excellent symmetrical circuits with an asymmetrical input. It’s worth looking at Fig. 5, and it immediately becomes clear (if a person has dealt with this!) that the problem of potentials here simply has not been solved in any way. It is partially solved by replacing the electrolytic capacitor with a battery of parallel-connected non-electrolytic ones, as if a temporary delay in connecting the speakers will solve everything. The time delay in connecting the acoustic systems to the UMZCH really eliminates clicks and surges when turned on, but it cannot solve the issue of additional distortion due to different potentials and different output impedances of the phase inverter. This phase inverter amplifier circuit (Fig. 2) has been successfully used with various UMZCHs, including symmetrical tube ones.

Recently, in periodicals you can find UMZCH circuits based on powerful KP901 and KP904. But the authors do not mention that field-effect transistors should be rejected for leakage currents. If, for example, VT1 and VT2 (in the circuit of Fig. 2) it is clearly necessary to use high-quality copies, then in cascades with large amplitudes of voltages and currents, and most importantly - where the input resistance of the MOS transistor (its reduction) does not play a role, you can use even worse examples. Having reached maximum leakage values, MOS transistors are, as a rule, stable in the future and further deterioration of their parameters is no longer observed over time (in most cases).

The number of transistors with increased leakages in the gate circuit, for example, in one pack (standard - 50 pcs.) can range from 10 to 20 pcs. (or even more). Rejecting powerful transistors is not difficult - just assemble a kind of stand, for example, according to Fig. 6 and include a digital ammeter in the gate circuit (pointer instruments in this case are too sensitive to overloads and are inconvenient due to the need for repeated switching from range to range).

And now that the bass reflex has already been manufactured, you can proceed to the circuit in Fig. 1, i.e. return directly to UMZCH. The widely used connectors (sockets) SSh-3, SSh-5 and the like cannot be used at all, as many designers do and manufacturers did. The contact resistance of such a connection is significant (0.01 - 0.1 Ohm!) and also fluctuates depending on the flowing current (with increasing current, the resistance increases!). Therefore, you should use powerful connectors (for example, from old military radio equipment) with low contact resistance. The same applies to the relay contacts in the AC protection unit against the possible appearance of constant voltage at the output of the UMZCH. And there is no need to cover them (contact groups) with any feedback to reduce distortion. Take my word for it that by ear (subjective examination) they are practically inaudible (with sufficiently low contact resistances), which cannot be said about the “electronic” distortions introduced by all amplifier stages, capacitors and other components of the UMZCH, which certainly bring bright colors to the overall picture of sound reproduction. All kinds of distortion can be minimized by rational use of amplification stages (this is especially true for voltage amplifiers - the fewer of them, the better the quality of the amplified signal). In this UMZCH there is only one voltage amplification stage - transistor VT3 (left shoulder) and VT4 (right shoulder). The cascade on transistors VT6 and VT5 are just matching (current) emitter followers. Transistors VT3 and VT4 are selected with h21 e more than 50, VT6 and VT5 - more than 150. In this case, no problems will arise when operating the UMZCH at high powers. The negative feedback voltage for direct and alternating current is supplied to the bases of transistors VT6 and VT5 through resistors R24 and R23. The depth of this feedback is only about 20 dB, so there is no dynamic distortion in the UMZCH, but such feedback is quite sufficient to maintain the modes of the output transistors VT7 and VT8 within the required limits. UMZCH is quite resistant to HF self-excitation. The simplicity of the circuit allows it to be quickly disassembled, since the power supply (-40 V) of the driver and the final transistors (2 x 38 V) can be turned off independently. Full symmetry of the amplifier helps to reduce nonlinear distortions and reduce sensitivity to supply voltage ripples, as well as additional suppression of common-mode interference arriving at both inputs of the UMZCH. The disadvantage of the amplifier is the significant dependence of nonlinear distortions on h21 e of the transistors used, but if the transistors have h21 out = 70 W) is equal to 1.7 V (effective value).

Transistors VT1 and VT2 are used as a source (current generator) that powers the differential stage (driver). The value of this current 20...25 mA is set with trimming resistor R3 (470 Ohm). Since the quiescent current also depends on this current, for thermal stabilization of the latter, transistor VT1 is placed on the heat sink of one of the output stage transistors (VT7 or VT8). The increase in the temperature of the heat sink of the output transistor is accordingly transferred to the transistor VT1 located on this heat sink, and when the latter is heated, the negative potential at the base of the transistor VT2 decreases. This closes transistor VT2, the current through it decreases, which corresponds to a decrease in the quiescent current of the output transistors VT7 and VT8. In this way, the quiescent current of the output transistors is stabilized when their heat sinks are significantly heated. Despite the apparent simplicity of implementing such thermal stabilization, it is quite effective and there were no problems with the reliability of the UMZCH. It is very convenient to monitor the currents of differential transistors (VT3 and VT4) by the voltage drop across resistors R7 and R15 or R21 and R26. Trimmer resistor R11 is a balancing resistor, used to set the zero potential on the loudspeaker (at the output of the UMZCH).

The diagram of the loudspeaker protection unit (Fig. 7) is made according to the traditional scheme. Since the design of placing the UMZCH in separate housings was chosen, then Each UMZCH had its own acoustic system protection units. The speaker protection circuit is simple and reliable; this option has undergone long-term testing in many designs and has proven itself to be good and reliable, more than once “saving” the lives of expensive loudspeakers. Satisfactory operation of the circuit can be considered when relay K1 is activated when a constant voltage of 5 V is applied between points A and B. It is very easy to check this using an adjustable power supply (with variable output voltage). Different types of relays were used in different designs, and the voltage of the power supply of this unit also changed within 30-50 V (for higher values of this voltage, transistors VT1 and VT2 should be replaced with higher voltage units, for example KT503E, etc.)

Preference for use in the protection unit should be given to relays with the highest current groups of contacts, with a large area of contact contact surfaces. But relays RES-9 or RES-10 should not be used at all - at high output powers of the UMZCH, they begin to introduce their “unique” colors into the amplified signal. The AC protection unit is powered from a separate rectifier, and it is necessary to exclude any galvanic connections of this unit with the UMZCH, with the exception of only the output voltage sensors - points A and B are connected to the outputs of the UMZCH.

The drivers of both channels can be powered from one common voltage regulator. In this case, both channels of the UMZCH are combined into one housing, and the power supplies are assembled in another housing. Naturally, there is a wide field of choice for each specific case, for whom what is more suitable in design. The diagram of one of the stabilizer options for powering drivers is shown in Fig. 8. VT1 is assembled on transistor the current generator feeding transistor VT2, the required voltage at the output of the stabilizer is set by trimming resistor R6. It should be emphasized that the maximum output power of the UMZCH primarily depends on the voltage of this stabilizer. But increasing the voltage above 50 V is not recommended due to the possible failure of driver transistors VT3 and VT4. The total stabilization voltage of the zener diodes should be in the range of 27-33 V. The current through the zener diodes is selected by resistor R4. Resistor R1 is limiting (current) and prevents failure of the control transistor VT2. The latter is quite likely during the setup process, while increasing the driver power supply can disable the entire UMZCH. After installing the UMZCH, resistor R1 in the stabilizer can be closed with a piece of wire, or you don’t have to do this, since the drivers consume a current of only a little more than 50 mA - the influence of resistor R1 on the parameters of the stabilizer is negligible at low load currents.

With a block design, you will have to completely separate the power supplies of both UMZCHs, including the drivers. But in any case, to power the driver you need a separate rectifier with its own winding in the transformer. The rectifier circuit is shown in Fig.9. Each UMZCH channel uses its own power transformer. This design option has several advantages over the traditional use of a single transformer. The first thing that is possible is to reduce the height of the block as a whole, since the size (height) of the network transformer is significantly reduced with separate supply transformers for each UMZCH. Further, it is easier to wind, since the diameter of the winding wires can be reduced by 1.4 times without compromising the power of the UMZCH. In this regard, the network windings can be switched on in antiphase to reduce network interference (this greatly helps to compensate for the radiation of transformer fields, especially when other amplifier circuits are placed in the same housing with the UMZCH - tone blocks, volume control, etc.). Separating the supply circuits of the UMZCH output transistors makes it possible to increase the quality of the reproduced signal, especially at low frequencies (transient distortions in low-frequency channels are also reduced). To reduce the level of intermodulation distortion caused by mains power, electrostatic screens (one layer of wire wound turn to turn) are introduced into the transformers.

All UMZCH design options use toroidal magnetic cores for transformers. Winding was done manually using shuttles. We can also recommend a simplified version of the power supply design. For this, a factory-made LATR is used (a nine-amp copy is good). The primary winding, as the most difficult one in the winding process, is already ready, you just need to wind the screen winding and all the secondary windings and the transformer will work perfectly. Its window is spacious enough to accommodate the windings for both channels of the UMZCH. In addition, it is possible to power the drivers and phase inverter amplifiers from common stabilizers, “saving” in this case two windings. The disadvantage of such a transformer is its large height (except, of course, for the above circumstances).

Now about the details. You should not install low-frequency diodes (like D242 and the like) to power the UMZCH - distortion at high frequencies (from 10 kHz and above) will increase; in addition, ceramic capacitors were additionally introduced into the rectifier circuits to reduce intermodulation distortion caused by changes in the conductivity of the diodes in the moment of their commutation. This reduces the influence of the mains power on the UMZCH when it operates at high frequencies in the audio range. The situation is even better with quality when shunting electrolytic capacitors in high-current rectifiers (UMZCH output stages) with non-electrolytic ones. At the same time, both the first and second additions to the rectifier circuits were quite clearly perceived by a subjective examination - an auditory test of the operation of the UMZCH; its more natural operation was noted when reproducing several HF components of different frequencies.

