The high-fidelity audio power amplifier (AMP), developed in 1989 by Nikolai Sukhov, can rightfully be called legendary. During its development, a professional approach was used, based on knowledge and experience in the field of analog circuitry. As a result, the parameters of this amplifier turned out to be so high that even today this design has not lost its relevance. This article describes a slightly improved version of the amplifier. Improvements come down to the use of a new element base and the use of a microcontroller control system.
A power amplifier (PA) is an integral part of any sound reproduction complex. Many descriptions of the design of such amplifiers are available. But in the vast majority of cases, even with very good characteristics, there is a complete lack of service amenities. But nowadays, when microcontrollers have become widespread, creating a sufficiently advanced control system is not particularly difficult. At the same time, in terms of functionality, a homemade device may not be inferior to the best branded samples. A version of the UMZCH BB with a microcontroller control system is shown in Fig. 1:
Rice. 1. Appearance of the amplifier.
The original circuit of the UMZCH VV has sufficient parameters to ensure that the amplifier is not the dominant source of nonlinearity in the sound-reproducing path over the entire range of output powers. Therefore, further improvement of the characteristics no longer provides noticeable advantages.
At least, the sound quality of different soundtracks differs much more than the sound quality of amplifiers. On this topic you can quote from the magazine “Audio”: “ There are aurally obvious differences in categories such as speakers, microphones, LP pickups, listening rooms, studio spaces, concert halls, and especially the studio and recording equipment configurations used by different record companies. If you want to hear subtle differences in soundstage, compare John Eargle's Delos recordings to Jack Renner's Telarc recordings, not the preamps. Or if you want to hear subtle differences in transitions, compare dmp studio jazz recordings with Chesky studio jazz recordings, rather than two interconnects.»
Despite this fact, Hi-End lovers continue to search for the “right” sound, which also affects the mind. In fact, the PA is an example of a very simple linear path. The current level of development of circuit technology makes it possible to provide such a device with sufficiently high parameters so that the introduced distortions become invisible. Therefore, in practice, any two modern, non-eccentrically designed PAs sound the same. On the contrary, if a mind has some special, specific sound, this only means one thing: the distortions introduced by such a mind are large and clearly noticeable by ear.
This does not mean that designing a high-quality mind is very easy. There are many subtleties, both circuitry and design. But all these subtleties have long been known to serious PA manufacturers, and gross errors in the designs of modern PAs are usually not found. The exception is expensive Hi-End amplifiers, which are often very poorly designed. Even if the distortion introduced by the PA is pleasant to the ear (as fans of tube amplifiers claim), this has nothing to do with high fidelity of sound reproduction.
In addition to the traditional requirements of broadband and good linearity, a high-quality PA is subject to a number of additional requirements. Sometimes you can hear that an amplifier power of 20-35 W is sufficient for home use. If we are talking about average power, then this statement is true. But a real music signal can have a peak power level that is 10 to 20 times higher than the average level. Therefore, in order to obtain undistorted reproduction of such a signal with an average power of 20 W, it is necessary to have a PA power of about 200 W. Here, for example, is the conclusion of an expert assessment for the amplifier described in: “ The only criticism was the insufficient volume of the sound of large percussion instruments, which is explained by the insufficient output power of the amplifier (120 W peak into a 4 Ohm load).»
Acoustic systems (AS) represent a complex load and have a very complex dependence of impedance on frequency. At some frequencies it may be 3 to 4 times less than the nominal value. The PA must be able to operate without distortion on such a low-impedance load. For example, if the nominal impedance of the speaker system is 4 ohms, then the PA should work normally with a load of 1 ohm. This requires very large output currents, which must be taken into account when designing the PA. The described amplifier satisfies these requirements.
Recently, the topic of optimal amplifier output impedance from the point of view of minimizing speaker distortion has been quite often discussed. However, this topic is relevant only when designing active speakers. Passive speaker crossover filters are designed on the assumption that the signal source will have a negligibly low output impedance. If the PA has a high output impedance, then the frequency response of such speakers will be greatly distorted. Therefore, there is nothing else left to do but to provide low output impedance for the PA.
It can be noted that new developments of PAs are mainly taking the path of reducing costs, improving manufacturability of the design, increasing output power, increasing efficiency, and improving consumer qualities. This article focuses on service functions that are implemented thanks to a microcontroller control system.
The amplifier is made in a MIDI format case, its overall dimensions are 348x180x270 mm, weight is about 20 kg. The built-in microcontroller allows you to control the amplifier using an IR remote control (shared with the pre-amplifier). In addition, the microcontroller measures and displays average and quasi-peak output power, radiator temperature, implements timer shutdowns, and processes emergency situations. The amplifier protection system, as well as power on and off control, are implemented with the participation of a microcontroller. The amplifier has a separate standby power supply, which allows it to be in “STANDBY” mode when the main power sources are turned off.
The described amplifier is called NSM (National Sound Machines), model PA-9000, since the name of the device is part of its design and must be present. The implemented set of service functions in some cases may turn out to be redundant; for such situations, a “minimalist” version of the amplifier (model PA-2020) has been developed, which has only a power switch and a two-color LED on the front panel, and the built-in microcontroller only controls the process of turning the power on and off, complements the protection system and provides remote control of the “STANDBY” mode.
All controls and indications of the amplifier are located on the front panel. Its appearance and the purpose of the controls are shown in Fig. 2:
Rice. 2. Front panel of the amplifier.
| 1 - LED for turning on external consumers EXT | 9 - minus button |
| 2 - DUTY power supply LED | 10 - peak power indication button PEAK |
| 3 - button to switch to standby mode STANDBY | 11 - TIMER indication button |
| 4 - POWER button | 12 - temperature indication button°C |
| 5 - main power LED MAIN | 13 - plus button |
| 6 - LED for normal operation OPERATE | 14 - left channel failure LED FAIL L |
| 7 - load switch LED LOAD | 15 - right channel failure LED FAIL R |
| 8 - display |
POWER button ensures complete disconnection of the amplifier from the network. Physically, this button disconnects only the standby power source from the network; accordingly, it can be designed for a small current. The main power sources are switched on using relays, the windings of which are powered from a standby source. Therefore, when the “POWER” button is disabled, all amplifier circuits are guaranteed to be de-energized.
When the POWER button is turned on, the amplifier is fully turned on. The switching process occurs as follows: the standby source is immediately turned on, as evidenced by the “DUTY” standby power supply LED. After some time required to reset the microcontroller, power is turned on to the external sockets and the “EXT” LED lights up. Then the “MAIN” LED lights up, and the first stage of turning on the main sources occurs. Initially, the main transformers are switched on through limiting resistors, which prevent the initial inrush current due to discharged filter capacitors. The capacitors are gradually charged, and when the measured supply voltage reaches a set threshold, the limiting resistors are removed from the circuit. At the same time, the “OPERATE” LED lights up. If within the allotted time the supply voltage has not reached the set threshold, the process of turning on the amplifier is interrupted and an alarm indication is turned on. If the switching on of the main sources was successful, the microcontroller checks the status of the protection system. In the absence of emergency situations, the microcontroller allows the load relay to turn on and the “LOAD” LED lights up.
STANDBY button controls the standby mode. A short press of the button puts the amplifier into standby mode or, conversely, turns on the amplifier. In practice, you may need to turn on external sockets while leaving the PA in standby mode. This is required, for example, when listening to soundtracks on stereo phones or when dubbing without sound control. External sockets can be independently turned on and off by long (until the sound signal) presses the “STANDBY” button. The option when the PA is turned on and the sockets are turned off does not make sense, so it is not implemented.
The front panel contains a 4-digit digital display and 5 display control buttons. The display can operate in the following modes (Fig. 3a):
Rice. 3. Display options.
PEAK button turns on the output power display and switches between average/quasi-peak power modes. In the output power indication mode, “W” lights up on the display, and for quasi-peak power, “PEAK” also lights up. The output power is indicated in watts with a resolution of 0.1 watts. The measurement is made by multiplying the current and voltage across the load, so the readings are valid for any permissible load resistance value. Holding the PEAK button until a beep turns off the display. Turning off the display, as well as its switching between different display modes, occurs smoothly (one image “flows” into another). This effect is implemented in software.
TIMER button displays the current state of the timer, and the letter “M” lights up. The timer allows you to set a time interval after which the amplifier goes into standby mode and the external sockets are turned off. It should be noted that when using this function, other components of the complex must allow power to be turned off “on the fly”. For a tuner and CD player, this is usually acceptable, but for some cassette decks, when the power is turned off, the CVL may not go into the “STOP” mode. These decks cannot be powered off during playback or recording. However, among branded devices such decks are extremely rare. On the contrary, most decks have a “Timer” switch, which has 3 positions: “Off”, “Record” and “Play”, which allows you to immediately turn on playback or recording mode by simply turning on the power. You can also turn off these modes by simply removing the power. The amplifier timer can be programmed at the following intervals (Fig. 3b): 5, 15, 30, 45, 60, 90 and 120 minutes. If the timer is not used, it must be set to “OFF”. It is in this state immediately after turning on the power.
The timer interval is set "+" and "-" buttons in timer display mode. If the timer is turned on, the “TIMER” LED is always lit on the display, and turning on the timer indication shows the real current state, i.e. how many minutes are left before shutdown? In such a situation, the interval can be extended by pressing the “+” button.
"°C" button turns on the display of the temperature of the radiators, and the “°C” symbol lights up. Each radiator has a separate thermometer, but the maximum temperature value is displayed on the display. The same thermometers are used to control the fan and for temperature protection of the amplifier's output transistors.
For accident indication There are two LEDs on the front panel: “FAIL LEFT” and “FAIL RIGHT”. When the protection is triggered, the corresponding LED lights up in one of the PA channels, and the letter name of the cause of the accident is indicated on the display (Fig. 3c). In this case, the amplifier goes into standby mode. The amplifier implements the following types of protection:
It reacts to the DC component of the PA output voltage, greater than 2 V. It protects the speakers and is also implemented in hardware. Indicated on the display as “dcF”.
Reacts to a drop in the supply voltage of any arm below a specified level. A significant violation of the symmetry of the supply voltages can cause the appearance of a constant component at the output of the PA, which is dangerous for the speaker system. Indicated on the display as “UF”.
Reacts to loss of several periods of mains voltage in a row. The purpose of this protection is to turn off the load before the supply voltage drops and a transient begins. Implemented in hardware, the microcontroller only reads its state. Indicated on the display as “prF”.
overheat protection output transistors are implemented in software; it uses information from thermometers that are installed on radiators. Indicated on the display as “tF”.
The mind has the ability remote control. Since many control buttons are not required, the same remote control is used as for controlling the preamplifier. This remote control operates in the RC-5 standard and has three buttons specifically designed for controlling the PA. The “STANDBY” button completely duplicates the similar button on the front panel. The “DISPLAY” button allows you to switch the display mode in a ring (Fig. 3a). Holding down the DISPLAY button until a beep turns off the display. The “MODE” button allows you to change the time interval of the timer (Fig. 3b), i.e. it replaces the “+” and “-” buttons.
On rear panel amplifier (Fig. 4) there are sockets installed for powering other components of the complex. These sockets have an independent shutdown, which allows you to turn off the power to the entire complex using the remote control.
Rice. 4. Rear panel of the amplifier.
As noted earlier, the basis for the described amplifier is the UMZCH VV circuit of Nikolai Sukhov, which is described in. The basic principles of building a high-fidelity mind are set out in. Schematic diagram amplifier main board shown in Fig. 5.
width=710>
Rice. 5. Schematic diagram of the main amplifier board.
Compared to the original design, minor changes have been made to the amplifier. These changes are not fundamental and mainly represent a transition to a newer element base.
Changed quiescent current temperature stabilization circuit. In the original design, along with the output transistors, a transistor was installed on the radiators - a temperature sensor, which set the bias voltage of the output stage. In this case, only the temperature of the output transistors was taken into account. But the temperature of the pre-terminal transistors, due to the rather large power dissipated by them, also increased significantly during operation. Because these transistors were mounted on small, separate heatsinks, their temperature could fluctuate quite dramatically, for example due to changes in power dissipation or even due to external air currents. This led to the same sharp fluctuations in the quiescent current. And any other element of the PA can get quite hot during operation, since heat sources are located in one case (radiators of output transistors, transformers, etc.). This also applies to the very first composite emitter follower transistors, which did not have radiators at all. As a result, the quiescent current could increase several times when the PA heats up. A solution to this problem was proposed by Alexey Belov.
Typically, to temperature stabilize the quiescent current of the PA output stages, the following circuit is used (Fig. 6a):
Rice. 6. Scheme of temperature stabilization of the quiescent current.
