High quality preamplifier "NATALY". My version. Schematic diagram, drawing of the printed circuit board of the NATALY pre-amplifier Natali pre-amplifier

What I have at the moment:

1. The amplifier itself:

2. Naturally, the power supply of the final amplifier:

When setting up the PA, I use a device that ensures a safe connection of the PA transformer to the network (via a lamp). It is made in a separate box with its own cord and socket and, if necessary, connects to any device. The diagram is shown below in the figure. This device requires a relay with a 220 AC winding and two groups of contacts for closure, one momentary button (S2), one latching button or switch (S1). When S1 is closed, the transformer is connected to the network through the lamp, if all modes of the PA are normal, when you press the S2 button, the relay closes the lamp through one group of contacts and connects the transformer directly to the network, and the second group of contacts, duplicating the S2 button, constantly connects the relay to the network. The device remains in this state until S1 opens, or the voltage decreases below the holding voltage of the relay contacts (including short circuit). The next time you turn on S1, the transformer is again connected to the network through the lamp, and so on...

Noise immunity of various methods of shielding signal wires

3. We also have assembled AC protection against DC voltage:

The protection includes:
speaker connection delay
protection against constant output, against short circuit
airflow control and turning off the speakers when the radiators overheat

Setting up:
Let's assume that everything is assembled from serviceable transistors and diodes tested by a tester. Initially, place the trimmer engines in the following positions: R6 - in the middle, R12, R13 - in the top according to the diagram.
Do not solder the VD7 zener diode at first. The protection board contains Zobel circuits, which are necessary for the stability of the amplifier; if they are already present on the UMZCH boards, then they do not need to be soldered, and the coils can be replaced with jumpers. Otherwise, the coils are wound on a mandrel with a diameter of 10 mm, for example, on the tail of a drill - with a wire with a diameter of 1 mm. The length of the resulting winding should be such that the coil fits into the holes provided for it on the board. After winding, I recommend impregnating the wire with varnish or glue, for example, epoxy or BFom - for rigidity.
For now, connect the wires going from the protection to the amplifier outputs to the common wire, disconnecting them from its outputs, of course. It is necessary to connect the earth protection polygon, marked on the PCB with the mark “Main GND”, to the “Mecca” UMZCH, otherwise the protection will not work correctly. And, of course, GND pads next to the coils.
Having turned on the protection with the speakers connected, we begin to reduce the resistance R6 until the relay clicks. After unscrewing the trimmer one or two more turns, we turn off the network protection, connect two speakers in parallel on any of the channels and check whether the relays work. If they don’t work, then everything works as intended; with a load of 2 Ohms, the amplifiers will not connect to it, in order to avoid damage.
Next, we disconnect the wires “From UMZCH LC” and “From UMZCH PC” from the ground, turn everything on again and check whether the protection will work if a constant voltage of about two or three volts is applied to these wires. The relays should turn off the speakers - there will be a click.
You can enter the “Protection” indication if you connect a chain of a red LED and a 10 kOhm resistor between ground and the VT6 collector. This LED will indicate a fault.
Next, we set up thermal control. We put the thermistors in a waterproof tube (attention! they should not get wet during the test!).
It often happens that a radio amateur does not have the thermistors indicated on the diagram. Two identical ones from those available will do, with a resistance of 4.7 kOhm, but in this case the resistance of R15 should be equal to twice the resistance of the thermistors connected in series. Thermistors must have a negative coefficient of resistance (reduce it with heating), posistors work the other way around and have no place here. Boil a glass of water. Let it cool for 10-15 minutes in calm air and lower the thermistors into it. Turn R13 until the “Overheat” LED goes out, which should have been lit initially.
When the water cools down to 50 degrees (this can be accelerated, exactly how is a big secret) - turn R12 so that the “Blowing” or FAN On LED goes out.
We solder the VD7 zener diode into place.
If no glitches are detected from the sealing of this zener diode, then everything is fine, but it happened that without it the transistor part works flawlessly, but with it it does not want to connect the relay to any. In this case, we change it to any one with a stabilization voltage from 3.3 V to 10 V. The reason is a zener diode leak.
When the thermistors heat up to 90*C, the “Overheat” LED should light up - Overheating and the relay will disconnect the speakers from the amplifier. When the radiators cool down a bit, everything will be connected back, but this mode of operation of the device should at least alert the owner. If the fan is working properly and the tunnel is not clogged with dust, thermal activation should not be observed at all.
If everything is fine, solder the wires to the amplifier outputs and enjoy.
The airflow (its intensity) is adjusted by selecting resistors R24 and R25. The first determines the cooler's performance when the fan is turned on (maximum), the second - when the radiators are only slightly warm. R25 can be excluded altogether, but then the fan will operate in ON-OFF mode.
If the relays have 24V windings, then they must be connected in parallel, but if they have 12V windings, then they must be connected in series.
Replacement of parts. As an op-amp, you can use almost any dual cheap op-amp in SOIK8 (from 4558 to OPA2132, although, I hope, it will not come to the latter), for example, TL072, NE5532, NJM4580, etc.
Transistors - 2n5551 are replaced with BC546-BC548, or with our KT3102. We can replace BD139 with 2SC4793, 2SC2383, or with a similar current and voltage, it is possible to install even KT815.
The polevik is replaced with one similar to the one used, the choice is huge. A radiator is not required for the field worker.
Diodes 1N4148 are replaced with 1N4004 - 1N4007 or with KD522. In the rectifier, you can put 1N4004 - 1N4007 or use a diode bridge with a current of 1 A.
If blowing control and protection against overheating of the UMZCH are not needed, then the right side of the circuit is not soldered - the op-amp, thermistors, field switch, etc., except for the diode bridge and filter capacitor. If you already have a 22..25V power source in the amplifier, then you can use it, not forgetting about the protection current consumption of about 0.35A when the blower is turned on.