About transistors. It is not worth replacing transistors VT3 and VT4 with copies that are worse in terms of frequency properties (KT814, for example), as the harmonic coefficient increases at least twice (in the HF section, and even more so). This is very noticeable by ear; mid frequencies are reproduced unnaturally. In order to simplify the design of the UMZCH, composite transistors of the KT827A series are used in the output stage. And although they, in principle, are quite reliable, they still need to be checked for the maximum withstandable (each instance has its own) collector-emitter voltage (meaning forward voltage Uke max. for a closed transistor). To do this, the base of the transistor is connected to the emitter through a 100 Ohm resistor and the voltage is applied, gradually increasing: to the collector - plus, to the emitter - minus. Instances that detect the flow of current (ammeter limit - 100 μA) for Uke = 100 V are not suitable for this design. They may work, but not for long... Instances without such “leaks” work reliably for years without creating any problems. The test bench diagram is shown in Fig. 10. Naturally, the parameters the KT827 series want to be the best, especially with regard to their frequency properties. Therefore, they were replaced with “composite” transistors assembled on KT940 and KT872. It is only necessary to select KT872 with the largest possible h21 e, since KT940 does not have I to max large enough. This equivalent works just fine across the entire audio range, and especially at high frequencies. The circuit diagram for connecting two transistors instead of one composite type KT827A is shown in Fig. 11. Transistor VT1 can be replaced with KT815G, and VT2 with almost any powerful one (P to > 50 W and with U e > 30).

Resistors used are types C2-13 (0.25 W), MLT. Capacitors types K73-17, K50-35, etc. Setting up a correctly (without errors) assembled UMZCH consists in setting the quiescent current of the UMZCH output stage transistors - VT7 and VT8 within 40-70 mA. It is very convenient to monitor the value of the quiescent current by the voltage drop across resistors R27 and R29. The quiescent current is set by resistor R3. A close-to-zero constant output voltage at the output of the UMZCH is set with a balancing resistor R11 (a potential difference of no more than 100 mV is achieved).

LITERATURE

Sukhov N.E. and others. High-quality sound reproduction technology - Kyiv, "Technique", 1985
Sukhov N.E. High fidelity UMZCH. - "Radio", 1989 - No. 6, No. 7.
Sukhov N.E. On the issue of assessing nonlinear distortions of UMZCH. - "Radio", No. 5. 1989.

A few words about installation errors:
In order to improve the readability of the circuits, let's consider a power amplifier with two pairs of final field-effect transistors and a power supply of ±45 V.
As a first mistake, let's try to "solder" the zener diodes VD1 and VD2 with the wrong polarity (correct connection is shown in Figure 11). The voltage map will take the form shown in Figure 12.

Figure 11 Pinout of zener diodes BZX84C15 (however, the pinout on diodes is the same).

Figure 12 Voltage map of a power amplifier with incorrect installation of zener diodes VD1 and VD2.

These zener diodes are needed to generate the supply voltage for the operational amplifier and were selected at 15 V solely because this voltage is optimal for this operational amplifier. The amplifier retains its performance without loss of quality even when using nearby ratings - 12 V, 13 V, 18 V (but not more than 18 V). If installed incorrectly, instead of the required supply voltage, the oprection amplifier receives only the drop voltage at the n-p junction of the zener diodes. The current is regulated normally, there is a small constant voltage at the output of the amplifier, and there is no output signal.
It is also possible that diodes VD3 and VD4 are installed incorrectly. In this case, the quiescent current is limited only by the values of resistors R5, R6 and can reach a critical value. There will be a signal at the output of the amplifier, but fairly rapid heating of the final transistors will definitely lead to their overheating and failure of the amplifier. The voltage and current map for this error are shown in Figures 13 and 14.

Figure 13 Amplifier voltage map with incorrect installation of thermal stabilization diodes.

Figure 14 Amplifier current map with incorrect installation of thermal stabilization diodes.

The next popular installation mistake may be incorrect installation of transistors of the penultimate stage (drivers). In this case, the voltage map of the amplifier takes on the form shown in Figure 15. In this case, the transistors of the terminal cascade are completely closed and there is no sign of sound at the amplifier output, and the DC voltage level is as close as possible to zero.

Figure 15 Voltage map for incorrect installation of transistors in the driver stage.

Next, the most dangerous mistake is that the transistors of the driver stage are mixed up, and the pinout is also mixed up, as a result of which what is applied to the terminals of transistors VT1 and VT2 is correct and they operate in emitter follower mode. In this case, the current through the final stage depends on the position of the trimming resistor slider and can be from 10 to 15 A, which in any case will cause an overload of the power supply and rapid heating of the final transistors. Figure 16 shows the currents at the middle position of the trimming resistor.

Figure 16 Current map when the transistors of the driver stage are installed incorrectly, the pinout is also confused.

It is unlikely that it will be possible to solder the output of the final field-effect transistors IRFP240 - IRFP9240 in reverse, but it is possible to swap them in places quite often. In this case, the diodes installed in transistors are in a difficult situation - the voltage applied to them has a polarity corresponding to their minimum resistance, which causes maximum consumption from the power supply and how quickly they burn out depends more on luck than on the laws of physics.
Fireworks on the board can happen for one more reason - 1.3 W zener diodes in a package the same as 1N4007 diodes are on sale, so before installing zener diodes on the board, if they are in a black case, you should take a closer look at the inscriptions on the case. When installing diodes instead of zener diodes, the supply voltage of the operational amplifier is limited only by the values of resistors R3 and R4 and the current consumption of the operational amplifier itself. In any case, the resulting voltage value is significantly greater than the maximum supply voltage for a given op-amp, which leads to its failure, sometimes with the shooting of part of the housing of the op-amp itself, and then a constant voltage may appear at its output, close to the supply voltage of the amplifier, which will lead to the appearance of a constant voltage at the output of the power amplifier itself. As a rule, the final cascade in this case remains operational.
And finally, a few words about the values of resistors R3 and R4, which depend on the supply voltage of the amplifier. 2.7 kOhm is the most universal, however, when powering the amplifier with a voltage of ±80 V (only to an 8 Ohm load), these resistors will dissipate about 1.5 W, so it must be replaced with a 5.6 kOhm or 6.2 kOhm resistor, which will reduce the generated thermal power to 0.7 W.

E K B BD135; BD137

H&S IRF240 - IRF9240

This amplifier deservedly gained its fans and began to acquire new versions. First of all, the bias voltage generation chain of the first transistor stage was changed. In addition, overload protection was introduced into the circuit.
As a result of modifications, the circuit diagram of a power amplifier with field-effect transistors at the output acquired the following form:

INCREASE

PCB options are shown in graphical format (needs to be scaled)

The appearance of the resulting modification of the power amplifier is shown in the photographs below:

All that remains is to add a fly in the ointment...
The fact is that the IRFP240 and IRFP9240 field-effect transistors used in the amplifier were discontinued by the developer International Rectifier (IR), which paid more attention to the quality of its products. The main problem with these transistors is that they were designed for use in power supplies, but turned out to be quite suitable for audio amplification equipment. International Rectifier's increased attention to the quality of manufactured components made it possible, without selecting transistors, to connect several transistors in parallel without worrying about differences in the characteristics of the transistors - the spread did not exceed 2%, which is quite acceptable.
Today, transistors IRFP240 and IRFP9240 are produced by Vishay Siliconix, which is not so sensitive to its products and the parameters of the transistors have become suitable only for power supplies - the spread in the “gain factor” of transistors of one batch exceeds 15%. This eliminates parallel connection without preliminary selection, and the number of tested transistors for selection 4 equally exceeds several dozen copies.
In this regard, before assembling this amplifier, you should first of all find out which brand of transistors you can get. If Vishay Siliconix is sold in your stores, then it is strongly recommended that you refuse to assemble this power amplifier - you risk spending quite a lot of money and not achieving anything.
However, the work on developing “VERSION 2” of this power amplifier and the lack of decent and inexpensive field-effect transistors for the output stage made us think a little about the future of this circuitry. As a result, “VERSION 3” was simulated, using instead of field-effect transistors IRFP240 - IRFP9240 from Vishay Siliconix a bipolar pair from TOSHIBA - 2SA1943 - 2SC5200, which today are still of quite decent quality.
The schematic diagram of the new version of the amplifier has incorporated improvements from “VERSION 2” and has undergone changes in the output stage, making it possible to abandon the use of field-effect transistors. The circuit diagram is shown below:

Schematic diagram using field-effect transistors as repeaters ENLARGE

In this version, field-effect transistors are retained, but they are used as voltage followers, which significantly relieves the load on the driver stage. A small positive connection has been introduced into the protection system to avoid excitation of the power amplifier at the protection operation limit.
The printed circuit board is in the process of development, approximately the results of real measurements and a working printed circuit board will appear at the end of November, but for now we can offer a THD measurement graph obtained by MICROCAP. You can read more about this program.

UMZCH with complementary field-effect transistors

We present to readers a version of a hundred-watt UMZCH with field-effect transistors. In this design, the housings of power transistors can be mounted on a common heat sink without insulating spacers, and this significantly improves heat transfer. As a second option for the power supply, a powerful pulse converter is proposed, which should have a fairly low level of self-interference.

The use of field-effect transistors (FETs) in UMZCHs has until recently been hampered by a meager range of complementary transistors, as well as their low operating voltage. The quality of sound reproduction through UMZCH on PT is often rated at the level of tube amplifiers and even higher due to the fact that, compared to amplifiers based on bipolar transistors, they create less nonlinear and intermodulation distortion, and also have a smoother increase in distortion during overloads. They are superior to tube amplifiers both in load damping and in the width of the operating audio frequency band. The cutoff frequency of such amplifiers without negative feedback is significantly higher than that of UMZCHs based on bipolar transistors, which has a beneficial effect on all types of distortion.