The bias voltage is applied to points A and B. It is allocated on a two-terminal network, which consists of transistor VT1 and resistors R1, R2. The initial bias voltage is set by resistor R2. Transistor VT1 is usually mounted on a common radiator with VT6, VT7. Stabilization is carried out as follows: when transistors VT6, VT7 are heated, the base-emitter drop decreases, which, at a fixed bias voltage, leads to an increase in the quiescent current. But along with these transistors, VT1 also heats up, which causes a decrease in the voltage drop across the two-terminal network, i.e. decrease in quiescent current. The disadvantage of this scheme is that the transition temperature of the remaining transistors included in the composite emitter follower is not taken into account. To take it into account, the junction temperature of all transistors must be known. The easiest way is to make it the same. To do this, it is enough to install all the transistors included in the composite emitter follower on a common radiator. Moreover, to obtain a quiescent current that does not depend on temperature, the bias voltage of the composite emitter follower must have a temperature coefficient the same as that of six p-n junctions connected in series. Approximately, we can assume that the forward voltage drop across the pn junction decreases linearly with a coefficient K approximately equal to 2.3 mV/°C. For a composite emitter follower, this coefficient is 6*K. To ensure such a temperature coefficient of the bias voltage is the task of a two-terminal network, which is connected between points A and B. The two-terminal network shown in Fig. 6a, has a temperature coefficient equal to (1+R2/R1)*K. When adjusting the quiescent current with resistor R2, the temperature coefficient also changes, which is not entirely correct. The simplest practical solution is the circuit shown in Fig. 6b. In this circuit, the temperature coefficient is equal to (1+R3/R1)*K, and the initial quiescent current is set by the position of the resistor R2 slider. The voltage drop across resistor R2, which is shunted by a diode, can be considered almost constant. Therefore, adjusting the initial quiescent current does not affect the temperature coefficient. With such a circuit, when the PA heats up, the quiescent current changes by no more than 10-20%. In order for all transistors in a composite emitter follower to be placed on a common heatsink, they must have packages suitable for mounting on the heatsink (transistors in TO-92 packages are not suitable). Therefore, other types of transistors are used in the PA, at the same time more modern ones.
In the amplifier circuit (Fig. 5), the two-terminal circuit for temperature stabilization of the quiescent current is shunted by capacitor C12. This capacitor is optional, although it does no harm either. The fact is that between the bases of the transistors of the composite emitter follower it is necessary to provide a bias voltage, which must be constant for the selected quiescent current and independent of the amplified signal. In short, the alternating component of the voltage on the two-terminal network, as well as on resistors R26 and R29 (Fig. 5) should be equal to zero. Therefore, all these elements can be bypassed with capacitors. But due to the low dynamic resistance of the two-terminal network, as well as the low resistance values of these resistors, the presence of shunt capacitors has a very weak effect. Therefore, these capacitances are not necessary, especially since to bypass R26 and R29 their ratings must be quite large (about 1 μF and 10 μF, respectively).
Output transistors The PAs are replaced by transistors KT8101A, KT8102A, which have a higher cutoff frequency of the current transfer coefficient. In high-power transistors, the effect of a drop in the current transfer coefficient with an increase in the collector current is quite pronounced. This effect is extremely undesirable for PAs, since here the transistors have to operate at high output currents. Modulation of the current transfer coefficient leads to a significant deterioration in the linearity of the amplifier output stage. To reduce the influence of this effect, parallel connection of two transistors is used in the output stage (and this is the minimum that can be afforded).
When connecting transistors in parallel, to reduce the influence of the spread of their parameters and equalize the operating currents, separate emitter resistors are used. For normal operation of the overcurrent protection system, a circuit has been added to isolate the maximum voltage value on diodes VD9 - VD12 (Fig. 5), since now it is necessary to remove the drop from not two, but from four emitter resistors.
Other transistors composite emitter follower - these are KT850A, KT851A (TO-220 housing) and KT940A, KT9115A (TO-126 housing). The quiescent current stabilization circuit uses a composite transistor KT973A (TO-126 package).
Replacement has also been made OU to more modern ones. The main op amp U1 is replaced by the AD744, which has increased speed and good linearity. Op-amp U2, which operates in the circuit for maintaining zero potential at the output of the UMZCH, is replaced by OP177, which has a low zero offset (no more than 15 µV). This made it possible to eliminate the trimming resistor for adjusting the bias. It should be noted that due to the peculiarities of the AD744 circuit design, op-amp U2 must provide an output voltage close to the supply voltage (pin 8 of the AD744 op-amp in terms of constant voltage is only two pn junctions away from pin 4). Therefore, not all types of precision op amps will be suitable. As a last resort, you can use a “pull-up” resistor from the output of the op-amp to –15 V. Op-amp U3, which operates in the impedance compensation circuit of the connecting speaker wires, is replaced by AD711. The parameters of this op-amp are not so critical, so a cheap op-amp with sufficient speed and a fairly low zero offset was chosen.
Resistor dividers R49 – R51, R52 – R54 and R47, R48 are added to the circuit, which are used to remove current and voltage signals for the power measurement circuit.
Implementation changed earthen chains. Since each amplifier channel is now completely assembled on a single board, there is no need for multiple ground wires that must be connected to a single point on the chassis. The special PCB topology ensures star-shaped ground circuit routing. The earth star is connected by one conductor to the common terminal of the power source. It should be noted that this topology is only suitable with completely separate power supplies for the left and right channels.
In the original amplifier circuit, the AC feedback loop covers both relay contacts, which connect the load. This measure was taken to reduce the influence of contact nonlinearity. However, this may cause problems with the operation of the constant component protection. The fact is that when the amplifier is turned on, power is supplied before the load relay turns on. At this time, a signal may be present at the input of the PA, and the transmission coefficient of the amplifier due to the broken feedback loop is very high. In this mode, the PA limits the signal, and the bias voltage compensation circuit is generally unable to maintain a zero DC component at the PA output. Therefore, even before connecting the load, it may be discovered that there is a constant component at the output of the PA, and then the protection system will work. It is very easy to eliminate this effect if you use a relay with changeover contacts.
Normally closed contacts must close the OOS loop in the same way as normally open contacts. In this case, when the relay is activated, the feedback is interrupted only for a very short time, during which all relay contacts are open. During this time, the relatively inertial protection for the constant component does not have time to operate. In Fig. Figure 7 shows the relay switching process recorded with a digital oscilloscope. As you can see, 4 ms after voltage is applied to the relay winding, the normally closed contacts open. After about another 3 ms, the normally open contacts close (with a noticeable chatter that lasts about 0.7 ms). Thus, the contacts are in “flight” for approximately 3 ms, and it is during this time that the feedback will be interrupted.
Rice. 7. AJS13113 relay switching process.
Protection circuit completely redesigned (Fig. 8). Now it is located on the main board. Thus, each channel has its own independent circuit. This is somewhat redundant, but each main board is completely autonomous and is a complete mono amplifier. Some of the protective functions are carried out by the microcontroller, but to increase reliability, a sufficient set of them is implemented in hardware. In principle, the amplifier board can operate without a microcontroller at all. Since the PA has a separate standby power supply, the protection circuit is powered from it (at +12V level). This makes the behavior of the protection circuit more predictable in the event of a failure of one of the main power sources.
width=710>
The drawing does not fit on the page and is therefore compressed!
To view it in full, click .
Rice. 8. Amplifier protection circuit.
Overcurrent protection includes a trigger assembled on transistors VT3, VT4 (Fig. 5), which turns on when transistor VT13 opens. VT13 receives a signal from the current sensor and opens when the current reaches the value set using trimming resistor R30. The trigger turns off the current generators VT5, VT6, which leads to the blocking of all transistors of the composite emitter follower. Zero output voltage is maintained in this mode using resistor R27 (Fig. 5). In addition, the state of the trigger is read through the chain VD13, R63 (Fig. 8), and when it is turned on, a low logic level is set at the inputs of the logic element U4D. Transistor VT24 provides an open collector output for the IOF (I Out Fail) signal, which is polled by the microcontroller.
DC protection implemented on transistors VT19 – VT22 and logic elements U4B, U4A. The signal from the output of the amplifier through the divider R57, R59 is fed to the low-pass filter R58C23 with a cutoff frequency of about 0.1 Hz, which selects the constant component of the signal. If a constant component of positive polarity appears, then transistor VT19, connected according to the OE circuit, opens. He, in turn, opens transistor VT22, and a high logic level appears at the inputs of logic element U4B. If a constant component of negative polarity appears, then transistor VT21, connected with OB, opens. This asymmetry is a necessary measure associated with the unipolar power supply of the protection circuit. In order to increase the current transfer coefficient, cascode connection of transistors VT21, VT20 (OB - OK) was used. Next, as in the first case, transistor VT22 opens, etc. Transistor VT23 is connected to the output of logic element U4A, which provides an open collector output for the DCF (DC Fail) signal.
Power failure protection contains an auxiliary rectifier (Fig. 13) VD1, VD2 (VD3, VD4), which has an anti-aliasing filter with a very small time constant. If several periods of mains voltage fail in a row, the output voltage of the rectifier drops, and a low logic level is set at the inputs of the U4C logic element (Fig. 8).
Logical signals from the three protection circuits described above are supplied to the “OR” element U5C, the output of which is generated at a low logic level if any of the circuits is triggered. In this case, capacitor C24 is discharged through diode VD17, and a low logic level appears at the inputs of logic element U5B (also at output U5A). This causes transistor VT27 to close and relay K1 to turn off. The R69C24 chain provides a certain minimum delay when turning on the power in case the microcontroller for some reason does not generate the initial delay. Transistor VT25 provides an open collector output for the OKL (OK Left) or OKR (OK Right) signal. The microcontroller may prohibit the relay from turning on. For this purpose, a VT26 transistor is installed. This feature is necessary to implement software protection against overheating, a software delay for turning on the relay, and to synchronize the operation of the left and right channel protection systems.
Interaction of the microcontroller with the hardware protection circuit the following: when the amplifier is turned on, after the supply voltage has reached the nominal value, the microcontroller polls the OKL and OKR hardware protection readiness signals. All this time, turning on the relay is prohibited by the microcontroller by maintaining the ENB (Enable) signal in a high logic level state. As soon as the microcontroller receives the ready signals, it creates a time delay and allows the relay to turn on. During operation of the amplifier, the microcontroller constantly monitors the readiness signal. If such a signal disappears for one of the channels, the microcontroller removes the ENB signal, thus turning off the relay in both channels. It then interrogates the security status signals to identify the channel and type of security.
overheat protection implemented entirely in software. If the radiators overheat, the microcontroller removes the ENB signal, which causes the load relay to turn off. To measure the temperature, a DS1820 thermometer from Dallas is attached to each radiator. The protection is triggered when the radiators reach a temperature of 59.8 °C. A little earlier, at a temperature of 55.0 °C, a preliminary message about overheating appears on the display - the temperature of the radiators is automatically displayed. The amplifier restarts automatically when the radiators cool down to 35.0 °C. Turning on the radiators at higher temperatures is only possible manually.
To improve the cooling conditions of the elements inside the amplifier housing, a small-sized fan, which is located on the rear panel. A fan with a brushless DC motor with a rated supply voltage of 12 V is used, designed to cool the computer processor. Since the operation of the fan creates some noise, which can be noticeable during pauses, a rather complex control algorithm is used. When the radiator temperature is 45.0 °C, the fan starts working, and when the radiators cool down to 35.0 °C, the fan turns off. When the output power is less than 2 W, fan operation is prohibited so that its noise is not noticeable. To prevent periodic fan switching on and off when the output power fluctuates around the threshold value, the minimum fan switch-off time is software limited to 10 seconds. At a radiator temperature of 55.0 °C and above, the fan operates without switching off, since this temperature is close to the emergency temperature. If the fan turned on while the amplifier was operating, then when entering the “STANDBY” mode, if the temperature of the radiators is above 35.0 °C, the fan continues to operate even at zero output power. This allows the amplifier to cool down quickly.
Power supply failure protection also implemented entirely in software. The microcontroller, using an ADC, monitors the supply voltages of both channels of the amplifier. This voltage is supplied to the processor from the main boards through resistors R55, R56 (Fig. 8).
The main power sources are switched on in stages. This is necessary for the reason that the load of the rectifiers is completely discharged filter capacitors, and when turned on suddenly there will be a strong current surge. This surge is dangerous to the rectifier diodes and can cause fuses to blow. Therefore, when the amplifier is turned on, relay K2 closes first (Fig. 12), and the transformers are connected to the network through limiting resistors R1 and R2. At this time, the threshold for the measured supply voltages is set by software to ±38 V. If this voltage threshold is not reached within the set time, the switching process is interrupted. This may occur if the current consumed by the amplifier circuit is significantly increased (the amplifier is damaged). In this case, the “UF” power supply failure indication is turned on.
If the ±38 V threshold is reached, then relay K3 is activated (Fig. 12), which excludes resistors from the primary circuits of the main transformers. Then the threshold is reduced to ±20 V, and the microcontroller continues to monitor the supply voltages. If during operation of the amplifier the supply voltage drops below ±20 V, the protection is triggered and the amplifier is turned off. Reducing the threshold in normal operation is necessary so that when the supply voltage drops under load, the protection does not trigger falsely.
Schematic diagram processor boards shown in Fig. 9. The basis of the processor is a microcontroller U1 type AT89C51 from Atmel, which operates at a clock frequency of 12 MHz. To increase system reliability, the U2 supervisor is used, which has a built-in watchdog timer and power monitor. To reset the watchdog timer, a separate WD line is used, on which a periodic signal is generated by software. The program is constructed in such a way that this signal will be present only if the timer interrupt handler and the main program loop are executed. Otherwise, the watchdog timer will reset the microcontroller.
width=710>
The drawing does not fit on the page and is therefore compressed!
To view it in full, click .
Rice. 9. Schematic diagram of the processor board.
The display is connected to the processor using an 8-bit bus (sockets XP4 - XP6). To gate the registers of the display board, signals C0..C4 are used, which are generated by the address decoder U4. Register U3 is the low address byte latch, only bits A0, A1, A2 are used. The high byte of the address is not used at all, which frees up the P2 port for other purposes.