Recommendations for assembling and configuring the UMZCH:
Before you begin assembling the printed circuit board, you should perform relatively simple operations on the board, namely, look in the light to see if there are any short circuits between the tracks that are barely noticeable under normal lighting. Factory production does not exclude manufacturing defects, unfortunately. Soldering is recommended to be done with POS-61 solder or similar with a melting point of no higher than 200* C.

First you need to decide on the op amp used. The use of op-amps from Analog Devices is highly discouraged - in this UMZCH their sound character is somewhat different from that intended by the author, and an excessively high speed can lead to irreparable self-excitation of the amplifier. Replacing OPA134 with OPA132, OPA627 is welcome, because they have less distortion at HF. The same applies to op-amp DA1 - it is recommended to use OPA2132, OPA2134 (in order of preference). It is acceptable to use OPA604, OPA2604, but there will be slightly more distortion. Of course, you can experiment with the type of op-amp, but at your own peril and risk. The UMZCH will work with KR544UD1, KR574UD1, but the level of zero offset at the output will increase and the harmonics will increase. The sound... I think no comments are needed.

From the very beginning of installation, it is recommended to select transistors in pairs. This is not a necessary measure, because the amplifier will work even with a spread of 20-30%, but if your goal is to get maximum quality, then pay attention to this. Particular attention should be paid to the selection of T5, T6 - they are best used with maximum H21e - this will reduce the load on the op-amp and improve its output spectrum. T9, T10 should also have the gain as close as possible. For latch transistors, selection is optional. Output transistors - if they are from the same batch, you don’t have to select them, because The production culture in the West is slightly higher than what we are used to and the spread is within 5-10%.

Next, instead of the terminals of resistors R30, R31, it is recommended to solder pieces of wire a couple of centimeters long, since it will be necessary to select their resistances. An initial value of 82 Ohms will give a quiescent current of approximately 20..25 mA, but statistically it turned out to be from 75 to 100 Ohms, this greatly depends on the specific transistors.
As already noted in the topic on the amplifier, you should not use transistor optocouplers. Therefore, you should focus on AOD101A-G. Imported diode optocouplers were not tested due to unavailability, this is temporary. The best results are obtained on the AOD101A of one batch for both channels.