Nonlinear distortions in the UMZCH are mainly introduced by the output stage, and to reduce them, general OOS is usually used. Distortion in the input differential stage, used as a summator of signals from the source and the general OOS circuit, may be small, but it is impossible to reduce them using the general OOS

The overload capacity of the differential cascade using field-effect transistors is approximately 100...200 times higher than with bipolar transistors.

The use of field-effect transistors in the output stage of the UMZCH makes it possible to abandon traditional two- and three-stage Darlington repeaters with their inherent disadvantages.

Good results are obtained by using field-effect transistors with a metal-dielectric-semiconductor (MDS) structure in the output stage. Due to the fact that the current in the output circuit is controlled by the input voltage (similar to electric vacuum devices), at high currents the performance of the cascade on field-effect MOS transistors in the switching mode is quite high (τ = 50 ns). Such cascades have good transfer properties at high frequencies and have a temperature self-stabilization effect.

The advantages of field-effect transistors include:

low control power in static and dynamic modes;
absence of thermal breakdown and low susceptibility to secondary breakdown;
thermal stabilization of the drain current, providing the possibility of parallel connection of transistors;
the transfer characteristic is close to linear or quadratic;
high performance in switching mode, thereby reducing dynamic losses;
absence of the phenomenon of accumulation of excess carriers in the structure;
low noise level,
small dimensions and weight, long service life.

But besides the advantages, these devices also have disadvantages:

failure due to electrical overvoltage;
Thermal distortion may occur at low frequencies (below 100 Hz). At these frequencies, the signal changes so slowly that in one half-cycle the temperature of the crystal has time to change and, consequently, the threshold voltage and transconductance of the transistors change.

The last noted disadvantage limits the output power, especially at low supply voltages; The way out is to switch on transistors in parallel and introduce OOS.

It should be noted that recently foreign companies (for example, Exicon, etc.) have developed many field-effect transistors suitable for audio equipment: EC-10N20, 2SK133-2SK135, 2SK175, 2SK176 with a n-type channel; EC-10P20, 2SJ48-2SJ50, 2SJ55, 2SJ56 with a p-type channel. Such transistors are distinguished by a weak dependence of the transconductance (forward transfer admittance) on the drain current and smoothed output I-V characteristics

The parameters of some field-effect transistors, including those produced by the Minsk Production Association "Integral", are given in Table. 1.

Most transistor transformerless UMZCHs are made using a half-bridge circuit. In this case, the load is connected to the diagonal of the bridge formed by two power supplies and two output transistors of the amplifier (Fig. 1).

When there were no complementary transistors, the output stage of the UMZCH was performed mainly on transistors of the same structure with a load and a power source connected to a common wire (Fig. 1, a). Two possible options for controlling the output transistors are presented in Fig. 2.

In the first of them (Fig. 2,a), the control of the lower arm of the output stage is in more favorable conditions. Since the change in supply voltage is small, the Miller effect (dynamic input capacitance) and the Earley effect (dependence of the collector current on the emitter-collector voltage) practically do not appear. The control circuit of the upper arm is connected here in series with the load itself, therefore, without taking additional measures (for example, cascode switching on of devices), these effects manifest themselves to a significant extent. A number of successful UMZCHs have been developed based on this principle.

According to the second option (Fig. 2.6 - MIS transistors are more consistent with this structure), a number of UMZCHs were also developed, for example. However, even in such cascades it is difficult to ensure symmetry of control of the output transistors, even with the use of current generators. Another example of balancing by input resistance is the implementation of amplifier arms in a quasi-complementary circuit or the use of complementary transistors (see Fig. 1, b) c.

The desire to balance the arms of the output stage of amplifiers made on transistors of the same conductivity led to the development of amplifiers with an ungrounded load according to the circuit in Fig. 1,g. However, even here it is not possible to achieve complete symmetry of the previous cascades. The negative feedback circuits from each arm of the output stage are unequal; The OOS circuits of these stages control the voltage on the load in relation to the output voltage of the opposite side. In addition, such a circuit solution requires isolated power supplies. Due to these shortcomings, it has not found widespread use.

With the advent of complementary bipolar and field-effect transistors, the output stages of the UMZCH are mainly built according to the circuits in Fig. 1, b, c. However, even in these options, it is necessary to use high-voltage devices to drive the output stage. Transistors of the pre-output stage operate with a high voltage gain, and therefore are subject to the Miller and Earley effects and, without general feedback, introduce significant distortion, which requires high dynamic characteristics from them. Powering the preliminary stages with increased voltage also reduces the efficiency of the amplifier.

If in Fig. 1, b, c move the connection point with the common wire to the opposite arm of the bridge diagonal, we get the options in Fig. 1,d and 1,f, respectively. In the cascade structure according to the diagram in Fig. 1,e automatically solves the problem of isolating the output transistors from the housing. Amplifiers made according to such circuits are free from a number of the listed disadvantages.

Amplifier circuit design features

We offer radio amateurs an inverting UMZCH (Fig. 3), corresponding to the block diagram of the output stage in Fig. 1,e.

(click to enlarge)

The input differential stage is made using field-effect transistors (VT1, VT2 and DA1) in a symmetrical circuit. Their advantages in a differential cascade are well known: high linearity and overload capacity, low noise. The use of field-effect transistors significantly simplified this cascade, since there was no need for current generators. To increase the gain with the feedback loop open, the signal is removed from both arms of the differential stage, and an emitter follower on transistors VT3, VT4 is installed in front of the subsequent voltage amplifier.

The second stage is made using transistors VT5-VT10 using a combined cascode circuit with tracking power. This power supply of the OE cascade neutralizes the input dynamic capacitance in the transistor and the dependence of the collector current on the emitter-collector voltage. The output stage of this stage uses high-frequency BSIT transistors, which, compared to bipolar transistors (KP959 versus KT940), have twice the cutoff frequency and four times the drain (collector) capacitance.

The use of an output stage powered by separate isolated sources made it possible to dispense with a low-voltage supply (9 V) for the preamplifier.

The output stage is made of powerful MOS transistors, and their drain terminals (and the heat-dissipating flanges of the housings) are connected to a common wire, which simplifies the design and assembly of the amplifier.

Powerful MOS transistors, unlike bipolar ones, have a smaller spread of parameters, which makes their parallel connection easier. The main spread of currents between devices arises due to the inequality of threshold voltages and the spread of input capacitances. The introduction of additional resistors with a resistance of 50-200 Ohms in the gate circuit ensures almost complete equalization of the on and off delays and eliminates the spread of currents during switching.

All amplifier stages are covered by local and general OOS.

Main technical characteristics

With open feedback (R6 replaced by 22 MOhm, C4 excluded)
Cutoff frequency, kHz......300
Voltage gain, dB......43
Harmonic coefficient in AB mode, %, no more......2

With OOS enabled

Output power, W at 4 Ohm load......100
at a load of 8 Ohms......60
Reproducible frequency range, Hz......4...300000
Harmonic coefficient, %, no more......0.2
Rated input voltage, V......2
Quiescent current of the output stage, A......0.15
Input resistance, kOhm.....24

Due to the fact that the cutoff frequency of the open-loop amplifier is relatively high, the feedback depth and harmonic distortion are virtually constant across the entire frequency range.

From below, the operating frequency band of the UMZCH is limited by the capacitance of capacitor C1, from above - by C4 (with a capacitance of 1.5 pF, the cutoff frequency is 450 kHz).

Construction and details

The amplifier is made on a board made of double-sided foil fiberglass (Fig. 4).

The board on the side where the elements are installed is filled as much as possible with foil connected to a common wire. Transistors VT8, VT9 are equipped with small plate heat sinks in the form of a “flag”. Pistons are installed in the holes for the drain terminals of powerful field-effect transistors; The drain terminals of transistors VT11, VT14 are connected to the common wire on the foil side (marked with crosses in the figure).

Pistons are installed in holes 5-7 of the board for connecting the leads of the network transformer and the holes for jumpers. Resistors R19, R20, R22, R23 are made of manganin wire with a diameter of 0.5 and a length of 150 mm. To suppress inductance, the wire is folded in half and, folded (bifilar), wound on a mandrel with a diameter of 4 mm.

Inductor L1 is wound with PEV-2 wire 0.8 turn to turn over the entire surface of a 2 W resistor (MLT or similar).

Capacitors C1, C5, C10, C11 - K73-17, with C10 and C11 soldered from the printed circuit side to the terminals of capacitors C8 and C9. Capacitors C2, C3 - oxide K50-35; capacitor C4 - K10-62 or KD-2; C12 - K10-17 or K73-17.

Field-effect transistors with an n-type channel (VT1, VT2) must be selected with approximately the same initial drain current as the transistors in the DA1 assembly. In terms of cutoff voltage, they should not differ by more than 20%. Microassembly DA1 K504NTZB can be replaced with K504NT4B. It is possible to use a selected pair of KP10ZL transistors (also with indices G, M, D); KP307V - KP307B (also A, E), KP302A or transistor assembly KPS315A, KPS315B (in this case the board will have to be redesigned).

In positions VT8, VT9, you can also use complementary transistors of the KT851, KT850 series, as well as KT814G, KT815G (with a cutoff frequency of 40 MHz) from the Minsk Association "Integral".

In addition to those indicated in the table, you can use, for example, the following pairs of MIS transistors: IRF530 and IRF9530; 2SK216 and 2SJ79; 2SK133-2SK135 and 2SJ48-2SJ50; 2SK175-2SK176 and 2SJ55-2SJ56.

For the stereo version, power is supplied to each amplifier from a separate transformer, preferably with a ring or rod (PL) magnetic circuit, with a power of 180...200 W. A layer of shielding winding with PEV-2 0.5 wire is placed between the primary and secondary windings; one of its terminals is connected to the common wire. The leads of the secondary windings are connected to the amplifier board with a shielded wire, and the screen is connected to the common wire of the board. On one of the network transformers the windings for the rectifiers of the preamplifiers are placed. Voltage stabilizers are made on IL7809AC (+9 V), IL7909AC (-9 V) microcircuits - not shown in the diagram. To supply 2x9 V power to the board, the ONP-KG-26-3 (XS1) connector is used.