When you press the control buttons, sound signals are generated programmatically. For this, the BPR line is used, to which the transistor switch VT1 is connected, loaded onto the dynamic emitter HA1.
The main left and right channel boards are connected to the processor board using connectors XP1 and XP2, respectively. Through these connectors, the processor receives the status signals of the IOF overcurrent protection system and the DC protection at the output of the DCF amplifier. These signals are common to the left and right channels, and their combination is possible thanks to the outputs of the open collector protection circuit. The OKL and OKR protection system readiness signals are channel-separate so that the processor can identify the channel in which the protection circuit has been triggered. The ENB signal, which comes from the processor to the protection system, allows the load relay to turn on. This signal is common to the two channels, which automatically synchronizes the operation of the two relays.
The TRR and TRL lines are used to read the thermometers installed on the right and left channel radiators, respectively. The temperature measured by thermometers can be displayed on the display if the appropriate display mode is turned on. The maximum temperature value of the two is displayed for the left and right channels. The measured value is also used for software implementation of overheating protection.
Additionally, connectors XP1 and XP2 contain WUR, WIR, WUL and WIL signals, which are used by the output power measurement circuitry.
The processor board is powered from a standby source via the XP3 connector. 4 levels are used for power supply: ±15 V, +12 V and +5 V. The ±15 V levels are turned off when entering standby mode, and the remaining levels are always present. Consumption from the +5 V and +12 V levels in standby mode is minimized due to the software shutdown of the main consumers. In addition, through this connector, several control logical signals are sent to the standby power supply: PEN - controls the standby power supply, REX - turns on the external socket relay, RP1 and RP2 - turns on the main power supply relay, FAN - turns on the fan. The protection circuits located on the main boards are powered from the processor board at +12 V, and the display board is powered at +5 V.
To measure output power and monitor supply voltages, a 12-bit ADC U6 type AD7896 from Analog Devices is used. One ADC channel is not enough, so a U5 switch is used at the input (it would be even better to use an 8-channel ADC, for example, type AD7888). Data is read from the ADC in serial form. The SDATA (serial data) and SCLK (clock) lines are used for this purpose. The conversion process is started programmatically by the START signal. REF195 (U7) was used as a reference source and at the same time a voltage regulator for the ADC supply. Since the ±15 V supply voltage is turned off in standby mode, all logic signals are connected to the ADC through resistors R9 - R11, which limit possible current surges when switching to standby mode and back.
Of the eight inputs of the switch, six are used: two for measuring power, four for monitoring supply voltages. The desired channel is selected using address lines AX0, AX1, AX2.
Let's consider power measurement circuit left channel. The applied circuit provides multiplication of the load current and voltage, so the load impedance is automatically taken into account and the readings always correspond to the real active power in the load. Through resistor dividers R49 - R54 located on the main board (Fig. 5), the voltage from the current sensors (emitter resistors of the output transistors) is supplied to the differential amplifier U8A (Fig. 9), which produces the current signal. From the output U8A, through the tuning resistor R17, the signal is supplied to the Y input of the analog multiplier U9 type K525PS2. The voltage signal is simply removed from the divider and fed to the X input of the analog multiplier. At the output of the multiplier, a low-pass filter R18C13 is installed, which produces a signal proportional to the quasi-peak output power with an integration time of about 10 ms. This signal goes to one of the inputs of the switch, then to the ADC. Diode VD1 protects the switch input from negative voltage.
In order to compensate for the initial zero offset of the multipliers, when the amplifier is turned on (when the load relay is not yet turned on and the output power is zero), the zero auto-calibration process occurs. The measured offset voltage is subtracted from the ADC readings during further operation.
The power in the left and right channels is measured separately, and the maximum value for the channels is indicated. Since the indicator must display both quasi-peak and average output power, and the displayed values must be easy to understand, the values measured using an ADC are subject to software processing. The timing characteristics of a power level meter are characterized by integration time and flyback time. For a quasi-peak power meter, the integration time is set by the hardware filtering chain and is approximately 10 ms. The average power meter differs only in its increased integration time, which is implemented in software. When calculating the average power, a moving average of 256 points is used. The return time in both cases is set by software. For ease of reading, this time should be relatively long. In this case, the reverse motion of the indicator is realized by subtracting 1/16 of the current power code once every 20 ms. In addition, when indicating, peak values are held for 1.4 seconds. Since updating the indicator readings too frequently is not perceived well, the update occurs every 320 ms. In order not to miss the next peak and display it synchronously with the input signal, when a peak is detected, an extraordinary update of the readings occurs.
As mentioned above, the PA shares a common with pre-amplifier remote control, which operates in the RC-5 standard. The receiver of the SFH-506 type remote control system is located on the display board. From the output of the photodetector, the signal is sent to the SER (INT1) input of the microcontroller. Decoding the RC-5 code is done in software. The number of the system used is 0AH, the “STANDBY” button has the code 0CH, the “DISPLAY” button is 21H, the “MODE” button is 20H. If necessary, these codes can be easily changed, since a conversion table is used, which can be found at the end of the source text of the microcontroller program.
On display board(Fig. 10) two two-digit seven-segment indicators HG1 and HG2 of type LTD6610E are installed. They are controlled by parallel registers U1 – U4. Dynamic display is not used as this may cause increased noise levels.
width=710>
The drawing does not fit on the page and is therefore compressed!
To view it in full, click .
Rice. 10. Schematic diagram of the indication board.
Register U5 is used to control LEDs. A limiting resistor is connected in series with each segment and each LED. The OC inputs of all registers are combined and connected to the PEN signal of the microcontroller. During reset and register initialization, this signal is logic high. This prevents accidental lighting of the indication during transient processes.
The display board also contains control buttons SB1 – SB6. They are connected to the data bus lines and to the RET return line. Diodes VD1 – VD6 prevent short-circuiting of data lines when two or more buttons are pressed simultaneously. When scanning the keyboard, the microcontroller uses port P0 as a simple output port, generating a running zero on its lines. At the same time, the RET line is polled. This way the code of the pressed button is determined.
An integrated remote control photodetector U6 is installed next to the indicators under a common protective glass. The signal from the output of the photodetector through connector XP6 is supplied to the input of the microcontroller SER (INT1).
Duty source(Fig. 11) provides 4 output levels: +5 V, +12 V and ±15 V. The ±15 V levels are disabled in standby mode. The source uses a small toroidal transformer wound on a 50x20x25 mm core. The standby transformer has a large power reserve, and the number of turns per volt is selected greater than the calculated one. Thanks to these measures, the transformer practically does not heat up, which increases its reliability (after all, it must operate continuously throughout the entire service life of the amplifier). Winding data and wire diameter are indicated in the diagram. Voltage stabilizers have no special features. Stabilizer chips U1 and U2 are installed on a small common radiator. To turn off the ±15 V levels, switches are used on transistors VT1 - VT4, which are controlled by the PEN signal coming from the processor board.
Rice. 11. Schematic diagram of the standby power supply board.
In addition to voltage stabilizers, the board of the standby power supply contains switches on transistors VT5 - VT12 to control the relay and fan. Since the microcontrollers of the MCS-51 family have ports in a high logical level state during the “Reset” signal, all actuators must be turned on at a low level. Otherwise, there will be false alarms when the power is turned on or when the watchdog timer is triggered. For this reason, single npn transistors with OE or ULN2003 driver chips and the like cannot be used as keys.
Relays, fuses and limiting resistors are located on relay board(Fig. 12). All network wires are connected via screw terminal blocks. Each main transformer, standby transformer and external socket block have separate fuses. For safety reasons, external sockets are turned off by two groups of relay contacts K1, which break both wires. Main transformers are tapped from the middle of the primary winding. This tap can be used to provide 110 V to power other components in the complex. Devices that meet the American standard are somewhat cheaper than multi-system devices, which is why they are sometimes found on our territory. There are points on the relay board where 110V can be drawn, but the basic version does not use this voltage.
Rice. 12. Schematic diagram of the relay board.
Block connection diagram for amplifier chassis shown in Fig. 13. Bridge rectifiers assembled on diodes VD5 - VD12 type KD2997A are connected to the secondary windings of the main transformers T1 and T2. Filter capacitors with a total capacity of more than 100,000 μF are connected to the output of the rectifiers. This high capacitance is necessary to achieve low ripple and improve the amplifier's ability to reproduce pulsed signals. From the filter capacitors, a supply voltage of ±45 V is supplied to the main boards of the amplifier. Additionally, there are low-power rectifiers assembled on diodes VD1 - VD4, the output voltage of which is filtered with a relatively small time constant by capacitors C1 and C2. Through resistors R1 and R2, the output voltage of these auxiliary rectifiers is supplied to the protection circuits, which are assembled on the main boards of the amplifier. If several half-cycles of the mains voltage fail, the output voltage of the auxiliary rectifiers drops, which is detected by the protection circuits, and the load relays are switched off. At this time, the output voltage of the main rectifiers is still quite high due to the large capacitors, so the transient process in the amplifier does not begin with a connected load.
width=710>
The drawing does not fit on the page and is therefore compressed!
To view it in full, click .
Rice. 13. Connection diagram of amplifier blocks.
For power amplifier design and layout no less important than circuit design. The main problem is that the output transistors need to provide efficient heat dissipation. With the natural cooling method, this results in massive radiators, which become almost the main structural elements. The common arrangement, when the rear wall also serves as a radiator, is not suitable, since then there is no space left at the back to install the necessary terminals and connectors. Therefore, in the described PA, a layout with a side arrangement of radiators was chosen (Fig. 14):
Rice. 14. General layout of the amplifier.
The radiators are slightly raised (this is clearly visible in Fig. 4), which ensures better cooling. The main amplifier boards are fixed parallel to the radiators. This minimizes the length of the wires between the board and the output transistors. Another dimensional element of the amplifier is network transformers. In this case, two toroidal transformers are used, which are installed on top of each other in a common cylindrical screen. This screen occupies a significant part of the internal volume of the amplifier case. The main rectifiers are mounted on a common radiator, which is located vertically behind the transformer shield. The filter capacitors are located at the bottom of the amplifier chassis and are covered with a tray. The relay board is also located there. The standby power supply is mounted on a special bracket near the rear panel. The processor and display boards are placed in the thickness of the front panel, which has a box-shaped section.
When developing the design of the amplifier, much attention was paid to the manufacturability of the design and ease of access to any component. More details on the amplifier layout can be found in Fig. 15 and 18:
Rice. 15. Location of assembled amplifier components.
The basis of the amplifier housing is Aluminum alloy chassis D16T 4mm thick (4 in Fig. 18). Attached to the chassis radiators(1 in Fig. 18) which are milled from an aluminum plate or casting. The required radiator area greatly depends on the operating conditions of the amplifier, but it should not be less than 2000 cm 2 . To facilitate access to the amplifier boards, the radiators are fixed to the chassis using hinges (10 in Fig. 18), which allows the radiators to be tilted. To ensure that the wires of the input and output connectors do not interfere with this, the rear panel is divided into three parts (Fig. 4). The middle part is fixed to the chassis using a bracket, and the two side parts are fixed to the radiators. The connectors are installed on the sides of the panel, which fold down along with the radiators. Thus, the radiator assembly is a monophonic PA, which is connected only by power wires and a flat control cable. In Fig. 18, for clarity, the radiators are only partially folded back, and the rear panel is not disassembled.
Main amplifier boards They are also secured to the radiators using hinges (12 in Fig. 18), which allows them to be folded back to gain access to the solder side. The rotation axis of the board runs along the line of holes for connecting the wires of the output transistors. This made it possible to practically not increase the length of these wires while simultaneously being able to fold back the board. The upper mounting points for the boards are regular 15mm high threaded posts. Wiring of one-sided main boards of the left and right channels is completed mirrored(Fig. 16), which made it possible to optimize the connections. Naturally, the mirroring of the topology is not complete, since elements are used that cannot simply be arranged in a mirror manner (microcircuits and relays). The figure gives an approximate idea of the topology of the boards; the topology of all boards is available in the archive (see the Download section) in the form of files in PCAD 4.5 format.
width=710>
The drawing does not fit on the page and is therefore compressed!
To view it in full, click .
Rice. 16. Layout of the main boards of the amplifier.
Each radiator 1 (Fig. 17) has a smooth surface 2, which is processed after blackening. Nine transistors 4 are installed on it through ceramic gaskets 2.
Rice. 17. Radiator design:
Studies have shown that mica, and even more so modern elastic gaskets, do not have sufficient thermal conductivity. The best material for insulating gaskets is BeO-based ceramics. However, for transistors in plastic cases such gaskets are almost never found. Quite good results were obtained by making spacers from hybrid chip substrates. This is pink ceramics (unfortunately, the material is not exactly known, most likely something based on Al 2 O 3). To compare the thermal conductivity of different gaskets, a stand was assembled in which two identical transistors in a TO-220 housing were mounted on the radiator: one directly, the other through the gasket under study. The base current for both transistors was the same. The transistor on the gasket dissipated power of about 20 W, but the other transistor did not dissipate power (no voltage was supplied to the collector). The difference in the B-E drops of two transistors was measured, and from this difference the difference in junction temperatures was calculated. All gaskets used thermal paste, without it the results were worse and inconsistent. The comparison results are presented in the table:
The output transistors are pressed with pads 5, the remaining transistors are secured with screws. This is not very convenient, since it requires drilling ceramic gaskets, which can only be done using diamond drills, and even then with great difficulty.