In addition to transistors, it is worth choosing complementary UNA resistors in pairs. The spread should not exceed 1%. Particular care must be taken to select R36=R39, R34=R35, R40=R41. As a guide, I note that with a spread of more than 0.5%, it is better not to switch to the option without environmental protection, because there will be an increase in even harmonics. It was the inability to obtain precise details that at one time stopped the author’s experiments in the non-OOS direction. The introduction of balancing into the current feedback circuit does not completely solve the problem.

Resistors R46, R47 can be soldered at 1 kOhm, but if you want to more accurately adjust the current shunt, then it is better to do the same as with R30, R31 - solder in the wiring for soldering.
As it turned out during the repetition of the circuit, under certain circumstances it is possible to excite an EA in the tracking circuit. This manifested itself in the form of an uncontrolled drift of the quiescent current, and especially in the form of oscillations with a frequency of about 500 kHz on the collectors T15, T18.
The necessary adjustments were initially included in this version, but it’s still worth checking with an oscilloscope.

Diodes VD14, VD15 are placed on the radiator for temperature compensation of the quiescent current. This can be done by soldering the wires to the leads of the diodes and gluing them to the radiator with “Moment” type glue or similar.

Before turning it on for the first time, you must thoroughly wash the board from traces of flux, check for any short circuits in the tracks with solder, and make sure that the common wires are connected to the midpoint of the power supply capacitors. It is also strongly recommended to use a Zobel circuit and a coil at the output of the UMZCH; they are not shown in the diagram, because the author considers their use to be a rule of good form. The ratings of this circuit are common - these are a series-connected 10 Ohm 2 W resistor and a K73-17 capacitor or similar with a capacity of 0.1 μF. The coil is wound with varnished wire with a diameter of 1 mm on an MLT-2 resistor, the number of turns is 12...15 (until filling). On the protection PP this circuit is completely separated.

All transistors VK and T9, T10 in UN are mounted on the radiator. Powerful VK transistors are installed through mica spacers and a paste of the KPT-8 type is used to improve thermal contact. It is not recommended to use computer pastes - there is a high probability of counterfeiting, and tests confirm that KPT-8 is often the best choice, and also very inexpensive. To avoid getting caught by a fake, use KPT-8 in metal tubes, like toothpaste. We haven't gotten to that point yet, fortunately.

For transistors in an insulated housing, the use of a mica spacer is not necessary and even undesirable, because worsens the conditions of thermal contact.
Be sure to turn on a 100-150W light bulb in series with the primary winding of the network transformer - this will save you from many troubles.

Short-circuit the D2 optocoupler LED leads (1 and 2) and turn on. If everything is assembled correctly, the current consumed by the amplifier should not exceed 40 mA (the output stage will operate in mode B). The DC bias voltage at the output of the UMZCH should not exceed 10 mV. Unwrap the LED. The current consumed by the amplifier should increase to 140...180 mA. If it increases more, then check (it is recommended to do this with a pointer voltmeter) collectors T15, T18. If everything works correctly, there should be voltages that differ from the supply ones by about 10-20 V. In the case when this deviation is less than 5 V, and the quiescent current is too high, try changing the diodes VD14, VD15 to others, it is very desirable that they were from the same party. The UMZCH quiescent current, if it does not fall within the range from 70 to 150 mA, can also be set by selecting resistors R57, R58. Possible replacement for diodes VD14, VD15: 1N4148, 1N4001-1N4007, KD522. Or reduce the current flowing through them by simultaneously increasing R57, R58. In my thoughts there was the possibility of implementing a bias of such a plan: instead of VD14, VD15, use transitions of BE transistors from the same batches as T15, T18, but then you would have to significantly increase R57, R58 - until the resulting current mirrors are fully adjusted. In this case, the newly introduced transistors must be in thermal contact with the radiator, as well as the diodes in which they are installed.