When setting up, the optimal current of the differential stage is set by adjusting resistor R3 to minimize distortion at maximum power (approximately in the middle of the working section). Resistors R4, R5 are designed for a current of about 2...3 mA in each arm with an initial drain current of about 4...6 mA. With a lower initial drain current, the resistance of these resistors must be proportionally increased.

The quiescent current of the output transistors in the range of 120... 150 mA is set by trimming resistor R3, and, if necessary, by selecting resistors R13, R14.

Impulse power block

For those radio amateurs who have difficulty purchasing and winding large network transformers, a switching power supply is offered for the output stages of the UMZCH. In this case, the pre-amplifier can be powered from a low-power stabilized power supply.

A pulse power supply (its circuit is shown in Fig. 5) is an unregulated self-oscillating half-bridge inverter. The use of proportional current control of the inverter transistors in combination with a saturable switching transformer allows the active transistor to be automatically removed from saturation at the time of switching. This reduces the charge dissipation time in the base and eliminates through current, and also reduces power losses in control circuits, increasing the reliability and efficiency of the inverter.

UPS Specifications

Output power, W, no more......360
Output voltage......2x40
Efficiency, %, not less......95
Conversion frequency, kHz......25

An interference suppression filter L1C1C2 is installed at the input of the mains rectifier. Resistor R1 limits the surge current charging capacitor C3. There is a jumper X1 in series with the resistor on the board, instead of which you can turn on a choke to improve filtering and increase the “hardness” of the output load characteristic.

The inverter has two positive feedback circuits: the first - for voltage (using windings II in transformer T1 and III - in T2); the second - by current (with a current transformer: turn 2-3 and windings 1-2, 4-5 of transformer T2).

The triggering device is made on a unijunction transistor VT3. After the converter starts, it is turned off due to the presence of the VD15 diode, since the time constant of the R6C8 circuit is significantly longer than the conversion period.

The peculiarity of the inverter is that when low-voltage rectifiers operate on large filter capacitances, it needs a smooth start. The smooth start of the unit is facilitated by chokes L2 and L3 and, to some extent, by resistor R1.

The power supply is made on a printed circuit board made of one-sided foil fiberglass 2 mm thick. The board drawing is shown in Fig. 6.

(click to enlarge)

Winding data of transformers and information about magnetic cores are given in table. 2. All windings are made with PEV-2 wire.

Before winding transformers, the sharp edges of the rings must be dulled with sandpaper or a block and wrapped with varnished cloth (for T1 - rings folded together in three layers). If this pre-treatment is not done, then it is possible that the varnished fabric will be pressed through and the turns of the wire will be shorted to the magnetic circuit. As a result, the no-load current will sharply increase and the transformer will heat up. Between windings 1-2, 5-6-7 and 8-9-10, shielding windings are wound with PEV-2 0.31 wire in one layer turn to turn, one end of which (E1, E2) is connected to the common wire of the UMZCH.

Winding 2-3 of the T2 transformer is a coil of wire with a diameter of 1 mm on top of winding 6-7, soldered at the ends into a printed circuit board.

Chokes L2 and L3 are made on BZO armored magnetic cores made of 2000NM ferrite. The windings of the chokes are wound in two wires until the frame is filled with PEV-2 0.8 wire. Considering that the chokes operate with direct current bias, it is necessary to insert gaskets made of non-magnetic material 0.3 mm thick between the cups.

Choke L1 is type D13-20, it can also be made on an armored magnetic core B30 similar to chokes L2, L3, but without a gasket, by winding the windings in two MGTF-0.14 wires until the frame is filled.

Transistors VT1 and VT2 are mounted on heat sinks made of ribbed aluminum profile with dimensions 55x50x15 mm through insulating gaskets. Instead of those indicated in the diagram, you can use KT8126A transistors from the Minsk Integral Production Association, as well as MJE13007. Between the power supply outputs +40 V, -40 V and “their” midpoint (ST1 and ST2), additional oxide capacitors K50-6 (not shown in the diagram) with a capacity of 2000 μF at 50 V are connected. These four capacitors are installed on a textolite plate with dimensions 140x100 mm, fixed with screws on the heat sinks of powerful transistors.

Capacitors C1, C2 - K73-17 for voltage 630 V, C3 - oxide K50-35B for 350 V, C4, C7 - K73-17 for 250 V, C5, C6 - K73-17 for 400 V, C8 - K10-17 .

The pulse power supply is connected to the PA board in close proximity to the terminals of capacitors C6-C11. In this case, the diode bridge VD5-VD8 is not mounted on the PA board.

To delay the connection of speaker systems to the UMZCH for the duration of the attenuation of transient processes that occur during power-on, and to turn off the speakers when a direct voltage of any polarity appears at the output of the amplifier, you can use a simple or more complex protective device.

Literature

Khlupnov A. Amateur low frequency amplifiers. -M.: Energy, 1976, p. 22.
Akulinichev I. Low-frequency amplifier with common-mode stabilizer. - Radio, 1980, No. Z.s.47.
Garevskikh I. Broadband power amplifier. - Radio, 1979, No. 6. p. 43.
Kolosov V. Modern amateur tape recorder. - M.: Energy, 1974.
Borisov S. MOS transistors in low-frequency amplifiers. - Radio. 1983, No. 11, p. 36-39.
Dorofeev M. Mode B in AF power amplifiers. - Radio, 1991, No. 3, p. 53.
Syritso A. Powerful bass amplifier. - Radio, 1978. No. 8, p. 45-47.
Syritso A. Power amplifier based on integrated op-amps. - Radio, 1984, No. 8, p. 35-37.
Yakimenko N. Field-effect transistors in the bridge UMZCH. - Radio. 1986, no. 9, p. 38, 39.
Vinogradov V. AC protection device. - Radio, 1987, No. 8. p. thirty.

To date, many versions of UMZCH with output stages based on field-effect transistors have been developed. The attractiveness of these transistors as powerful amplifying devices has been repeatedly noted by various authors. At audio frequencies, field-effect transistors (FETs) act as current amplifiers, so the load on the pre-stages is negligible and the insulated gate FET output stage can be directly connected to the pre-amplifier stage operating in class A linear mode.
When using powerful PTs, the nature of nonlinear distortions changes (fewer higher harmonics than when using bipolar transistors), dynamic distortions are reduced, and the level of intermodulation distortions is significantly lower. However, due to a lower transconductance than that of bipolar transistors, the nonlinear distortion of the source follower turns out to be large, since the transconductance depends on the level of the input signal.
The output stage on powerful PTs, where they can withstand a short circuit in the load circuit, has the property of thermal stabilization. Some disadvantage of such a cascade is the lower utilization of the supply voltage, and therefore it is necessary to use a more efficient heat sink.
The main advantages of powerful PTs include the low order of nonlinearity of their pass-through characteristics, which brings the sound features of PT amplifiers and tube amplifiers closer together, as well as a high power gain for signals in the audio frequency range.
Among the latest publications in the journal about UMZCH with powerful PTs, articles can be noted. The undoubted advantage of the amplifier is the low level of distortion, but the disadvantage is low power (15 W). The amplifier has more power, sufficient for residential use, and an acceptable level of distortion, but appears to be relatively complex to manufacture and configure. Hereinafter we are talking about UMZCHs intended for use with household speakers with a power of up to 100 W.
UMZCH parameters, focused on compliance with international IEC recommendations, determine the minimum requirements for hi-fi equipment. They are quite justified both from the psychophysiological side of human perception of distortion, and from the actually achievable distortion of audio signals in acoustic systems (AS), on which the UMZCH actually works.
In accordance with the requirements of IEC 581-7 for hi-fi speakers, the total harmonic distortion factor should not exceed 2% in the frequency range 250 ... 1000 Hz and 1% in the range above 2 kHz at a sound pressure level of 90 dB at a distance of 1 m. The characteristic sensitivity of household speakers is 86 dB/W/m, this corresponds to an UMZCH output power of only 2.5 W. Taking into account the peak factor of music programs, taken equal to three (as for Gaussian noise), the output power of the UMZCH should be about 20 W. In a stereophonic system, the sound pressure at the low frequencies approximately doubles, which allows the listener to move away from the speaker by 2 m. At a distance of 3 m, the power of a stereo amplifier of 2x45 W is quite sufficient.
It has been repeatedly noted that distortions in UMZCHs on field-effect transistors are caused mainly by the second and third harmonics (as in working speakers). If we assume that the causes of nonlinear distortions in the speakers and the UMZCH are independent, then the resulting harmonic coefficient for sound pressure is determined as the square root of the sum of the squares of the harmonic coefficients of the UMZCH and the speaker. In this case, if the total harmonic distortion coefficient in the UMZCH is three times lower than the distortion in the speakers (i.e. does not exceed 0.3%), then it can be neglected.
The range of effectively reproduced frequencies of the UMZCH should be no longer audible to humans - 20...20,000 Hz. As for the rate of rise of the output voltage of the UMZCH, in accordance with the results obtained in the author’s work, a speed of 7 V/μs is sufficient for a power of 50 W when operating at a load of 4 Ohms and 10 V/μs when operating at a load of 8 Ohms.
The basis for the proposed UMZCH was an amplifier in which a high-speed op-amp with tracking power was used to “drive” the output stage in the form of composite repeaters on bipolar transistors. Tracking power was also used for the output stage bias circuit.

The following changes have been made to the amplifier: the output stage based on complementary pairs of bipolar transistors has been replaced by a cascade with a quasi-complementary structure using inexpensive IRFZ44 insulated gate PTs and the depth of the total SOS is limited to 18 dB. The circuit diagram of the amplifier is shown in Fig. 1.