A thermometer 9 is installed next to the transistors. As experience has shown, when attaching DS1820 thermometers, great pressure cannot be applied to their body, otherwise the readings will be distorted, and quite significantly (it is generally better to glue thermometers using glue that has high thermal conductivity).
A board 6 is attached to the radiator under the transistors. There are no conductors on the back side of this board, so it can be mounted directly on the surface of the radiator. The leads of all transistors are soldered to pads on the top side of the board. The connections between the board and the main board are made with short wires, which are soldered into hollow rivets 7. To prevent the rivets from shorting to the radiator, a recess 8 is made in it.
Basic toroidal transformers(7 in Fig. 18) are installed on each other through elastic gaskets. To reduce interference from transformers to other equipment (a cassette deck, for example), it is recommended to place the transformers in a screen made of annealed steel with a thickness of at least 1.5 mm. The screen consists of a steel cylinder and two covers held together with a pin. To avoid short-circuited turns, the top cover has a dielectric sleeve. However, if you intend to operate the PA at high average power, then you should provide ventilation holes in the screen or abandon the screen altogether. It would seem that to mutually compensate the stray fields of transformers, it is enough to simply turn on their primary windings out of phase. But in practice this measure is very ineffective. The stray field of a toroidal transformer, despite its apparent axial symmetry, has a very complex spatial distribution. Therefore, reversing the polarity of one of the primary windings leads to a weakening of the stray field at one point in space, but to an increase in another. In addition, the configuration of the stray field depends significantly on the transformer load.
Rice. 18. Main components of the amplifier:
| 1 - radiators | 12 - board fastening loop |
| 2 - main amplifier boards | 13 - board mounting stand |
| 3 - platform on the radiator for installing transistors | 14 - control cable connector (from the processor board) |
| 4 - load-bearing plate | 15 - wire from additional output. rectifier |
| 5 - front panel support plate | 16 - duty transformer in the screen |
| 6 - box-section front panel | 17 - standby power supply board |
| 7 - main transformers in the screen | 18 - radiator for voltage stabilizers |
| 8 - rectifier diode radiator | 19 - relay block control wires |
| 9 - power supply to the boards | 20 - rear panel |
| 10 - mounting radiators on hinges | 21 - output terminals |
| 11 - radiator mounting bracket | 22 - input connectors |
Very stringent requirements are imposed on the PA power transformer. This is due to the fact that it is loaded onto a rectifier with very large filter capacitors. This leads to the fact that the current consumed from the secondary winding of the transformer is pulsed in nature, and the value of the current in the pulse is many times greater than the average current consumed. To keep transformer losses low, the windings must have very low resistance. In other words, the transformer must be designed to handle significantly more power than is averagely drawn from it. The described amplifier uses two toroidal transformers, each of which is wound on a 110x60x40 mm core made of E-380 steel tape. The primary windings contain 2x440
UMZCH VV with microcontroller control system
Views today: 32347, total: 32347
UMZCH VVS-2011 Ultimate version
Amplifier specifications:
1. Large power: 150W/8ohm,
2. High linearity - 0.000.2...0.000.3% at 20 kHz 100 W / 4 Ohm,
Full set of service units:
1. Maintain zero constant voltage,
2. Compensator for resistance of AC wires,
3. Current protection,
4. DC output voltage protection,
5. Smooth start.
UMZCH VVS2011 scheme
The layout of printed circuit boards was carried out by a participant in many popular projects LepekhinV (Vladimir Lepekhin). It turned out very well).
UMZCH-VVS2011 board
ULF amplifier board VVS-2011 was designed for tunnel ventilation (parallel to the radiator). Installation of transistors UN (voltage amplifier) and VK (output stage) is somewhat difficult, because installation/disassembly has to be done with a screwdriver through holes in the PP with a diameter of about 6 mm. When access is open, the projection of the transistors does not fall under the PP, which is much more convenient. I had to modify the board a little.
I didn’t take one point into account in the new software— this is the convenience of setting up protection on the amplifier board:
C25 0.1n, R42* 820 Ohm and R41 1k all elements are SMD and are located on the solder side, which is not very convenient when setting up, because You will need to unscrew and tighten the bolts securing the PCB to the stands and the transistors to the radiators several times. Offer: R42* 820 consists of two SMD resistors located in parallel, from here the proposal: we solder one SMD resistor immediately, we solder the other output resistor overhang to VT10, one output to the base, the other to the emitter, we select it to the appropriate one. Selected, change output to smd, for clarity:
UMZCH VVS-2011 Ultimate version
Amplifier specifications:
High power: 150W/8ohm
High linearity: 0.0002 – 0.0003% (at 20 kHz 100 W / 4 ohms)
Full set of service units:
Maintain zero constant voltage
AC wire resistance compensator
Current protection
Output DC voltage protection
Smooth start
Electrical diagram
The layout of printed circuit boards was carried out by a participant in many popular projects LepekhinV (Vladimir Lepekhin). It turned out very well).
VVS-2011 amplifier board
Start-protective device
AC amplifier protection board VVS-2011
The VHF VVS-2011 amplifier board was designed for tunnel ventilation (parallel to the radiator). Installation of transistors UN (voltage amplifier) and VK (output stage) is somewhat difficult, because installation/disassembly has to be done with a screwdriver through holes in the PP with a diameter of about 6 mm. When access is open, the projection of the transistors does not fall under the PP, which is much more convenient. I had to modify the board a little.
Amplifier board
VVS-2011 amplifier wiring diagram
One thing I didn’t take into account in the new PCBs is the ease of setting up protection on the amplifier board
C25 = 0.1 nF, R42* = 820 Ohm and R41 = 1 kOhm. All SMD elements are located on the solder side, which is very inconvenient when setting up, because You will need to unscrew and tighten the bolts securing the PCB to the stands and the transistors to the radiators several times.
Offer: R42* 820 Ohm consists of two SMD resistors located in parallel, from here the proposal: we solder one SMD resistor immediately, we solder the other output resistor overhang to VT10, one output to the base, the other to the emitter, we select the appropriate one. We picked it up and changed the output to SMD, for clarity.
UMZCH BB-2010 is a new development from the well-known line of UMZCH BB (high fidelity) amplifiers. A number of technical solutions used were influenced by Ageev’s work.
Specifications:
Harmonic distortion at 20000 Hz: 0.001% (150 W/8 ohms)
Small signal bandwidth -3 dB: 0 – 800000 Hz
Output voltage slew rate: 100 V/µs
Signal-to-noise and signal-to-background ratio: 120 dB
Electrical diagram of VVS-2010
Thanks to the use of an op-amp operating in a lightweight mode, as well as the use in the voltage amplifier of only cascades with OK and OB, covered by deep local OOS, the UMZCH BB is characterized by high linearity even before the general OOS is covered. In the very first high-fidelity amplifier back in 1985, solutions were used that until then were used only in measuring technology: DC modes are supported by a separate service unit, to reduce the level of interface distortion, the transition resistance of the contact group of the AC switching relay is covered by a common negative feedback, and a special unit effectively compensates for the influence of the resistance of speaker cables on these distortions. The tradition has been preserved in the UMZCH BB-2010, however, the general OOS also covers the resistance of the output low-pass filter.
In the vast majority of designs of other UMZCHs, both professional and amateur, many of these solutions are still missing. At the same time, high technical characteristics and audiophile advantages of the UMZCH BB are achieved by simple circuit solutions and a minimum of active elements. In fact, this is a relatively simple amplifier: one channel can be assembled in a couple of days without haste, and the setup only involves setting the required quiescent current of the output transistors. Especially for novice radio amateurs, a method of node-by-node, cascade testing and adjustment has been developed, using which you can be guaranteed to localize possible errors and prevent their possible consequences even before the UMZCH is fully assembled. All possible questions about this or similar amplifiers have detailed explanations, both on paper and on the Internet.
At the input of the amplifier there is a high-pass filter R1C1 with a cutoff frequency of 1.6 Hz, Fig. 1. But the efficiency of the mode stabilization device allows the amplifier to work with an input signal containing up to 400 mV of DC component voltage. Therefore, C1 is excluded, which realizes the eternal audiophile dream of a path without capacitors and significantly improves the sound of the amplifier.
The capacitance of capacitor C2 of the input low-pass filter R2C2 is selected so that the cutoff frequency of the input low-pass filter, taking into account the output resistance of the preamplifier 500 Ohm -1 kOhm, is in the range from 120 to 200 kHz. At the input of op amp DA1 there is a frequency correction circuit R3R5C3, which limits the band of processed harmonics and interference coming through the OOS circuit from the output side of the UMZCH, with a band of 215 kHz at a level of -3 dB and increases the stability of the amplifier. This circuit allows you to reduce the difference signal above the cutoff frequency of the circuit and thereby eliminate unnecessary overload of the voltage amplifier with high-frequency interference signals, interference and harmonics, eliminating the possibility of dynamic intermodulation distortion (TIM; DIM).
Next, the signal is fed to the input of a low-noise operational amplifier with field-effect transistors at the DA1 input. Many “claims” to the UMZCH BB are made by opponents regarding the use of an op-amp at the input, which supposedly worsens the sound quality and “steals the virtual depth” of the sound. In this regard, it is necessary to pay attention to some quite obvious features of the operation of the op amp in the UMZCH VV.
Operational amplifiers of pre-amplifiers, post-DAC op-amps are forced to develop several volts of output voltage. Since the gain of the op amp is small and ranges from 500 to 2000 times at 20 kHz, this indicates their operation with a relatively high voltage difference signal - from several hundred microvolts at LF to several millivolts at 20 kHz and a high probability of intermodulation distortion being introduced by the input stage of the op amp. The output voltage of these op-amps is equal to the output voltage of the last voltage amplification stage, usually performed according to a circuit with an OE. An output voltage of several volts indicates that this stage operates with fairly large input and output voltages, and as a result, it introduces distortion into the amplified signal. The op-amp is loaded by the resistance of the parallel-connected OOS and load circuits, sometimes amounting to several kilo-ohms, which requires up to several milliamps of output current from the output repeater of the amplifier. Therefore, changes in the current of the output repeater of the IC, the output stages of which consume a current of no more than 2 mA, are quite significant, which also indicates that they introduce distortions into the amplified signal. We see that the input stage, voltage amplification stage and op-amp output stage can introduce distortion.
But the circuit design of the high-fidelity amplifier, due to the high gain and input resistance of the transistor part of the voltage amplifier, provides very gentle operating conditions for op-amp DA1. Judge for yourself. Even in a UMZCH that has developed a nominal output voltage of 50 V, the input differential stage of the op-amp operates with difference signals with voltages from 12 μV at frequencies of 500 Hz to 500 μV at a frequency of 20 kHz. The ratio of the high input overload capacity of the differential stage, made on field-effect transistors, and the scanty voltage of the difference signal ensures high linearity of signal amplification. The output voltage of the op-amp does not exceed 300 mV. which indicates the low input voltage of the voltage amplification stage with a common emitter from the operational amplifier - up to 60 μV - and the linear mode of its operation. The output stage of the op-amp supplies an alternating current of no more than 3 µA to the load of about 100 kOhm from the VT2 base side. Consequently, the output stage of the op-amp also operates in an extremely light mode, almost at idle. On a real musical signal, voltages and currents are most of the time an order of magnitude less than the given values.
From a comparison of the voltages of the difference and output signals, as well as the load current, it is clear that in general the operational amplifier in the UMZCH BB operates in a hundreds of times lighter, and therefore linear, mode than the op-amp mode of preamplifiers and post-DAC op-amps of CD players that serve as sources signal for UMZCH with any depth of environmental protection, as well as without it at all. Consequently, the same op-amp will introduce much less distortion in the UMZCH BB than in a single connection.
Occasionally there is an opinion that the distortions introduced by the cascade ambiguously depend on the voltage of the input signal. This is mistake. The dependence of the manifestation of cascade nonlinearity on the voltage of the input signal may obey one or another law, but it is always unambiguous: an increase in this voltage never leads to a decrease in the introduced distortions, but only to an increase.
It is known that the level of distortion products at a given frequency decreases in proportion to the depth of negative feedback for this frequency. The open-circuit gain, before the amplifier reaches the OOS, at low frequencies cannot be measured due to the smallness of the input signal. According to calculations, the open-circuit gain developed to cover the negative feedback allows one to achieve a negative feedback depth of 104 dB at frequencies up to 500 Hz. Measurements for frequencies starting from 10 kHz show that the OOS depth at a frequency of 10 kHz reaches 80 dB, at a frequency of 20 kHz - 72 dB, at a frequency of 50 kHz - 62 dB and 40 dB - at a frequency of 200 kHz. Figure 2 shows the amplitude-frequency characteristics of the UMZCH VV-2010 and, for comparison, similar in complexity.
High gain up to OOS coverage is the main feature of the circuit design of BB amplifiers. Since the goal of all circuit tricks is to achieve high linearity and high gain to maintain deep OOS in the widest possible frequency band, this means that such structures are the only circuit methods for improving amplifier parameters. Further reduction in distortion can only be achieved by design measures aimed at reducing the interference of harmonics of the output stage on the input circuits, especially on the inverting input circuit, from which the gain is maximum.