Next you need to set the quiescent current UNA. Leave the amplifier turned on and after 20-30 minutes check the voltage drop across resistors R42, R43. 200...250 mV should drop there, which means a quiescent current of 20-25 mA. If it is greater, then it is necessary to reduce the resistances R30, R31; if it is less, then increase it accordingly. It may happen that the quiescent current of the UA will be asymmetrical - 5-6mA in one arm, 50mA in the other. In this case, unsolder the transistors from the latch and continue without them for now. The effect did not find a logical explanation, but disappeared when replacing transistors. In general, there is no point in using transistors with large H21e in the latch. A gain of 50 is enough.

After setting up the UN, we again check the quiescent current of the VK. It should be measured by the voltage drop across resistors R79, R82. A current of 100 mA corresponds to a voltage drop of 33 mV. Of these 100 mA, about 20 mA is consumed by the pre-final stage and up to 10 mA can be spent on controlling the optocoupler, so in the case when, for example, 33 mV drops across these resistors, the quiescent current will be 70...75 mA. It can be clarified by measuring the voltage drop across the resistors in the emitters of the output transistors and subsequent summation. The quiescent current of the output transistors from 80 to 130 mA can be considered normal, while the declared parameters are completely preserved.

Based on the results of voltage measurements on the collectors T15, T18, we can conclude that the control current through the optocoupler is sufficient. If T15, T18 are almost saturated (the voltages on their collectors differ from the supply voltages by less than 10 V), then you need to reduce the ratings of R51, R56 by about one and a half times and re-measure. The situation with voltages should change, but the quiescent current should remain the same. The optimal case is when the voltages on the collectors T15, T18 are equal to approximately half of the supply voltages, but a deviation from the supply of 10-15V is quite sufficient; this is a reserve that is needed to control the optocoupler on a music signal and a real load. Resistors R51, R56 can heat up to 40-50*C, this is normal.

Instantaneous power in the most severe case - with an output voltage close to zero - does not exceed 125-130 W per transistor (according to technical conditions, up to 150 W is allowed) and it acts almost instantly, which should not lead to any consequences.

The actuation of the latch can be determined subjectively by a sharp decrease in output power and a characteristic “dirty” sound, in other words, there will be a highly distorted sound in the speakers.

4. Pre-amplifier and its power supply

High quality PU material:

Serves for timbre correction and loudness compensation when adjusting the volume. Can be used to connect headphones.

The well-proven Matyushkin TB was used as a tone block. It has a 4-stage low-frequency adjustment and smooth high-frequency adjustment, and its frequency response corresponds well to auditory perception; in any case, the classic bridge TB (which can also be used) is rated lower by listeners. The relay allows, if necessary, to disable any frequency correction in the path; the output signal level is adjusted by a trimming resistor to equalize the gain at a frequency of 1000 Hz in the TB mode and when bypassing.

Design characteristics:

Kg in the frequency range from 20 Hz to 20 kHz - less than 0.001% (typical value about 0.0005%)

Rated input voltage, V 0.775

Overload capacity in TB bypass mode is at least 20 dB.

The minimum load resistance at which operation of the output stage is guaranteed in mode A is with a maximum peak-to-peak output voltage swing of 58V 1.5 kOhm.

When using the control unit only with CD players, it is permissible to reduce the buffer supply voltage to +\-15V because the output voltage range of such signal sources is obviously limited from above, this will not affect the parameters.

A complete set of boards consists of two PU channels, Matyushkin RT (one board for both channels) and a power supply. Printed circuit boards were designed by Vladimir Lepekhin.

Measurement results:

What I have at the moment:

1. The amplifier itself:

2. Naturally, the power supply of the final amplifier:

Noise immunity of various methods of shielding signal wires

3. We also have assembled AC protection against DC voltage:

The protection includes:
speaker connection delay
protection against constant output, against short circuit
airflow control and turning off the speakers when the radiators overheat

4. Pre-amplifier and its power supply

High quality PU material:

Serves for timbre correction and loudness compensation when adjusting the volume. Can be used to connect headphones.