The KR544UD2A op-amp with high input impedance and increased speed was used as a pre-amplifier. It contains an input differential stage on a PT with a p-n junction and an output push-pull voltage follower. Internal frequency equalization elements provide stability in various feedback modes, including voltage follower.
The input signal comes through the low-pass filter RnC 1 with a cutoff frequency of about 70 kHz (here the internal resistance of the signal source = 22 kOhm). which is used to limit the spectrum of the signal entering the power amplifier input. Circuit R1C1 ensures the stability of the UMZCH when the value of RM changes from zero to infinity. To the non-inverting input of op-amp DA1, the signal passes through a high-pass filter built on elements C2, R2 with a cutoff frequency of 0.7 Hz, which serves to separate the signal from the constant component. Local OOS for the operational amplifier is made on elements R5, R3, SZ and provides a gain of 43 dB.
The voltage stabilizer for the bipolar supply of op-amp DA1 is made on elements R4, C4, VDI and R6, Sat. VD2 respectively. The stabilization voltage is chosen to be 16 V. Resistor R8 together with resistors R4, R6 form a divider of the output voltage of the UMZCH to supply “tracking” power to the op-amp, the swing of which should not exceed the limit values of the common-mode input voltage of the op-amp, i.e. +/-10 V "Tracking" power supply allows you to significantly increase the range of the op-amp's output signal.
As is known, for the operation of a field-effect transistor with an insulated gate, in contrast to a bipolar one, a bias of about 4 V is required. For this, in the circuit shown in Fig. 1, for transistor VT3, a signal level shift circuit is used on elements R10, R11 and УУЗ.У04 to 4.5 V. The signal from the output of the op-amp through the circuit VD3VD4C8 and resistor R15 is supplied to the gate of transistor VT3, the constant voltage on which relative to the common wire is +4, 5 V.
The electronic analogue of the zener diode on elements VT1, VD5, VD6, Rl2o6ecne4H shifts the voltage by -1.5 V relative to the op-amp output to ensure the required operating mode of transistor VT2. The signal from the output of the op-amp through circuit VT1C9 also goes to the base of transistor VT2, which is connected according to a common emitter circuit, which inverts the signal.
On R17 elements. VD7, C12, R18 an adjustable level shift circuit is assembled, which allows you to set the required bias for transistor VT4 and thereby set the quiescent current of the final stage. The capacitor SY provides “tracking power” to the level shift circuit by supplying the UMZCH output voltage to the connection point of resistors R10, R11 to stabilize the current in this circuit. The connection of transistors VT2 and VT4 forms a virtual field-effect transistor with a p-type channel. i.e., a quasi-complementary pair is formed with the output transistor VT3 (with an n-type channel).
Circuit C11R16 increases the stability of the amplifier in the ultrasonic frequency range. Ceramic capacitors C13. C14. installed in close proximity to the output transistors serve the same purpose. Protection of the UMZCH from overloads during short circuits in the load is provided by fuses FU1-FU3. since IRFZ44 field-effect transistors have a maximum drain current of 42 A and can withstand overloads until the fuses blow.
To reduce the DC voltage at the output of the UMZCH, as well as to reduce nonlinear distortions, a general OOS has been introduced on elements R7, C7. R3, NW. The AC OOS depth is limited to 18.8 dB, which stabilizes the harmonic distortion coefficient in the audio frequency range. For direct current, the op-amp, together with the output transistors, operates in the voltage follower mode, providing a constant component of the UMZCH output voltage of no more than a few millivolts.

– The neighbor stopped knocking on the radiator. I turned the music up so I couldn't hear him.
(From audiophile folklore).

The epigraph is ironic, but the audiophile is not necessarily “sick in the head” with the face of Josh Ernest at a briefing on relations with the Russian Federation, who is “thrilled” because his neighbors are “happy.” Someone wants to listen to serious music at home as in the hall. For this purpose, the quality of the equipment is needed, which among lovers of decibel volume as such simply does not fit where sane people have a mind, but for the latter it goes beyond reason from the prices of suitable amplifiers (UMZCH, audio frequency power amplifier). And someone along the way has a desire to join useful and exciting areas of activity - sound reproduction technology and electronics in general. Which in the age of digital technology are inextricably linked and can become a highly profitable and prestigious profession. The optimal first step in this matter in all respects is to make an amplifier with your own hands: It is UMZCH that allows, with initial training on the basis of school physics on the same table, to go from the simplest designs for half an evening (which, nevertheless, “sing” well) to the most complex units, through which a good rock band will play with pleasure. The purpose of this publication is highlight the first stages of this path for beginners and, perhaps, convey something new to those with experience.

Protozoa

So, first, let's try to make an audio amplifier that just works. In order to thoroughly delve into sound engineering, you will have to gradually master quite a lot of theoretical material and not forget to enrich your knowledge base as you progress. But any “cleverness” is easier to assimilate when you see and feel how it works “in hardware.” In this article further, too, we will not do without theory - about what you need to know at first and what can be explained without formulas and graphs. In the meantime, it will be enough to know how to use a multitester.

Note: If you haven’t soldered electronics yet, keep in mind that its components cannot be overheated! Soldering iron - up to 40 W (preferably 25 W), maximum allowable soldering time without interruption - 10 s. The soldered pin for the heat sink is held 0.5-3 cm from the soldering point on the side of the device body with medical tweezers. Acid and other active fluxes cannot be used! Solder - POS-61.

On the left in Fig.- the simplest UMZCH, “which just works.” It can be assembled using both germanium and silicon transistors.

On this baby it is convenient to learn the basics of setting up an UMZCH with direct connections between cascades that give the clearest sound:

Before turning on the power for the first time, turn off the load (speaker);
Instead of R1, we solder a chain of a constant resistor of 33 kOhm and a variable resistor (potentiometer) of 270 kOhm, i.e. first note four times less, and the second approx. twice the denomination compared to the original according to the scheme;
We supply power and, by rotating the potentiometer, at the point marked with a cross, we set the indicated collector current VT1;
We remove the power, unsolder the temporary resistors and measure their total resistance;
As R1 we set a resistor with a value from the standard series closest to the measured one;
We replace R3 with a constant 470 Ohm chain + 3.3 kOhm potentiometer;
Same as according to paragraphs. 3-5, V. And we set the voltage equal to half the supply voltage.

Point a, from where the signal is removed to the load, is the so-called. midpoint of the amplifier. In UMZCH with unipolar power supply, it is set to half its value, and in UMZCH with bipolar power supply - zero relative to the common wire. This is called adjusting the amplifier balance. In unipolar UMZCHs with capacitive decoupling of the load, it is not necessary to turn it off during setup, but it is better to get used to doing this reflexively: an unbalanced 2-polar amplifier with a connected load can burn out its own powerful and expensive output transistors, or even a “new, good” and very expensive powerful speaker.

Note: components that require selection when setting up the device in the layout are indicated on the diagrams either with an asterisk (*) or an apostrophe (‘).

In the center of the same fig.- a simple UMZCH on transistors, already developing power up to 4-6 W at a load of 4 ohms. Although it works like the previous one, in the so-called. class AB1, not intended for Hi-Fi sound, but if you replace a pair of these class D amplifiers (see below) in cheap Chinese computer speakers, their sound improves noticeably. Here we learn another trick: powerful output transistors need to be placed on radiators. Components that require additional cooling are outlined in dotted lines in the diagrams; however, not always; sometimes - indicating the required dissipative area of the heat sink. Setting up this UMZCH is balancing using R2.

On the right in Fig.- not yet a 350 W monster (as was shown at the beginning of the article), but already quite a solid beast: a simple amplifier with 100 W transistors. You can listen to music through it, but not Hi-Fi, operating class is AB2. However, it is quite suitable for scoring a picnic area or an outdoor meeting, a school assembly hall or a small shopping hall. An amateur rock band, having such a UMZCH per instrument, can perform successfully.

There are 2 more tricks in this UMZCH: firstly, in very powerful amplifiers, the drive stage of the powerful output also needs to be cooled, so VT3 is placed on a radiator of 100 kW or more. see. For output VT4 and VT5 radiators from 400 sq.m. are needed. see. Secondly, UMZCHs with bipolar power supply are not balanced at all without load. First one or the other output transistor goes into cutoff, and the associated one goes into saturation. Then, at full supply voltage, current surges during balancing can damage the output transistors. Therefore, for balancing (R6, guessed it?), the amplifier is powered from +/–24 V, and instead of a load, a wirewound resistor of 100...200 Ohms is switched on. By the way, the squiggles in some resistors in the diagram are Roman numerals, indicating their required heat dissipation power.

Note: A power source for this UMZCH needs a power of 600 W or more. Anti-aliasing filter capacitors - from 6800 µF at 160 V. In parallel with the electrolytic capacitors of the IP, 0.01 µF ceramic capacitors are included to prevent self-excitation at ultrasonic frequencies, which can instantly burn out the output transistors.

On the field workers

On the trail. rice. - another option for a fairly powerful UMZCH (30 W, and with a supply voltage of 35 V - 60 W) on powerful field-effect transistors:

The sound from it already meets the requirements for entry-level Hi-Fi (if, of course, the UMZCH works on the corresponding acoustic systems, speakers). Powerful field drivers do not require a lot of power to drive, so there is no pre-power cascade. Even more powerful field-effect transistors do not burn out the speakers in the event of any malfunction - they themselves burn out faster. Also unpleasant, but still cheaper than replacing an expensive loudspeaker bass head (GB). This UMZCH does not require balancing or adjustment in general. As a design for beginners, it has only one drawback: powerful field-effect transistors are much more expensive than bipolar transistors for an amplifier with the same parameters. Requirements for individual entrepreneurs are similar to previous ones. case, but its power is needed from 450 W. Radiators – from 200 sq. cm.

Note: there is no need to build powerful UMZCHs on field-effect transistors for switching power supplies, for example. computer When trying to “drive” them into the active mode required for UMZCH, they either simply burn out, or the sound is weak and “no quality at all.” The same applies to powerful high-voltage bipolar transistors, for example. from line scan of old TVs.