Another feature of the UMZCH BB circuitry is the current control of the output stage of the voltage amplifier. The input op-amp controls the voltage-current conversion stage, made with OK and OB, and the resulting current is subtracted from the quiescent current of the stage, made according to the circuit with OB.
The use of a linearizing resistor R17 with a resistance of 1 kOhm in the differential stage VT1, VT2 on transistors of different structures with serial power increases the linearity of the conversion of the output voltage of the op-amp DA1 to the collector current VT2 by creating a local feedback loop with a depth of 40 dB. This can be seen from comparing the sum of the emitters' own resistances VT1, VT2 - approximately 5 Ohms each - with resistance R17, or the sum of thermal voltages VT1, VT2 - about 50 mV - with the voltage drop across resistance R17 amounting to 5.2 - 5.6 V .
For amplifiers built using the circuit design under consideration, a sharp, 40 dB per decade of frequency, decrease in gain above a frequency of 13...16 kHz is observed. The error signal, which is a product of distortion, at frequencies above 20 kHz is two to three orders of magnitude less than the useful audio signal. This makes it possible to convert the linearity of the differential stage VT1, VT2, which is excessive at these frequencies, into increasing the gain of the transistor part of the UN. Due to minor changes in the current of the differential cascade VT1, VT2, when amplifying weak signals, its linearity with a decrease in the depth of local feedback does not deteriorate significantly, but the operation of the op-amp DA1, on the operating mode of which at these frequencies the linearity of the entire amplifier depends, will make the gain margin easier, since all voltages, The distortions that determine the operational amplifier's distortion, starting from the difference signal to the output signal, decrease in proportion to the gain in gain at a given frequency.
The phase lead correction circuits R18C13 and R19C16 were optimized in the simulator to reduce the op amp differential voltage to frequencies of several megahertz. It was possible to increase the gain of the UMZCH VV-2010 compared to the UMZCH VV-2008 at frequencies of the order of several hundred kilohertz. The gain in gain was 4 dB at 200 kHz, 6 at 300 kHz, 8.6 at 500 kHz, 10.5 dB at 800 kHz, 11 dB at 1 MHz and from 10 to 12 dB at frequencies higher 2 MHz. This can be seen from the simulation results, Fig. 3, where the lower curve refers to the frequency response of the advance correction circuit of the UMZCH VV-2008, and the upper curve refers to the UMZCH VV-2010.
VD7 protects the emitter junction VT1 from reverse voltage arising due to the flow of recharging currents C13, C16 in the mode of limiting the output signal of the UMZCH by voltage and the resulting maximum voltages with a high rate of change at the output of the op-amp DA1.
The output stage of the voltage amplifier is made of transistor VT3, connected according to a common base circuit, which eliminates the penetration of the signal from the output circuits of the cascade into the input circuits and increases its stability. The OB cascade, loaded onto the current generator on transistor VT5 and the input resistance of the output stage, develops a high stable gain - up to 13,000...15,000 times. Choosing the resistance of resistor R24 to be half the resistance of resistor R26 guarantees equality of the quiescent currents VT1, VT2 and VT3, VT5. R24, R26 provide local feedback that reduces the Early effect - the change in p21e depending on the collector voltage and increases the initial linearity of the amplifier by 40 dB and 46 dB, respectively. Powering the UN with a separate voltage, modulo 15 V higher than the voltage of the output stages, makes it possible to eliminate the effect of quasi-saturation of transistors VT3, VT5, which manifests itself in a decrease in p21e when the collector-base voltage decreases below 7 V.
The three-stage output follower is assembled using bipolar transistors and does not require any special comments. Don't try to fight entropy by skimping on the quiescent current of the output transistors. It should not be less than 250 mA; in the author's version - 320 mA.
Before the activation relay AC K1 is activated, the amplifier is covered by OOS1, realized by switching on the divider R6R4. The accuracy of maintaining the resistance R6 and the consistency of these resistances in different channels is not essential, but to maintain the stability of the amplifier it is important that the resistance R6 is not much lower than the sum of the resistances R8 and R70. When relay K1 is triggered, OOS1 is turned off and the OOS2 circuit, formed by R8R70C44 and R4, and covering contact group K1.1, comes into operation, where R70C44 excludes the output low-pass filter R71L1 R72C47 from the OOS circuit at frequencies above 33 kHz. The frequency-dependent OOS R7C10 forms a roll-off in the frequency response of the UMZCH to the output low-pass filter at a frequency of 800 kHz at a level of -3 dB and provides a margin in the OOS depth above this frequency. The decrease in frequency response at the AC terminals above the frequency of 280 kHz at a level of -3 dB is ensured by the combined action of R7C10 and the output low-pass filter R71L1 -R72C47.
The resonant properties of loudspeakers lead to the emission by the diffuser of damped sound vibrations, overtones after pulse action and the generation of its own voltage when the turns of the loudspeaker coil cross the magnetic field lines in the gap of the magnetic system. The damping coefficient shows how large the amplitude of the diffuser's oscillations is and how quickly they attenuate when the AC load is applied as a generator to the full impedance of the UMZCH. This coefficient is equal to the ratio of the AC resistance to the sum of the output resistance of the UMZCH, the transition resistance of the contact group of the AC switching relay, the resistance of the output low-pass filter inductor usually wound with a wire of insufficient diameter, the transition resistance of the AC cable terminals and the resistance of the AC cables themselves.
In addition, the impedance of loudspeaker systems is nonlinear. The flow of distorted currents through the conductors of AC cables creates a voltage drop with a large proportion of harmonic distortion, which is also subtracted from the undistorted output voltage of the amplifier. Therefore, the signal at the AC terminals is distorted much more than at the output of the UMZCH. These are so-called interface distortions.
To reduce these distortions, compensation of all components of the amplifier's output impedance is applied. The UMZCH's own output resistance, together with the transition resistance of the relay contacts and the resistance of the inductor wire of the output low-pass filter, is reduced by the action of a deep general negative feedback taken from the right terminal of L1. In addition, by connecting the right terminal of R70 to the “hot” AC terminal, you can easily compensate for the transition resistance of the AC cable clamp and the resistance of one of the AC wires, without fear of generating UMZCH due to phase shifts in the wires covered by the OOS.
The AC wire resistance compensation unit is made in the form of an inverting amplifier with Ky = -2 on op-amps DA2, R10, C4, R11 and R9. The input voltage for this amplifier is the voltage drop across the “cold” (“ground”) speaker wire. Since its resistance is equal to the resistance of the “hot” wire of the AC cable, to compensate for the resistance of both wires it is enough to double the voltage on the “cold” wire, invert it and, through resistor R9 with a resistance equal to the sum of the resistances R8 and R70 of the OOS circuit, apply it to the inverting input of the op-amp DA1 . Then the output voltage of the UMZCH will increase by the sum of the voltage drops on the speaker wires, which is equivalent to eliminating the influence of their resistance on the damping coefficient and the level of interface distortion at the speaker terminals. Compensation for the drop in the AC wire resistance of the nonlinear component of the back-EMF of loudspeakers is especially necessary at the lower frequencies of the audio range. The signal voltage at the tweeter is limited by the resistor and capacitor connected in series with it. Their complex resistance is much greater than the resistance of the speaker cable wires, so compensating for this resistance at HF makes no sense. Based on this, the integrating circuit R11C4 limits the operating frequency band of the compensator to 22 kHz.
Of particular note: the resistance of the “hot” wire of the AC cable can be compensated by covering its general OOS by connecting the right terminal of R70 with a special wire to the “hot” AC terminal. In this case, only the resistance of the “cold” AC wire will need to be compensated and the gain of the wire resistance compensator must be reduced to the value Ku = -1 by choosing the resistance of resistor R10 equal to the resistance of resistor R11.
The current protection unit prevents damage to the output transistors during short circuits in the load. The current sensor is resistors R53 - R56 and R57 - R60, which is quite enough. The flow of amplifier output current through these resistors creates a voltage drop that is applied to the divider R41R42. A voltage with a value greater than the threshold opens transistor VT10, and its collector current opens VT8 of the trigger cell VT8VT9. This cell enters a stable state with the transistors open and bypasses the HL1VD8 circuit, reducing the current through the zener diode to zero and locking VT3. Discharging C21 with a small current from the VT3 base may take several milliseconds. After the trigger cell is triggered, the voltage on the lower plate of C23, charged by the voltage on the LED HL1 to 1.6 V, increases from the level of -7.2 V from the positive power supply bus to the level of -1.2 B1, the voltage on the upper plate of this capacitor also increases by 5 V. C21 is quickly discharged through resistor R30 to C23, transistor VT3 is turned off. In the meantime, VT6 opens and through R33, R36 opens VT7. VT7 bypasses the zener diode VD9, discharges capacitor C22 through R31 and turns off transistor VT5. Without receiving bias voltage, the output stage transistors are also turned off.
Restoring the initial state of the trigger and turning on the UMZCH is done by pressing the SA1 “Protection Reset” button. C27 is charged by the collector current of VT9 and bypasses the base circuit of VT8, locking the trigger cell. If by this moment the emergency situation has been eliminated and VT10 is locked, the cell goes into a state with stable closed transistors. VT6, VT7 are closed, the reference voltage is supplied to the bases VT3, VT5 and the amplifier enters operating mode. If the short circuit in the UMZCH load continues, the protection is triggered again, even if capacitor C27 is connected to SA1. The protection works so effectively that during work on setting up the correction, the amplifier was de-energized several times for small soldering by touching the non-inverting input. The resulting self-excitation led to an increase in the current of the output transistors, and the protection turned off the amplifier. Although this crude method cannot be suggested as a general rule, but due to the current protection, it did not cause any harm to the output transistors.
Operation of the AC cable resistance compensator
The efficiency of the UMZCH BB-2008 compensator was tested using the old audiophile method, by ear, by switching the compensator input between the compensating wire and the common wire of the amplifier. The improvement in sound was clearly noticeable, and the future owner was eager to get an amplifier, so measurements of the influence of the compensator were not carried out. The advantages of the “cable cleaning” circuit were so obvious that the “compensator + integrator” configuration was adopted as a standard unit for installation in all developed amplifiers.
It's surprising how much unnecessary debate has flared up on the Internet regarding the usefulness/uselessness of cable resistance compensation. As usual, those who especially insisted on listening to a nonlinear signal were those to whom the extremely simple cable cleaning scheme seemed complex and incomprehensible, the costs for it exorbitant, and the installation labor-intensive ©. There were even suggestions that since so much money is spent on the amplifier itself, it would be a sin to skimp on the sacred, but one should take the best, glamorous path that all civilized humanity follows and...purchase normal, human © super-expensive cables made of precious metals. To my great surprise, fuel was added to the fire by statements from highly respected specialists about the uselessness of the compensation unit at home, including those specialists who successfully use this unit in their amplifiers. It is very unfortunate that many fellow radio amateurs were distrustful of reports of improved sound quality in the low and midrange with the inclusion of a compensator, and did their best to avoid this simple way of improving the performance of the UMZCH, thereby robbing themselves.
Little research has been done to document the truth. From the GZ-118 generator, a number of frequencies were supplied to the UMZCH BB-2010 in the region of the resonant frequency of the AC, the voltage was controlled by an oscilloscope S1-117, and Kr at the AC terminals was measured by the INI S6-8, Fig. 4. Checking the effectiveness of wire resistanceResistor R1 is installed to avoid interference to the compensator input when switching it between the control and common wires. In the experiment, common and publicly available AC cables with a length of 3 m and a core cross-section of 6 square meters were used. mm, as well as the GIGA FS Il speaker system with a frequency range of 25-22000 Hz, a nominal impedance of 8 Ohms and a nominal power of 90 W from Acoustic Kingdom.
Unfortunately, the circuit design of harmonic signal amplifiers from C6-8 involves the use of high-capacity oxide capacitors in OOS circuits. This causes the low-frequency noise of these capacitors to affect the device's low-frequency resolution, causing its low-frequency resolution to deteriorate. When measuring a Kr signal with a frequency of 25 Hz from GZ-118 directly from C6-8, the instrument readings dance around the value of 0.02%. It is not possible to bypass this limitation using the notch filter of the GZ-118 generator in the case of measuring the efficiency of the compensator, because a number of discrete values of 2T filter tuning frequencies are limited at low frequencies to 20, 60, 120, 200 Hz and do not allow measuring Kr at the frequencies of interest to us. Therefore, reluctantly, the level of 0.02% was accepted as zero, the reference.
At a frequency of 20 Hz with a voltage at the AC terminals of 3 Vamp, which corresponds to an output power of 0.56 W into an 8 Ohm load, Kr was 0.02% with the compensator turned on and 0.06% with it turned off. At a voltage of 10 V ampl, which corresponds to an output power of 6.25 W, the Kr value is 0.02% and 0.08%, respectively, at a voltage of 20 V ampl and a power of 25 W - 0.016% and 0.11%, and at a voltage of 30 In amplitude and power 56 W - 0.02% and 0.13%.
Knowing the relaxed attitude of manufacturers of imported equipment to the meaning of inscriptions regarding power, and also remembering the wonderful, after the adoption of Western standards, the transformation of an acoustic system with a low-frequency loudspeaker power of 30 W into , long-term power of more than 56 W was not supplied to AC.
At a frequency of 25 Hz at a power of 25 W, Kr was 0.02% and 0.12% with the compensation unit on/off, and at a power of 56 W - 0.02% and 0.15%.