Straight up

If you have already taken the first steps, then it is quite natural to want to build Hi-Fi class UMZCH, without going too deep into the theoretical jungle. To do this, you will have to expand your instrumentation - you need an oscilloscope, an audio frequency generator (AFG) and an AC millivoltmeter with the ability to measure the DC component. It is better to take as a prototype for repetition the E. Gumeli UMZCH, described in detail in Radio No. 1, 1989. To build it, you will need a few inexpensive available components, but the quality meets very high requirements: power up to 60 W, band 20-20,000 Hz, frequency response unevenness 2 dB, nonlinear distortion factor (THD) 0.01%, self-noise level –86 dB. However, setting up the Gumeli amplifier is quite difficult; if you can handle it, you can take on any other. However, some of the currently known circumstances greatly simplify the establishment of this UMZCH, see below. Bearing in mind this and the fact that not everyone is able to get into the Radio archives, it would be appropriate to repeat the main points.

Schemes of a simple high-quality UMZCH

The Gumeli UMZCH circuits and specifications for them are shown in the illustration. Radiators of output transistors – from 250 sq. see for UMZCH in Fig. 1 and from 150 sq. see for option according to fig. 3 (original numbering). Transistors of the pre-output stage (KT814/KT815) are installed on radiators bent from 75x35 mm aluminum plates with a thickness of 3 mm. There is no need to replace KT814/KT815 with KT626/KT961; the sound does not noticeably improve, but setup becomes seriously difficult.

This UMZCH is very critical to power supply, installation topology and general, so it needs to be installed in a structurally complete form and only with a standard power source. When trying to power it from a stabilized power supply, the output transistors burn out immediately. Therefore, in Fig. Drawings of original printed circuit boards and setup instructions are provided. We can add to them that, firstly, if “excitement” is noticeable when you first turn it on, they fight it by changing the inductance L1. Secondly, the leads of parts installed on boards should be no longer than 10 mm. Thirdly, it is extremely undesirable to change the installation topology, but if it is really necessary, there must be a frame shield on the side of the conductors (ground loop, highlighted in color in the figure), and the power supply paths must pass outside it.

Note: breaks in the tracks to which the bases of powerful transistors are connected - technological, for adjustment, after which they are sealed with drops of solder.

Setting up this UMZCH is greatly simplified, and the risk of encountering “excitement” during use is reduced to zero if:

Minimize interconnect installation by placing the boards on radiators of powerful transistors.
Completely abandon the connectors inside, performing all installation only by soldering. Then there will be no need for R12, R13 in a powerful version or R10 R11 in a less powerful version (they are dotted in the diagrams).
Use oxygen-free copper audio wires of minimum length for internal installation.

If these conditions are met, there are no problems with excitation, and setting up the UMZCH comes down to the routine procedure described in Fig.

Wires for sound

Audio wires are not an idle invention. The need for their use at present is undeniable. In copper with an admixture of oxygen, a thin oxide film is formed on the faces of metal crystallites. Metal oxides are semiconductors and if the current in the wire is weak without a constant component, its shape is distorted. In theory, distortions on myriads of crystallites should compensate each other, but very little (apparently due to quantum uncertainties) remains. Sufficient to be noticed by discerning listeners against the background of the purest sound of modern UMZCH.

Manufacturers and traders shamelessly substitute ordinary electrical copper instead of oxygen-free copper - it is impossible to distinguish one from the other by eye. However, there is an area of application where counterfeiting is not clear: twisted pair cable for computer networks. If you put a grid with long segments on the left, it will either not start at all or will constantly glitch. Momentum dispersion, you know.

The author, when there was just talk about audio wires, realized that, in principle, this was not idle chatter, especially since oxygen-free wires by that time had long been used in special-purpose equipment, with which he was well acquainted by his line of work. Then I took and replaced the standard cord of my TDS-7 headphones with a homemade one made from “vitukha” with flexible multi-core wires. The sound, aurally, has steadily improved for end-to-end analogue tracks, i.e. on the way from the studio microphone to the disc, never digitized. Vinyl recordings made using DMM (Direct Metal Mastering) technology sounded especially bright. After this, the interconnect installation of all home audio was converted to “vitushka”. Then completely random people, indifferent to the music and not notified in advance, began to notice the improvement in sound.

How to make interconnect wires from twisted pair, see next. video.

Video: do-it-yourself twisted pair interconnect wires

Unfortunately, the flexible “vitha” soon disappeared from sale - it did not hold well in the crimped connectors. However, for the information of readers, flexible “military” wire MGTF and MGTFE (shielded) is made only from oxygen-free copper. Fake is impossible, because On ordinary copper, tape fluoroplastic insulation spreads quite quickly. MGTF is now widely available and costs much less than branded audio cables with a guarantee. It has one drawback: it cannot be done in color, but this can be corrected with tags. There are also oxygen-free winding wires, see below.

Theoretical Interlude

As we can see, already in the early stages of mastering audio technology, we had to deal with the concept of Hi-Fi (High Fidelity), high fidelity sound reproduction. Hi-Fi comes in different levels, which are ranked according to the following. main parameters:

Reproducible frequency band.
Dynamic range - the ratio in decibels (dB) of the maximum (peak) output power to the noise level.
Self-noise level in dB.
Nonlinear distortion factor (THD) at rated (long-term) output power. The SOI at peak power is assumed to be 1% or 2% depending on the measurement technique.
Unevenness of the amplitude-frequency response (AFC) in the reproducible frequency band. For speakers - separately at low (LF, 20-300 Hz), medium (MF, 300-5000 Hz) and high (HF, 5000-20,000 Hz) sound frequencies.

Note: the ratio of absolute levels of any values of I in (dB) is defined as P(dB) = 20log(I1/I2). If I1

You need to know all the subtleties and nuances of Hi-Fi when designing and building speakers, and as for a homemade Hi-Fi UMZCH for the home, before moving on to these, you need to clearly understand the requirements for their power required to sound a given room, dynamic range (dynamics), noise level and SOI. It is not very difficult to achieve a frequency band of 20-20,000 Hz from the UMZCH with a roll off at the edges of 3 dB and an uneven frequency response in the midrange of 2 dB on a modern element base.

Volume

The power of the UMZCH is not an end in itself; it must provide the optimal volume of sound reproduction in a given room. It can be determined by curves of equal loudness, see fig. There are no natural noises in residential areas quieter than 20 dB; 20 dB is the wilderness in complete calm. A volume level of 20 dB relative to the threshold of audibility is the threshold of intelligibility - a whisper can still be heard, but music is perceived only as the fact of its presence. An experienced musician can tell which instrument is being played, but not what exactly.

40 dB - the normal noise of a well-insulated city apartment in a quiet area or a country house - represents the intelligibility threshold. Music from the threshold of intelligibility to the threshold of intelligibility can be listened to with deep frequency response correction, primarily in the bass. To do this, the MUTE function (mute, mutation, not mutation!) is introduced into modern UMZCHs, including, respectively. correction circuits in UMZCH.

90 dB is the volume level of a symphony orchestra in a very good concert hall. 110 dB can be produced by an extended orchestra in a hall with unique acoustics, of which there are no more than 10 in the world, this is the threshold of perception: louder sounds are still perceived as distinguishable in meaning with an effort of will, but already annoying noise. The volume zone in residential premises of 20-110 dB constitutes the zone of complete audibility, and 40-90 dB is the zone of best audibility, in which untrained and inexperienced listeners fully perceive the meaning of the sound. If, of course, he is in it.

Power

Calculating the power of equipment at a given volume in the listening area is perhaps the main and most difficult task of electroacoustics. For yourself, in conditions it is better to go from acoustic systems (AS): calculate their power using a simplified method, and take the nominal (long-term) power of the UMZCH equal to the peak (musical) speaker. In this case, the UMZCH will not noticeably add its distortions to those of the speakers; they are already the main source of nonlinearity in the audio path. But the UMZCH should not be made too powerful: in this case, the level of its own noise may be higher than the threshold of audibility, because It is calculated based on the voltage level of the output signal at maximum power. If we consider it very simply, then for a room in an ordinary apartment or house and speakers with normal characteristic sensitivity (sound output) we can take the trace. UMZCH optimal power values:

Up to 8 sq. m – 15-20 W.
8-12 sq. m – 20-30 W.
12-26 sq. m – 30-50 W.
26-50 sq. m – 50-60 W.
50-70 sq. m – 60-100 W.
70-100 sq. m – 100-150 W.
100-120 sq. m – 150-200 W.
More than 120 sq. m – determined by calculation based on on-site acoustic measurements.

Dynamics

The dynamic range of the UMZCH is determined by curves of equal loudness and threshold values for different degrees of perception:

Symphonic music and jazz with symphonic accompaniment - 90 dB (110 dB - 20 dB) ideal, 70 dB (90 dB - 20 dB) acceptable. No expert can distinguish a sound with a dynamics of 80-85 dB in a city apartment from ideal.
Other serious music genres – 75 dB excellent, 80 dB “through the roof”.
Pop music of any kind and movie soundtracks - 66 dB is enough for the eyes, because... These opuses are already compressed during recording to levels of up to 66 dB and even up to 40 dB, so that you can listen to them on anything.

The dynamic range of the UMZCH, correctly selected for a given room, is considered equal to its own noise level, taken with the + sign, this is the so-called. signal-to-noise ratio.

SOI

Nonlinear distortions (ND) of UMZCH are components of the output signal spectrum that were not present in the input signal. Theoretically, it is best to “push” the NI under the level of its own noise, but technically this is very difficult to implement. In practice, they take into account the so-called. masking effect: at volume levels below approx. At 30 dB, the range of frequencies perceived by the human ear narrows, as does the ability to distinguish sounds by frequency. Musicians hear notes, but find it difficult to assess the timbre of the sound. In people without a hearing for music, the masking effect is observed already at 45-40 dB of volume. Therefore, an UMZCH with a THD of 0.1% (–60 dB from a volume level of 110 dB) will be assessed as Hi-Fi by the average listener, and with a THD of 0.01% (–80 dB) can be considered not distorting the sound.

Lamps

The last statement will probably cause rejection, even fury, among adherents of tube circuitry: they say, real sound is produced only by tubes, and not just some, but certain types of octal ones. Calm down, gentlemen - the special tube sound is not a fiction. The reason is the fundamentally different distortion spectra of electronic tubes and transistors. Which, in turn, are due to the fact that in the lamp the flow of electrons moves in a vacuum and quantum effects do not appear in it. A transistor is a quantum device, where minority charge carriers (electrons and holes) move in the crystal, which is completely impossible without quantum effects. Therefore, the spectrum of tube distortions is short and clean: only harmonics up to the 3rd - 4th are clearly visible in it, and there are very few combinational components (sums and differences in the frequencies of the input signal and their harmonics). Therefore, in the days of vacuum circuitry, SOI was called harmonic distortion (CHD). In transistors, the spectrum of distortions (if they are measurable, the reservation is random, see below) can be traced up to the 15th and higher components, and there are more than enough combination frequencies in it.

At the beginning of solid-state electronics, designers of transistor UMZCHs used the usual “tube” SOI of 1-2% for them; Sound with a tube distortion spectrum of this magnitude is perceived by ordinary listeners as pure. By the way, the very concept of Hi-Fi did not yet exist. It turned out that they sound dull and dull. In the process of developing transistor technology, an understanding of what Hi-Fi is and what is needed for it was developed.

Currently, the growing pains of transistor technology have been successfully overcome and side frequencies at the output of a good UMZCH are difficult to detect using special measurement methods. And lamp circuitry can be considered to have become an art. Its basis can be anything, why can’t electronics go there? An analogy with photography would be appropriate here. No one can deny that a modern digital SLR camera produces an image that is immeasurably clearer, more detailed, and deeper in the range of brightness and color than a plywood box with an accordion. But someone, with the coolest Nikon, “clicks pictures” like “this is my fat cat, he got drunk like a bastard and is sleeping with his paws outstretched,” and someone, using Smena-8M, uses Svemov’s b/w film to take a picture in front of which there is a crowd of people at a prestigious exhibition.

Note: and calm down again - not everything is so bad. Today, low-power lamp UMZCHs have at least one application left, and not the least important, for which they are technically necessary.

Experimental stand

Many audio lovers, having barely learned to solder, immediately “go into tubes.” This in no way deserves censure, on the contrary. Interest in the origins is always justified and useful, and electronics has become so with tubes. The first computers were tube-based, and the on-board electronic equipment of the first spacecraft was also tube-based: there were already transistors then, but they could not withstand extraterrestrial radiation. By the way, at that time lamp microcircuits were also created under the strictest secrecy! On microlamps with a cold cathode. The only known mention of them in open sources is in the rare book by Mitrofanov and Pickersgil “Modern receiving and amplifying tubes”.

But enough of the lyrics, let's get to the point. For those who like to tinker with the lamps in Fig. – diagram of a bench lamp UMZCH, intended specifically for experiments: SA1 switches the operating mode of the output lamp, and SA2 switches the supply voltage. The circuit is well known in the Russian Federation, a minor modification affected only the output transformer: now you can not only “drive” the native 6P7S in different modes, but also select the screen grid switching factor for other lamps in ultra-linear mode; for the vast majority of output pentodes and beam tetrodes it is either 0.22-0.25 or 0.42-0.45. For the manufacture of the output transformer, see below.

Guitarists and rockers

This is the very case when you can’t do without lamps. As you know, the electric guitar became a full-fledged solo instrument after the pre-amplified signal from the pickup began to be passed through a special attachment - a fuser - which deliberately distorted its spectrum. Without this, the sound of the string was too sharp and short, because the electromagnetic pickup reacts only to the modes of its mechanical vibrations in the plane of the instrument soundboard.

An unpleasant circumstance soon emerged: the sound of an electric guitar with a fuser acquires full strength and brightness only at high volumes. This is especially true for guitars with a humbucker-type pickup, which gives the most “angry” sound. But what about a beginner who is forced to rehearse at home? You can’t go to the hall to perform without knowing exactly how the instrument will sound there. And rock fans just want to listen to their favorite things in full juice, and rockers are generally decent and non-conflict people. At least those who are interested in rock music, and not shocking surroundings.

So, it turned out that the fatal sound appears at volume levels acceptable for residential premises, if the UMZCH is tube-based. The reason is the specific interaction of the signal spectrum from the fuser with the pure and short spectrum of tube harmonics. Here again an analogy is appropriate: a b/w photo can be much more expressive than a color one, because leaves only the outline and light for viewing.

Those who need a tube amplifier not for experiments, but due to technical necessity, do not have time to master the intricacies of tube electronics for a long time, they are passionate about something else. In this case, it is better to make the UMZCH transformerless. More precisely, with a single-ended matching output transformer that operates without constant magnetization. This approach greatly simplifies and speeds up the production of the most complex and critical component of a lamp UMZCH.

“Transformerless” tube output stage of the UMZCH and pre-amplifiers for it

On the right in Fig. a diagram of a transformerless output stage of a tube UMZCH is given, and on the left are pre-amplifier options for it. At the top - with a tone control according to the classic Baxandal scheme, which provides fairly deep adjustment, but introduces slight phase distortion into the signal, which can be significant when operating an UMZCH on a 2-way speaker. Below is a preamplifier with simpler tone control that does not distort the signal.

But let's get back to the end. In a number of foreign sources, this scheme is considered a revelation, but an identical one, with the exception of the capacitance of the electrolytic capacitors, is found in the Soviet “Radio Amateur Handbook” of 1966. A thick book of 1060 pages. There was no Internet and disk-based databases back then.

In the same place, on the right in the figure, the disadvantages of this scheme are briefly but clearly described. An improved one, from the same source, is given on the trail. rice. on right. In it, the screen grid L2 is powered from the midpoint of the anode rectifier (the anode winding of the power transformer is symmetrical), and the screen grid L1 is powered through the load. If, instead of high-impedance speakers, you turn on a matching transformer with regular speakers, as in the previous one. circuit, the output power is approx. 12 W, because the active resistance of the primary winding of the transformer is much less than 800 Ohms. SOI of this final stage with transformer output - approx. 0.5%

How to make a transformer?

The main enemies of the quality of a powerful signal low-frequency (sound) transformer are the magnetic leakage field, the lines of force of which are closed, bypassing the magnetic circuit (core), eddy currents in the magnetic circuit (Foucault currents) and, to a lesser extent, magnetostriction in the core. Because of this phenomenon, a carelessly assembled transformer “sings,” hums, or beeps. Foucault currents are combated by reducing the thickness of the magnetic circuit plates and additionally insulating them with varnish during assembly. For output transformers, the optimal plate thickness is 0.15 mm, the maximum allowable is 0.25 mm. You should not take thinner plates for the output transformer: the fill factor of the core (the central rod of the magnetic circuit) with steel will fall, the cross-section of the magnetic circuit will have to be increased to obtain a given power, which will only increase distortions and losses in it.

In the core of an audio transformer operating with constant bias (for example, the anode current of a single-ended output stage) there must be a small (determined by calculation) non-magnetic gap. The presence of a non-magnetic gap, on the one hand, reduces signal distortion from constant magnetization; on the other hand, in a conventional magnetic circuit it increases the stray field and requires a core with a larger cross-section. Therefore, the non-magnetic gap must be calculated at the optimum and performed as accurately as possible.

For transformers operating with magnetization, the optimal type of core is made of Shp (cut) plates, pos. 1 in Fig. In them, a non-magnetic gap is formed during core cutting and is therefore stable; its value is indicated in the passport for the plates or measured with a set of probes. The stray field is minimal, because the side branches through which the magnetic flux is closed are solid. Transformer cores without bias are often assembled from Shp plates, because Shp plates are made from high-quality transformer steel. In this case, the core is assembled across the roof (the plates are laid with a cut in one direction or the other), and its cross-section is increased by 10% compared to the calculated one.

It is better to wind transformers without magnetization on USH cores (reduced height with widened windows), pos. 2. In them, a decrease in the stray field is achieved by reducing the length of the magnetic path. Since USh plates are more accessible than Shp, transformer cores with magnetization are often made from them. Then the core assembly is carried out cut to pieces: a package of W-plates is assembled, a strip of non-conducting non-magnetic material is placed with a thickness equal to the size of the non-magnetic gap, covered with a yoke from a package of jumpers and pulled together with a clip.

Note:“sound” signal magnetic circuits of the ShLM type are of little use for output transformers of high-quality tube amplifiers; they have a large stray field.

At pos. 3 shows a diagram of the core dimensions for calculating the transformer, at pos. 4 design of the winding frame, and at pos. 5 – patterns of its parts. As for the transformer for the “transformerless” output stage, it is better to make it on the ShLMm across the roof, because the bias is negligible (the bias current is equal to the screen grid current). The main task here is to make the windings as compact as possible in order to reduce the stray field; their active resistance will still be much less than 800 Ohms. The more free space left in the windows, the better the transformer turned out. Therefore, the windings are wound turn to turn (if there is no winding machine, this is a terrible task) from the thinnest possible wire; the laying coefficient of the anode winding for the mechanical calculation of the transformer is taken 0.6. The winding wire is PETV or PEMM, they have an oxygen-free core. There is no need to take PETV-2 or PEMM-2; due to double varnishing, they have an increased outer diameter and a larger scattering field. The primary winding is wound first, because it is its scattering field that most affects the sound.