At the same time, the necessity and effectiveness of covering the output low-pass filter with a general OOS was tested. At a frequency of 25 Hz with a power of 56 W and connected in series to one of the AC cable wires of the output RL-RC low-pass filter, similar to that installed in an ultra-linear UMZCH, Kr with the compensator turned off reaches 0.18%. At a frequency of 30 Hz at a power of 56 W Kr 0.02% and 0.06% with the compensation unit on/off. At a frequency of 35 Hz at a power of 56 W Kr 0.02% and 0.04% with the compensation unit on/off. At frequencies of 40 and 90 Hz at a power of 56 W, Kr is 0.02% and 0.04% with the compensation unit on/off, and at a frequency of 60 Hz -0.02% and 0.06%.
The conclusions are obvious. The presence of nonlinear signal distortions at the AC terminals is observed. A deterioration in the linearity of the signal at the AC terminals is clearly detected when it is connected through the uncompensated, not covered by the OOS resistance of the low-pass filter containing 70 cm of relatively thin wire. The dependence of the distortion level on the power supplied to the AC suggests that it depends on the ratio of the signal power and the rated power of the AC woofers. Distortion is most pronounced at frequencies near the resonant one. The back EMF generated by the speakers in response to the influence of an audio signal is shunted by the sum of the output resistance of the UMZCH and the resistance of the AC cable wires, so the level of distortion at the AC terminals directly depends on the resistance of these wires and the output resistance of the amplifier.
The cone of a poorly damped low-frequency loudspeaker itself emits overtones, and, in addition, this loudspeaker generates a wide tail of non-linear and intermodulation distortion products that the mid-frequency loudspeaker reproduces. This explains the deterioration of sound at mid frequencies.
Despite the assumption of a zero Kr level of 0.02% adopted due to the imperfection of the INI, the influence of the cable resistance compensator on the AC signal distortion is clearly and unambiguously noted. It can be stated that there is complete agreement between the conclusions drawn after listening to the operation of the compensation unit on a musical signal and the results of instrumental measurements.
The improvement clearly audible when the cable cleaner is turned on can be explained by the fact that with the disappearance of distortion at the AC terminals, the midrange speaker stops producing all that dirt. Apparently, therefore, by reducing or eliminating the reproduction of distortions by the mid-frequency loudspeaker, the two-cable speaker circuit, the so-called. “Bi-wiring,” when the LF and MF-HF sections are connected with different cables, has an advantage in sound compared to a single-cable circuit. However, since in a two-cable circuit the distorted signal at the terminals of the AC low-frequency section does not disappear anywhere, this circuit is inferior to the version with a compensator in terms of the damping coefficient of free vibrations of the low-frequency loudspeaker cone.
You can’t fool physics, and for decent sound it’s not enough to get brilliant performance at the amplifier output with an active load, but you also need to not lose linearity after delivering the signal to the speaker terminals. As part of a good amplifier, a compensator made according to one scheme or another is absolutely necessary.
Integrator
The efficiency and error reduction capabilities of the integrator on DA3 were also tested. In the UMZCH BB with op-amp TL071, the output DC voltage is in the range of 6...9 mV and it was not possible to reduce this voltage by including an additional resistor in the non-inverting input circuit.
The effect of low-frequency noise, characteristic of an op-amp with a DC input, due to the coverage of deep feedback through the frequency-dependent circuit R16R13C5C6, manifests itself in the form of instability of the output voltage of several millivolts, or -60 dB relative to the output voltage at rated output power, at frequencies below 1 Hz , non-reproducible speakers.
The Internet mentioned the low resistance of the protective diodes VD1...VD4, which supposedly introduces an error into the operation of the integrator due to the formation of a divider (R16+R13)/R VD2|VD4.. To check the reverse resistance of the protective diodes, a circuit was assembled in Fig. 6. Here op-amp DA1, connected according to the inverting amplifier circuit, is covered by OOS through R2, its output voltage is proportional to the current in the circuit of the tested diode VD2 and the protective resistor R2 with a coefficient of 1 mV/nA, and the resistance of the circuit R2VD2 - with a coefficient of 1 mV/15 GOhm . To exclude the influence of additive errors of the op-amp - bias voltage and input current on the results of measuring the diode leakage current, it is necessary to calculate only the difference between the intrinsic voltage at the output of the op-amp, measured without the diode being tested, and the voltage at the output of the op-amp after its installation. In practice, a difference in the op-amp output voltages of several millivolts gives a diode reverse resistance value of the order of ten to fifteen gigaohms at a reverse voltage of 15 V. Obviously, the leakage current will not increase as the voltage on the diode decreases to a level of several millivolts, characteristic of the difference voltage of the op-amp integrator and compensator .
But the photoelectric effect characteristic of diodes placed in a glass case actually leads to a significant change in the output voltage of the UMZCH. When illuminated with a 60 W incandescent lamp from a distance of 20 cm, the constant voltage at the output of the UMZCH increased to 20...3O mV. Although it is unlikely that a similar level of illumination could be observed inside the amplifier case, a drop of paint applied to these diodes eliminated the dependence of the UMZCH modes on illumination. According to the simulation results, the decrease in the frequency response of the UMZCH is not observed even at a frequency of 1 millihertz. But the time constant R16R13C5C6 should not be reduced. The phases of the alternating voltages at the outputs of the integrator and compensator are opposite, and with a decrease in the capacitance of the capacitors or the resistance of the integrator resistors, an increase in its output voltage can worsen the compensation of the resistance of the speaker cables.
Comparison of the sound of amplifiers. The sound of the assembled amplifier was compared with the sound of several industrially produced foreign amplifiers. The source was a CD player from Cambridge Audio; a pre-amplifier was used to drive and adjust the sound level of the final UMZCHs; the Sugden A21a and NAD C352 used standard adjustment controls.
The first to be tested was the legendary, shocking and damn expensive English UMZCH “Sugden A21a”, operating in class A with an output power of 25 W. What is noteworthy is that in the accompanying documentation for the VX, the British considered it better not to indicate the level of nonlinear distortions. They say it’s not a matter of distortion, but of spirituality. “Sugden A21a>” lost to the UMZCH BB-2010 with comparable power both in level and in clarity, confidence, and noble sound at low frequencies. This is not surprising, given the features of its circuit design: just a two-stage quasi-symmetric output follower on transistors of the same structure, assembled according to the circuit design of the 70s of the last century with a relatively high output resistance and an electrolytic capacitor connected at the output, which further increases the total output resistance - this is the latter the solution itself worsens the sound of any amplifiers at low and mid frequencies. At medium and high frequencies, the UMZCH BB showed higher detail, transparency and excellent scene elaboration, when singers and instruments could be clearly localized by sound. By the way, speaking of the correlation of objective measurement data and subjective impressions of sound: in one of the journal articles of Sugden’s competitors, its Kr was determined at the level of 0.03% at a frequency of 10 kHz.
The next one was also the English amplifier NAD C352. The general impression was the same: the pronounced “bucket” sound of the Englishman at low frequencies left him no chance, while the work of the UMZCH BB was recognized as impeccable. Unlike NADA, the sound of which was associated with dense bushes, wool, and cotton wool, the sound of BB-2010 at medium and high frequencies made it possible to clearly distinguish the voices of performers in a general choir and instruments in an orchestra. The work of the NAD C352 clearly expressed the effect of better audibility of a more vocal performer, a louder instrument. As the owner of the amplifier himself put it, in the sound of the UMZCH BB the vocalists did not “scream and nod” at each other, and the violin did not fight in sound power with the guitar or trumpet, but all the instruments were peacefully and harmoniously “friends” in the overall sound image of the melody. At high frequencies, the UMZCH BB-2010, according to imaginative audiophiles, sounds “as if it were painting the sound with a thin, thin brush.” These effects can be attributed to differences in intermodulation distortion between the amplifiers.
The sound of the Rotel RB 981 UMZCH was similar to the sound of the NAD C352, with the exception of better performance at low frequencies, yet the BB-2010 UMZCH remained unrivaled in the clarity of AC control at low frequencies, as well as the transparency and delicacy of sound at mid and high frequencies.
The most interesting thing in terms of understanding the way of thinking of audiophiles was the general opinion that, despite their superiority over these three UMZCHs, they bring “warmth” to the sound, which makes it more pleasant, and the BB UMZCH works smoothly, “it is neutral to the sound.”
The Japanese Dual CV1460 lost its sound immediately after switching on in the most obvious way for everyone, and we didn’t waste time listening to it in detail. Its Kr was in the range of 0.04...0.07% at low power.
The main impressions from comparing the amplifiers were completely identical in their main features: the UMZCH BB was unconditionally and unequivocally ahead of them in sound. Therefore, further testing was deemed unnecessary. In the end, friendship won, everyone got what they wanted: for a warm, soulful sound - Sugden, NAD and Rotel, and to hear what was recorded on disk by the director - UMZCH BB-2010.
Personally, I like the high-fidelity UMZCH for its light, clean, impeccable, noble sound; it effortlessly reproduces passages of any complexity. As a friend of mine, an experienced audiophile, put it, he handles the sounds of drum kits at low frequencies without variations, like a press, at medium frequencies he sounds as if there is none, and at high frequencies he seems to be painting the sound with a thin brush. For me, the non-straining sound of the UMZCH BB is associated with the ease of operation of the cascades.
Victor Zhukovsky, Krasnoarmeysk, Donetsk region.
UMZCH BB-2010 is a new development from the well-known line of UMZCH BB (high fidelity) amplifiers [1; 2; 5]. A number of technical solutions used were influenced by the work of SI Ageev. .
The amplifier provides Kr of the order of 0.001% at a frequency of 20 kHz at Pout = 150 W into an 8 Ohm load, small signal frequency band at a level of -3 dB - 0 Hz ... 800 kHz, slew rate of the output voltage -100 V / µs, signal-to-noise ratio and signal/background -120 dB.
Thanks to the use of an op-amp operating in a lightweight mode, as well as the use in the voltage amplifier of only cascades with OK and OB, covered by deep local OOS, the UMZCH BB is characterized by high linearity even before the general OOS is covered. In the very first high-fidelity amplifier back in 1985, solutions were used that until then were used only in measuring technology: DC modes are supported by a separate service unit, to reduce the level of interface distortion, the transition resistance of the contact group of the AC switching relay is covered by a common negative feedback, and a special unit effectively compensates for the influence of the resistance of speaker cables on these distortions. The tradition has been preserved in the UMZCH BB-2010, however, the general OOS also covers the resistance of the output low-pass filter.
In the vast majority of designs of other UMZCHs, both professional and amateur, many of these solutions are still missing. At the same time, high technical characteristics and audiophile advantages of the UMZCH BB are achieved by simple circuit solutions and a minimum of active elements. In fact, this is a relatively simple amplifier: one channel can be assembled in a couple of days without haste, and the setup only involves setting the required quiescent current of the output transistors. Especially for novice radio amateurs, a method of node-by-node, cascade testing and adjustment has been developed, using which you can be guaranteed to localize possible errors and prevent their possible consequences even before the UMZCH is fully assembled. All possible questions about this or similar amplifiers have detailed explanations, both on paper and on the Internet.
At the input of the amplifier there is a high-pass filter R1C1 with a cutoff frequency of 1.6 Hz, Fig. 1. But the efficiency of the mode stabilization device allows the amplifier to work with an input signal containing up to 400 mV of DC component voltage. Therefore, C1 is excluded, which realizes the eternal audiophile dream of a path without capacitors © and significantly improves the sound of the amplifier.
The capacitance of capacitor C2 of the input low-pass filter R2C2 is selected so that the cutoff frequency of the input low-pass filter, taking into account the output resistance of the preamplifier 500 Ohm -1 kOhm, is in the range from 120 to 200 kHz. At the input of op amp DA1 there is a frequency correction circuit R3R5C3, which limits the band of processed harmonics and interference coming through the OOS circuit from the output side of the UMZCH, with a band of 215 kHz at a level of -3 dB and increases the stability of the amplifier. This circuit allows you to reduce the difference signal above the cutoff frequency of the circuit and thereby eliminate unnecessary overload of the voltage amplifier with high-frequency interference signals, interference and harmonics, eliminating the possibility of dynamic intermodulation distortion (TIM; DIM).
Next, the signal is fed to the input of a low-noise operational amplifier with field-effect transistors at the DA1 input. Many “claims” to the UMZCH BB are made by opponents regarding the use of an op-amp at the input, which supposedly worsens the sound quality and “steals the virtual depth” of the sound. In this regard, it is necessary to pay attention to some quite obvious features of the operation of the op amp in the UMZCH VV.
Operational amplifiers of pre-amplifiers, post-DAC op-amps are forced to develop several volts of output voltage. Since the gain of the op amp is small and ranges from 500 to 2,000 times at 20 kHz, this indicates that they operate with a relatively high voltage difference signal - from several hundred microvolts at LF to several millivolts at 20 kHz and a high probability of intermodulation distortion being introduced by the input stage of the op amp. The output voltage of these op-amps is equal to the output voltage of the last voltage amplification stage, usually performed according to a circuit with an OE. An output voltage of several volts indicates that this stage operates with fairly large input and output voltages, and as a result, it introduces distortion into the amplified signal. The op-amp is loaded by the resistance of the parallel-connected OOS and load circuits, sometimes amounting to several kilo-ohms, which requires up to several milliamps of output current from the output repeater of the amplifier. Therefore, changes in the current of the output repeater of the IC, the output stages of which consume a current of no more than 2 mA, are quite significant, which also indicates that they introduce distortions into the amplified signal. We see that the input stage, voltage amplification stage and op-amp output stage can introduce distortion.