You need to look for iron for this transformer with holes in the corners of the plates and clamping brackets (see figure on the right), because “for complete happiness,” the magnetic circuit is assembled as follows. order (of course, the windings with leads and external insulation should already be on the frame):

Prepare acrylic varnish diluted in half or, in the old fashioned way, shellac;
Plates with jumpers are quickly coated with varnish on one side and placed into the frame as quickly as possible, without pressing too hard. The first plate is placed with the varnished side inward, the next one with the unvarnished side to the first varnished, etc.;
When the frame window is filled, staples are applied and bolted tightly;
After 1-3 minutes, when the squeezing of varnish from the gaps apparently stops, add plates again until the window is filled;
Repeat paragraphs. 2-4 until the window is tightly packed with steel;
The core is pulled tightly again and dried on a battery, etc. 3-5 days.

The core assembled using this technology has very good plate insulation and steel filling. Magnetostriction losses are not detected at all. But keep in mind that this technique is not applicable for permalloy cores, because Under strong mechanical influences, the magnetic properties of permalloy irreversibly deteriorate!

On microcircuits

UMZCHs on integrated circuits (ICs) are most often made by those who are satisfied with the sound quality up to average Hi-Fi, but are more attracted by the low cost, speed, ease of assembly and the complete absence of any setup procedures that require special knowledge. Simply, an amplifier on microcircuits is the best option for dummies. The classic of the genre here is the UMZCH on the TDA2004 IC, which has been on the series, God willing, for about 20 years now, on the left in Fig. Power – up to 12 W per channel, supply voltage – 3-18 V unipolar. Radiator area – from 200 sq. see for maximum power. The advantage is the ability to work with a very low-resistance, up to 1.6 Ohm, load, which allows you to extract full power when powered from a 12 V on-board network, and 7-8 W when supplied with a 6-volt power supply, for example, on a motorcycle. However, the output of the TDA2004 in class B is not complementary (on transistors of the same conductivity), so the sound is definitely not Hi-Fi: THD 1%, dynamics 45 dB.

The more modern TDA7261 does not produce better sound, but is more powerful, up to 25 W, because The upper limit of the supply voltage has been increased to 25 V. The lower limit, 4.5 V, still allows it to be powered from a 6 V on-board network, i.e. The TDA7261 can be started from almost all on-board networks, except for the aircraft 27 V. Using attached components (strapping, on the right in the figure), the TDA7261 can operate in mutation mode and with the St-By (Stand By) function, which switches the UMZCH to the minimum power consumption mode when there is no input signal for a certain time. Convenience costs money, so for a stereo you will need a pair of TDA7261 with radiators from 250 sq. see for each.

Note: If you are somehow attracted to amplifiers with the St-By function, keep in mind that you should not expect speakers wider than 66 dB from them.

“Super economical” in terms of power supply TDA7482, on the left in the figure, operating in the so-called. class D. Such UMZCHs are sometimes called digital amplifiers, which is incorrect. For real digitization, level samples are taken from an analog signal with a quantization frequency that is no less than twice the highest of the reproduced frequencies, the value of each sample is recorded in a noise-resistant code and stored for further use. UMZCH class D – pulse. In them, the analogue is directly converted into a sequence of high-frequency pulse-width modulated (PWM), which is fed to the speaker through a low-pass filter (LPF).

Class D sound has nothing in common with Hi-Fi: SOI of 2% and dynamics of 55 dB for class D UMZCH are considered very good indicators. And TDA7482 here, it must be said, is not the optimal choice: other companies specializing in class D produce UMZCH ICs that are cheaper and require less wiring, for example, D-UMZCH of the Paxx series, on the right in Fig.

Among the TDAs, the 4-channel TDA7385 should be noted, see the figure, on which you can assemble a good amplifier for speakers up to medium Hi-Fi, inclusive, with frequency division into 2 bands or for a system with a subwoofer. In both cases, low-pass and mid-high-frequency filtering is done at the input on a weak signal, which simplifies the design of the filters and allows deeper separation of the bands. And if the acoustics are subwoofer, then 2 channels of the TDA7385 can be allocated for the sub-ULF bridge circuit (see below), and the remaining 2 can be used for MF-HF.

UMZCH for subwoofer

A subwoofer, which can be translated as “subwoofer” or, literally, “boomer,” reproduces frequencies up to 150-200 Hz; in this range, human ears are practically unable to determine the direction of the sound source. In speakers with a subwoofer, the “sub-bass” speaker is placed in a separate acoustic design, this is the subwoofer as such. The subwoofer is placed, in principle, as conveniently as possible, and the stereo effect is provided by separate MF-HF channels with their own small-sized speakers, for the acoustic design of which there are no particularly serious requirements. Experts agree that it is better to listen to stereo with full channel separation, but subwoofer systems significantly save money or labor on the bass path and make it easier to place acoustics in small rooms, which is why they are popular among consumers with normal hearing and not particularly demanding ones.

The “leakage” of mid-high frequencies into the subwoofer, and from it into the air, greatly spoils the stereo, but if you sharply “cut off” the sub-bass, which, by the way, is very difficult and expensive, then a very unpleasant sound jumping effect will occur. Therefore, channels in subwoofer systems are filtered twice. At the input, electric filters highlight midrange-high frequencies with bass “tails” that do not overload the midrange-high frequency path, but provide a smooth transition to sub-bass. Bass with midrange “tails” are combined and fed to a separate UMZCH for the subwoofer. The midrange is additionally filtered so that the stereo does not deteriorate; in the subwoofer it is already acoustic: a sub-bass speaker is placed, for example, in the partition between the resonator chambers of the subwoofer, which do not let the midrange out, see on the right in Fig.

A UMZCH for a subwoofer is subject to a number of specific requirements, of which “dummies” consider the most important to be as high a power as possible. This is completely wrong, if, say, the calculation of the acoustics for the room gave a peak power W for one speaker, then the power of the subwoofer needs 0.8 (2W) or 1.6W. For example, if S-30 speakers are suitable for the room, then a subwoofer needs 1.6x30 = 48 W.

It is much more important to ensure the absence of phase and transient distortions: if they occur, there will definitely be a jump in the sound. As for SOI, it is permissible up to 1%. Intrinsic bass distortion of this level is not audible (see curves of equal volume), and the “tails” of their spectrum in the best audible midrange region will not come out of the subwoofer.

To avoid phase and transient distortions, the amplifier for the subwoofer is built according to the so-called. bridge circuit: the outputs of 2 identical UMZCHs are switched on back-to-back through a speaker; signals to the inputs are supplied in antiphase. The absence of phase and transient distortions in the bridge circuit is due to the complete electrical symmetry of the output signal paths. The identity of the amplifiers forming the arms of the bridge is ensured by the use of paired UMZCHs on ICs, made on the same chip; This is perhaps the only case when an amplifier on microcircuits is better than a discrete one.

Note: The power of a bridge UMZCH does not double, as some people think, it is determined by the supply voltage.

An example of a bridge UMZCH circuit for a subwoofer in a room up to 20 sq. m (without input filters) on the TDA2030 IC is given in Fig. left. Additional midrange filtering is carried out by circuits R5C3 and R’5C’3. Radiator area TDA2030 – from 400 sq. see. Bridged UMZCHs with an open output have an unpleasant feature: when the bridge is unbalanced, a constant component appears in the load current, which can damage the speaker, and the sub-bass protection circuits often fail, turning off the speaker when not needed. Therefore, it is better to protect the expensive oak bass head with non-polar batteries of electrolytic capacitors (highlighted in color, and the diagram of one battery is given in the inset.

A little about acoustics

The acoustic design of a subwoofer is a special topic, but since a drawing is given here, explanations are also needed. Case material – MDF 24 mm. The resonator tubes are made of fairly durable, non-ringing plastic, for example, polyethylene. The internal diameter of the pipes is 60 mm, the protrusions inward are 113 mm in the large chamber and 61 in the small chamber. For a specific loudspeaker head, the subwoofer will have to be reconfigured for the best bass and, at the same time, the least impact on the stereo effect. To tune the pipes, they take a pipe that is obviously longer and, by pushing it in and out, achieve the required sound. The protrusions of the pipes outward do not affect the sound; they are then cut off. The pipe settings are interdependent, so you will have to tinker.

Headphone Amplifier

A headphone amplifier is most often made by hand for two reasons. The first is for listening “on the go”, i.e. outside the home, when the power of the audio output of the player or smartphone is not enough to drive “buttons” or “burdocks”. The second is for high-end home headphones. A Hi-Fi UMZCH for an ordinary living room is needed with dynamics of up to 70-75 dB, but the dynamic range of the best modern stereo headphones exceeds 100 dB. An amplifier with such dynamics costs more than some cars, and its power will be from 200 W per channel, which is too much for an ordinary apartment: listening at a power that is much lower than the rated power spoils the sound, see above. Therefore, it makes sense to make a low-power, but with good dynamics, a separate amplifier specifically for headphones: the prices for household UMZCHs with such an additional weight are clearly absurdly inflated.

The circuit of the simplest headphone amplifier using transistors is given in pos. 1 pic. The sound is only for Chinese “buttons”, it works in class B. It is also no different in terms of efficiency - 13 mm lithium batteries last for 3-4 hours at full volume. At pos. 2 – TDA’s classic for on-the-go headphones. The sound, however, is quite decent, up to average Hi-Fi depending on the track digitization parameters. There are countless amateur improvements to the TDA7050 harness, but no one has yet achieved the transition of sound to the next level of class: the “microphone” itself does not allow it. TDA7057 (item 3) is simply more functional; you can connect the volume control to a regular, not dual, potentiometer.

The UMZCH for headphones on the TDA7350 (item 4) is designed to drive good individual acoustics. It is on this IC that headphone amplifiers in most middle and high-class household UMZCHs are assembled. The UMZCH for headphones on KA2206B (item 5) is already considered professional: its maximum power of 2.3 W is enough to drive such serious isodynamic “mugs” as TDS-7 and TDS-15.