But the circuit design of the high-fidelity amplifier, due to the high gain and input resistance of the transistor part of the voltage amplifier, provides very gentle operating conditions for op-amp DA1. Judge for yourself. Even in a UMZCH that has developed a nominal output voltage of 50 V, the input differential stage of the op-amp operates with difference signals with voltages from 12 μV at frequencies of 500 Hz to 500 μV at a frequency of 20 kHz. The ratio of the high input overload capacity of the differential stage, made on field-effect transistors, and the scanty voltage of the difference signal ensures high linearity of signal amplification. The output voltage of the op-amp does not exceed 300 mV. which indicates the low input voltage of the voltage amplification stage with a common emitter from the operational amplifier - up to 60 μV - and the linear mode of its operation. The output stage of the op-amp supplies an alternating current of no more than 3 µA to the load of about 100 kOhm from the VT2 base side. Consequently, the output stage of the op-amp also operates in an extremely light mode, almost at idle. On a real musical signal, voltages and currents are most of the time an order of magnitude less than the given values.
From a comparison of the voltages of the difference and output signals, as well as the load current, it is clear that in general the operational amplifier in the UMZCH BB operates in a hundreds of times lighter, and therefore linear, mode than the op-amp mode of preamplifiers and post-DAC op-amps of CD players that serve as sources signal for UMZCH with any depth of environmental protection, as well as without it at all. Consequently, the same op-amp will introduce much less distortion in the UMZCH BB than in a single connection.
Occasionally there is an opinion that the distortions introduced by the cascade ambiguously depend on the voltage of the input signal. This is mistake. The dependence of the manifestation of cascade nonlinearity on the voltage of the input signal may obey one or another law, but it is always unambiguous: an increase in this voltage never leads to a decrease in the introduced distortions, but only to an increase.
It is known that the level of distortion products at a given frequency decreases in proportion to the depth of negative feedback for this frequency. The open-circuit gain, before the amplifier reaches the OOS, at low frequencies cannot be measured due to the smallness of the input signal. According to calculations, the open-circuit gain developed to cover the negative feedback allows one to achieve a negative feedback depth of 104 dB at frequencies up to 500 Hz. Measurements for frequencies starting from 10 kHz show that the OOS depth at a frequency of 10 kHz reaches 80 dB, at a frequency of 20 kHz - 72 dB, at a frequency of 50 kHz - 62 dB and 40 dB - at a frequency of 200 kHz. Figure 2 shows the amplitude-frequency characteristics of the UMZCH VV-2010 and, for comparison, the UMZCH Leonid Zuev, which is similar in complexity.
High gain up to OOS coverage is the main feature of the circuit design of BB amplifiers. Since the goal of all circuit tricks is to achieve high linearity and high gain to maintain deep OOS in the widest possible frequency band, this means that such structures are the only circuit methods for improving amplifier parameters. Further reduction in distortion can only be achieved by design measures aimed at reducing the interference of harmonics of the output stage on the input circuits, especially on the inverting input circuit, from which the gain is maximum.
Another feature of the UMZCH BB circuitry is the current control of the output stage of the voltage amplifier. The input op-amp controls the voltage-current conversion stage, made with OK and OB, and the resulting current is subtracted from the quiescent current of the stage, made according to the circuit with OB.
The use of a linearizing resistor R17 with a resistance of 1 kOhm in the differential stage VT1, VT2 on transistors of different structures with serial power increases the linearity of the conversion of the output voltage of the op-amp DA1 to the collector current VT2 by creating a local feedback loop with a depth of 40 dB. This can be seen from comparing the sum of the emitters’ own resistances VT1, VT2 - approximately 5 Ohms each - with resistance R17, or the sum of the thermal voltages VT1, VT2 - about 50 mV - with the voltage drop across resistance R17 amounting to 5.2 - 5.6 V .
For amplifiers built using the circuit design under consideration, a sharp, 40 dB per decade of frequency, decrease in gain above a frequency of 13...16 kHz is observed. The error signal, which is a product of distortion, at frequencies above 20 kHz is two to three orders of magnitude less than the useful audio signal. This makes it possible to convert the linearity of the differential stage VT1, VT2, which is excessive at these frequencies, into increasing the gain of the transistor part of the UN. Due to minor changes in the current of the differential cascade VT1, VT2, when amplifying weak signals, its linearity with a decrease in the depth of local feedback does not deteriorate significantly, but the operation of the op-amp DA1, on the operating mode of which at these frequencies the linearity of the entire amplifier depends, will make the gain margin easier, since all voltages, The distortions that determine the operational amplifier's distortion, starting from the difference signal to the output signal, decrease in proportion to the gain in gain at a given frequency.
The phase lead correction circuits R18C13 and R19C16 were optimized in the simulator to reduce the op amp differential voltage to frequencies of several megahertz. It was possible to increase the gain of the UMZCH VV-2010 compared to the UMZCH VV-2008 at frequencies of the order of several hundred kilohertz. The gain in gain was 4 dB at 200 kHz, 6 at 300 kHz, 8.6 at 500 kHz, 10.5 dB at 800 kHz, 11 dB at 1 MHz and from 10 to 12 dB at frequencies higher 2 MHz. This can be seen from the simulation results, Fig. 3, where the lower curve refers to the frequency response of the advance correction circuit of the UMZCH VV-2008, and the upper curve refers to the UMZCH VV-2010.
VD7 protects the emitter junction VT1 from reverse voltage arising due to the flow of recharging currents C13, C16 in the mode of limiting the output signal of the UMZCH by voltage and the resulting maximum voltages with a high rate of change at the output of the op-amp DA1.
The output stage of the voltage amplifier is made of transistor VT3, connected according to a common base circuit, which eliminates the penetration of the signal from the output circuits of the cascade into the input circuits and increases its stability. The OB stage, loaded onto the current generator on transistor VT5 and the input resistance of the output stage, develops a high stable gain - up to 13,000...15,000 times. Choosing the resistance of resistor R24 to be half the resistance of resistor R26 guarantees equality of the quiescent currents VT1, VT2 and VT3, VT5. R24, R26 provide local feedback that reduces the Early effect - the change in p21e depending on the collector voltage and increases the initial linearity of the amplifier by 40 dB and 46 dB, respectively. Powering the UN with a separate voltage, modulo 15 V higher than the voltage of the output stages, makes it possible to eliminate the effect of quasi-saturation of transistors VT3, VT5, which manifests itself in a decrease in p21e when the collector-base voltage decreases below 7 V.
The three-stage output follower is assembled using bipolar transistors and does not require any special comments. Don't try to fight entropy © by skimping on the quiescent current of the output transistors. It should not be less than 250 mA; in the author's version - 320 mA.
Before the activation relay AC K1 is activated, the amplifier is covered by OOS1, realized by switching on the divider R6R4. The accuracy of maintaining the resistance R6 and the consistency of these resistances in different channels is not essential, but to maintain the stability of the amplifier it is important that the resistance R6 is not much lower than the sum of the resistances R8 and R70. When relay K1 is triggered, OOS1 is turned off and the OOS2 circuit, formed by R8R70C44 and R4, and covering contact group K1.1, comes into operation, where R70C44 excludes the output low-pass filter R71L1 R72C47 from the OOS circuit at frequencies above 33 kHz. The frequency-dependent OOS R7C10 forms a roll-off in the frequency response of the UMZCH to the output low-pass filter at a frequency of 800 kHz at a level of -3 dB and provides a margin in the OOS depth above this frequency. The decrease in frequency response at the AC terminals above the frequency of 280 kHz at a level of -3 dB is ensured by the combined action of R7C10 and the output low-pass filter R71L1 -R72C47.
The resonant properties of loudspeakers lead to the emission by the diffuser of damped sound vibrations, overtones after pulse action and the generation of its own voltage when the turns of the loudspeaker coil cross the magnetic field lines in the gap of the magnetic system. The damping coefficient shows how large the amplitude of the diffuser's oscillations is and how quickly they attenuate when the AC load is applied as a generator to the full impedance of the UMZCH. This coefficient is equal to the ratio of the AC resistance to the sum of the output resistance of the UMZCH, the transition resistance of the contact group of the AC switching relay, the resistance of the output low-pass filter inductor usually wound with a wire of insufficient diameter, the transition resistance of the AC cable terminals and the resistance of the AC cables themselves.
In addition, the impedance of loudspeaker systems is nonlinear. The flow of distorted currents through the conductors of AC cables creates a voltage drop with a large proportion of harmonic distortion, which is also subtracted from the undistorted output voltage of the amplifier. Therefore, the signal at the AC terminals is distorted much more than at the output of the UMZCH. These are so-called interface distortions.
To reduce these distortions, compensation of all components of the amplifier's output impedance is applied. The UMZCH's own output resistance, together with the transition resistance of the relay contacts and the resistance of the inductor wire of the output low-pass filter, is reduced by the action of a deep general negative feedback taken from the right terminal of L1. In addition, by connecting the right terminal of R70 to the “hot” AC terminal, you can easily compensate for the transition resistance of the AC cable clamp and the resistance of one of the AC wires, without fear of generating UMZCH due to phase shifts in the wires covered by the OOS.
The AC wire resistance compensation unit is made in the form of an inverting amplifier with Ky = -2 on op-amps DA2, R10, C4, R11 and R9. The input voltage for this amplifier is the voltage drop across the “cold” (“ground”) speaker wire. Since its resistance is equal to the resistance of the “hot” wire of the AC cable, to compensate for the resistance of both wires it is enough to double the voltage on the “cold” wire, invert it and, through resistor R9 with a resistance equal to the sum of the resistances R8 and R70 of the OOS circuit, apply it to the inverting input of the op-amp DA1 . Then the output voltage of the UMZCH will increase by the sum of the voltage drops on the speaker wires, which is equivalent to eliminating the influence of their resistance on the damping coefficient and the level of interface distortion at the speaker terminals. Compensation for the drop in the AC wire resistance of the nonlinear component of the back-EMF of loudspeakers is especially necessary at the lower frequencies of the audio range. The signal voltage at the tweeter is limited by the resistor and capacitor connected in series with it. Their complex resistance is much greater than the resistance of the speaker cable wires, so compensating for this resistance at HF makes no sense. Based on this, the integrating circuit R11C4 limits the operating frequency band of the compensator to 22 kHz.
Of particular note: the resistance of the “hot” wire of the AC cable can be compensated by covering its general OOS by connecting the right terminal of R70 with a special wire to the “hot” AC terminal. In this case, only the resistance of the “cold” AC wire will need to be compensated and the gain of the wire resistance compensator must be reduced to the value Ku = -1 by choosing the resistance of resistor R10 equal to the resistance of resistor R11.
The current protection unit prevents damage to the output transistors during short circuits in the load. The current sensor is resistors R53 - R56 and R57 - R60, which is quite enough. The flow of amplifier output current through these resistors creates a voltage drop that is applied to the divider R41R42. A voltage with a value greater than the threshold opens transistor VT10, and its collector current opens VT8 of the trigger cell VT8VT9. This cell enters a stable state with the transistors open and bypasses the HL1VD8 circuit, reducing the current through the zener diode to zero and locking VT3. Discharging C21 with a small current from the VT3 base may take several milliseconds. After the trigger cell is triggered, the voltage on the lower plate of C23, charged by the voltage on the HL1 LED to 1.6 V, increases from the level of -7.2 V from the positive power supply bus to the level of -1.2 V 1, the voltage on the upper plate of this capacitor also increases at 5 V. C21 is quickly discharged through resistor R30 to C23, transistor VT3 is turned off. In the meantime, VT6 opens and through R33, R36 opens VT7. VT7 bypasses the zener diode VD9, discharges capacitor C22 through R31 and turns off transistor VT5. Without receiving bias voltage, the output stage transistors are also turned off.
Restoring the initial state of the trigger and turning on the UMZCH is done by pressing the SA1 “Protection Reset” button. C27 is charged by the collector current of VT9 and bypasses the base circuit of VT8, locking the trigger cell. If by this moment the emergency situation has been eliminated and VT10 is locked, the cell goes into a state with stable closed transistors. VT6, VT7 are closed, the reference voltage is supplied to the bases VT3, VT5 and the amplifier enters operating mode. If the short circuit in the UMZCH load continues, the protection is triggered again, even if capacitor C27 is connected to SA1. The protection works so effectively that during work on setting up the correction, the amplifier was de-energized several times for small soldering connections ... by touching the non-inverting input. The resulting self-excitation led to an increase in the current of the output transistors, and the protection turned off the amplifier. Although this crude method cannot be suggested as a general rule, but due to the current protection, it did not cause any harm to the output transistors.
Operation of the AC cable resistance compensator.
The efficiency of the UMZCH BB-2008 compensator was tested using the old audiophile method, by ear, by switching the compensator input between the compensating wire and the common wire of the amplifier. The improvement in sound was clearly noticeable, and the future owner was eager to get an amplifier, so measurements of the influence of the compensator were not carried out. The advantages of the “cable cleaning” circuit were so obvious that the “compensator + integrator” configuration was adopted as a standard unit for installation in all developed amplifiers.
It's surprising how much unnecessary debate has flared up on the Internet regarding the usefulness/uselessness of cable resistance compensation. As usual, those who especially insisted on listening to a nonlinear signal were those to whom the extremely simple cable cleaning scheme seemed complex and incomprehensible, the costs for it exorbitant, and installation labor-intensive ©. There were even suggestions that since so much money is spent on the amplifier itself, it would be a sin to skimp on the sacred, but one should take the best, glamorous path that all civilized humanity follows and...purchase normal, human © super-expensive cables made of precious metals. To my great surprise, fuel was added to the fire by statements from highly respected specialists about the uselessness of the compensation unit at home, including those specialists who successfully use this unit in their amplifiers. It is very unfortunate that many fellow radio amateurs were distrustful of reports of improved sound quality in the low and midrange with the inclusion of a compensator, and did their best to avoid this simple way of improving the performance of the UMZCH, thereby robbing themselves.
Little research has been done to document the truth. From the GZ-118 generator, a number of frequencies were supplied to the UMZCH BB-2010 in the region of the resonant frequency of the AC, the voltage was controlled by an oscilloscope S1-117, and Kr at the AC terminals was measured by the INI S6-8, Fig. 4. 
Resistor R1 is installed to avoid interference to the compensator input when switching it between the control and common wires. In the experiment, common and publicly available AC cables with a length of 3 m and a core cross-section of 6 square meters were used. mm, as well as the GIGA FS Il speaker system with a frequency range of 25 -22,000 Hz, a nominal impedance of 8 Ohms and a nominal power of 90 W from Acoustic Kingdom.
Unfortunately, the circuit design of harmonic signal amplifiers from C6-8 involves the use of high-capacity oxide capacitors in OOS circuits. This causes the low-frequency noise of these capacitors to affect the device's low-frequency resolution, causing its low-frequency resolution to deteriorate. When measuring a Kr signal with a frequency of 25 Hz from GZ-118 directly from C6-8, the instrument readings dance around the value of 0.02%. It is not possible to bypass this limitation using the notch filter of the GZ-118 generator in the case of measuring the efficiency of the compensator, because a number of discrete values of the tuning frequencies of the 2T filter are limited at low frequencies to 20.60, 120, 200 Hz and do not allow measuring Kr at the frequencies of interest to us. Therefore, reluctantly, the level of 0.02% was accepted as zero, the reference.
At a frequency of 20 Hz with a voltage at the AC terminals of 3 Vamp, which corresponds to an output power of 0.56 W into an 8 Ohm load, Kr was 0.02% with the compensator turned on and 0.06% with it turned off. At a voltage of 10 V ampl, which corresponds to an output power of 6.25 W, the Kr value is 0.02% and 0.08%, respectively, at a voltage of 20 V ampl and a power of 25 W - 0.016% and 0.11%, and at a voltage of 30 In amplitude and power 56 W - 0.02% and 0.13%.
Knowing the relaxed attitude of manufacturers of imported equipment to the meanings of inscriptions regarding power, and also remembering the wonderful, after the adoption of Western standards, transformation of the 35AC-1 speaker system with a subwoofer power of 30 W into the S-90, long-term power of more than 56 W was not supplied to the AC.
At a frequency of 25 Hz at a power of 25 W, Kr was 0.02% and 0.12% with the compensation unit on/off, and at a power of 56 W - 0.02% and 0.15%.
At the same time, the necessity and effectiveness of covering the output low-pass filter with a general OOS was tested. At a frequency of 25 Hz with a power of 56 W and connected in series to one of the AC cable wires of the output RL-RC low-pass filter, similar to that installed in an ultra-linear UMZCH, Kr with the compensator turned off reaches 0.18%. At a frequency of 30 Hz at a power of 56 W Kr 0.02% and 0.06% with the compensation unit on/off. At a frequency of 35 Hz at a power of 56 W Kr 0.02% and 0.04% with the compensation unit on/off. At frequencies of 40 and 90 Hz at a power of 56 W, Kr is 0.02% and 0.04% with the compensation unit on/off, and at a frequency of 60 Hz -0.02% and 0.06%.
The conclusions are obvious. The presence of nonlinear signal distortions at the AC terminals is observed. A deterioration in the linearity of the signal at the AC terminals is clearly detected when it is connected through the uncompensated, not covered by the OOS resistance of the low-pass filter containing 70 cm of relatively thin wire. The dependence of the distortion level on the power supplied to the AC suggests that it depends on the ratio of the signal power and the rated power of the AC woofers. Distortion is most pronounced at frequencies near the resonant one. The back EMF generated by the speakers in response to the influence of an audio signal is shunted by the sum of the output resistance of the UMZCH and the resistance of the AC cable wires, so the level of distortion at the AC terminals directly depends on the resistance of these wires and the output resistance of the amplifier.
The cone of a poorly damped low-frequency loudspeaker itself emits overtones, and, in addition, this loudspeaker generates a wide tail of non-linear and intermodulation distortion products that the mid-frequency loudspeaker reproduces. This explains the deterioration of sound at mid frequencies.
Despite the assumption of a zero Kr level of 0.02% adopted due to the imperfection of the INI, the influence of the cable resistance compensator on the signal distortion at the AC terminals is clearly and unambiguously noted. It can be stated that there is complete agreement between the conclusions drawn after listening to the operation of the compensation unit on a musical signal and the results of instrumental measurements.
The improvement clearly audible when the cable cleaner is turned on can be explained by the fact that with the disappearance of distortion at the AC terminals, the midrange speaker stops producing all that dirt. Apparently, therefore, by reducing or eliminating the reproduction of distortions by the mid-frequency loudspeaker, the two-cable speaker circuit, the so-called. “Bi-wiring,” when the LF and MF-HF sections are connected with different cables, has an advantage in sound compared to a single-cable circuit. However, since in a two-cable circuit the distorted signal at the terminals of the AC low-frequency section does not disappear anywhere, this circuit is inferior to the version with a compensator in terms of the damping coefficient of free vibrations of the low-frequency loudspeaker cone.
You can’t fool physics, and for decent sound it’s not enough to get brilliant performance at the amplifier output with an active load, but you also need to not lose linearity after delivering the signal to the speaker terminals. As part of a good amplifier, a compensator made according to one scheme or another is absolutely necessary.
Integrator.
The efficiency and error reduction capabilities of the integrator on DA3 were also tested. In the UMZCH BB with op-amp TL071, the output DC voltage is in the range of 6...9 mV and it was not possible to reduce this voltage by including an additional resistor in the non-inverting input circuit.
The effect of low-frequency noise, characteristic of an op-amp with a DC input, due to the coverage of deep feedback through the frequency-dependent circuit R16R13C5C6, manifests itself in the form of instability of the output voltage of several millivolts, or -60 dB relative to the output voltage at rated output power, at frequencies below 1 Hz , non-reproducible speakers.
The Internet mentioned the low resistance of the protective diodes VD1...VD4, which supposedly introduces an error in the operation of the integrator due to the formation of a divider (R16+R13)/R VD2|VD4 . . To check the reverse resistance of the protective diodes, a circuit was assembled in Fig. 6. Here, op-amp DA1, connected according to the inverting amplifier circuit, is covered by OOS through R2, its output voltage is proportional to the current in the circuit of the tested diode VD2 and the protective resistor R2 with a coefficient of 1 mV/nA, and the resistance of the circuit R2VD2 - with a coefficient of 1 mV/15 GOhm. To exclude the influence of additive errors of the op-amp - bias voltage and input current on the results of measuring the leakage current of the diode, it is necessary to calculate only the difference between the intrinsic voltage at the output of the op-amp, measured without the diode being tested, and the voltage at the output of the op-amp after its installation. In practice, a difference in the op-amp output voltages of several millivolts gives a diode reverse resistance value of the order of ten to fifteen gigaohms at a reverse voltage of 15 V. Obviously, the leakage current will not increase as the voltage on the diode decreases to a level of several millivolts, characteristic of the difference voltage of the op-amp integrator and compensator .
But the photoelectric effect characteristic of diodes placed in a glass case actually leads to a significant change in the output voltage of the UMZCH. When illuminated with a 60 W incandescent lamp from a distance of 20 cm, the constant voltage at the output of the UMZCH increased to 20...3O mV. Although it is unlikely that a similar level of illumination could be observed inside the amplifier case, a drop of paint applied to these diodes eliminated the dependence of the UMZCH modes on illumination. According to the simulation results, the decrease in the frequency response of the UMZCH is not observed even at a frequency of 1 millihertz. But the time constant R16R13C5C6 should not be reduced. The phases of the alternating voltages at the outputs of the integrator and compensator are opposite, and with a decrease in the capacitance of the capacitors or the resistance of the integrator resistors, an increase in its output voltage can worsen the compensation of the resistance of the speaker cables.
Comparison of the sound of amplifiers. The sound of the assembled amplifier was compared with the sound of several industrially produced foreign amplifiers. The source was a Cambridge Audio CD player; the Radiotekhnika UP-001 pre-amplifier was used to drive and adjust the sound level of the final UMZCHs; the Sugden A21a and NAD C352 used standard adjustment controls.
The first to be tested was the legendary, shocking and damn expensive English UMZCH “Sugden A21a”, operating in class A with an output power of 25 W. What is noteworthy is that in the accompanying documentation for the VX, the British considered it better not to indicate the level of nonlinear distortions. They say it’s not a matter of distortion, but of spirituality. “Sugden A21a>” lost to the UMZCH BB-2010 with comparable power both in level and in clarity, confidence, and noble sound at low frequencies. This is not surprising, given the features of its circuit design: just a two-stage quasi-symmetric output follower on transistors of the same structure, assembled according to the circuit design of the 70s of the last century with a relatively high output resistance and an electrolytic capacitor connected at the output, which further increases the total output resistance - this is the latter the solution itself worsens the sound of any amplifiers at low and mid frequencies. At medium and high frequencies, the UMZCH BB showed higher detail, transparency and excellent scene elaboration, when singers and instruments could be clearly localized by sound. By the way, speaking of the correlation of objective measurement data and subjective impressions of sound: in one of the journal articles of Sugden’s competitors, its Kr was determined at the level of 0.03% at a frequency of 10 kHz.
The next one was also the English amplifier NAD C352. The general impression was the same: the pronounced “bucket” sound of the Englishman at low frequencies left him no chance, while the work of the UMZCH BB was recognized as impeccable. Unlike NADA, the sound of which was associated with dense bushes, wool, and cotton wool, the sound of BB-2010 at medium and high frequencies made it possible to clearly distinguish the voices of performers in a general choir and instruments in an orchestra. The work of the NAD C352 clearly expressed the effect of better audibility of a more vocal performer, a louder instrument. As the owner of the amplifier himself put it, in the sound of the UMZCH BB the vocalists did not “scream and nod” at each other, and the violin did not fight in sound power with the guitar or trumpet, but all the instruments were peacefully and harmoniously “friends” in the overall sound image of the melody. At high frequencies, the UMZCH BB-2010, according to imaginative audiophiles, sounds “as if it were painting the sound with a thin, thin brush.” These effects can be attributed to differences in intermodulation distortion between the amplifiers.
The sound of the Rotel RB 981 UMZCH was similar to the sound of the NAD C352, with the exception of better performance at low frequencies, yet the BB-2010 UMZCH remained unrivaled in the clarity of AC control at low frequencies, as well as the transparency and delicacy of sound at mid and high frequencies.
The most interesting thing in terms of understanding the way of thinking of audiophiles was the general opinion that, despite their superiority over these three UMZCHs, they bring “warmth” to the sound, which makes it more pleasant, and the BB UMZCH works smoothly, “it is neutral to the sound.”
The Japanese Dual CV1460 lost its sound immediately after switching on in the most obvious way for everyone, and we didn’t waste time listening to it in detail. Its Kr was in the range of 0.04...0.07% at low power.
The main impressions from comparing the amplifiers were completely identical in their main features: the UMZCH BB was unconditionally and unequivocally ahead of them in sound. Therefore, further testing was deemed unnecessary. In the end, friendship won, everyone got what they wanted: for a warm, soulful sound - Sugden, NAD and Rotel, and to hear what was recorded on disk by the director - UMZCH BB-2010.
Personally, I like the high-fidelity UMZCH for its light, clean, impeccable, noble sound; it effortlessly reproduces passages of any complexity. As a friend of mine, an experienced audiophile, put it, he handles the sounds of drum kits at low frequencies without variations, like a press, at medium frequencies he sounds as if there is none, and at high frequencies he seems to be painting the sound with a thin brush. For me, the non-straining sound of the UMZCH BB is associated with the ease of operation of the cascades.
Literature
1. Sukhov I. UMZCH of high fidelity. "Radio", 1989, No. 6, pp. 55-57; No. 7, pp. 57-61.
2. Ridiko L. UMZCH BB on a modern element base with a microcontroller control system. “Radio Hobby”, 2001, No. 5, pp. 52-57; No. 6, pp. 50-54; 2002, no. 2, pp. 53-56.
3. Ageev S. Superlinear UMZCH with deep environmental protection “Radio”, 1999, No. 10... 12; "Radio", 2000, No. 1; 2; 4…6; 9…11.
4. Zuev. L. UMZCH with parallel OOS. "Radio", 2005, No. 2, p. 14.
5. Zhukovsky V. Why do you need the speed of the UMZCH (or “UMZCH VV-2008”)? “Radio Hobby”, 2008, No. 1, pp. 55-59; No. 2, pp. 49-55.