Design characteristics:

Kg in the frequency range from 20 Hz to 20 kHz - less than 0.001% (typical value about 0.0005%)

High quality preamplifier NATALY

Schematic diagram, description, printed circuit board

This preamplifier is used for timbre correction and loudness compensation when adjusting the volume. Can be used to connect headphones.

For a high-quality path that includes an UMZCH with nonlinear and intermodulation distortions of the order of 0.001%, the remaining stages become important, which should allow the full potential to be realized. Currently, there are many known options for implementing high parameters, including using op-amps. The reasons for developing our own version of the preamplifier were the following factors:

When assembling a preamplifier on an op-amp, the threshold of its output voltage, and therefore the overload capacity, is entirely determined by the supply voltage of the op-amp, and in the case of power supply from +\-15V it cannot be higher than this voltage.
The results of subjective examinations of preamplifiers based on op-amps in their pure form (without output repeaters) and with those, for example, based on a parallel amplifier, show listeners’ preference for the op-amp + repeater circuit, with almost identical parameters “from the point of view of Kg”, this is explained by the narrowing of the spectrum of op-amp distortion when working with a high-resistance load and operating its output stage without entering the AB mode, which produces switching distortions that are practically below the level of sensitivity of the devices (Kg OU ORA134, for example - 0.00008%), but clearly noticeable when listening. This is why, as well as for a number of other reasons, listeners clearly distinguish a preamplifier with a transistor output stage.
The well-known circuit solution containing an integrated repeater based on the BUF634 parallel amplifier is quite expensive (buffer price is at least 500 rubles), although the internal buffer circuit can be easily implemented in discrete form - for a much more reasonable amount.
Amplifiers in which the op-amp operates in a small-signal mode show high performance, but lose in audition results. In addition, they are very critical to set up and require, at a minimum, a square wave generator and a wideband oscilloscope. And all this with clearly worse subjective results.

The lack of output voltage in the PU circuit (op-amp + buffer) can be eliminated by implementing voltage amplification in the buffer, and deep local feedback eliminates distortion. A sufficiently high initial quiescent current in the output transistors of the buffer guarantees its operation without distortions characteristic of push-pull structures in the AV mode. The presence of only a twofold voltage amplification allows one to achieve an increase in overload capacity by 6 dB, and with a threefold amplification, this figure becomes equal to 9 dB. When the buffer operates from a +\-30V power source, its output voltage range is 58 volts peak to peak. If the buffer is powered from +\-45V, then the output voltage from peak to peak can be about 87V. This margin will be beneficial when listening to vinyl discs that have characteristic features in the form of clicks from dust.
The two-stage implementation of the preamplifier is due to the fact that the timbre block introduces attenuation into the signal up to 10...12 dB. Of course, you can compensate for this by increasing the gain of the second stage, but, as practice shows, it is better to apply as much voltage as possible to the tone block - this increases the signal to noise ratio.

In addition, it is quite common to find discs recorded with a high crest factor (loud peaks and rather low average volume). This is not a lack of mixing, rather, on the contrary, because sound engineers often abuse the compressor, trying to fit all levels of sound volume into the CD range. But we cannot pretend that such records do not exist. The listener turns up the volume. Thus, the second stage must have no less overload capacity; in addition, it must have low intrinsic noise, high input impedance and the ability to pass the real signal without distortion after the tone block, in which the extreme frequencies of the audio range are most elevated. An additional requirement is a linear frequency response when the tone control is turned off, an even response when testing with a meander, and subjective invisibility of the control unit in the path.
Matyushkin’s well-proven tone block was used as a tone block. It has a 4-stage low-frequency adjustment and smooth high-frequency adjustment, and its frequency response corresponds well to auditory perception; in any case, the classic bridge TB (which can also be used) is rated lower by listeners. The relay allows, if necessary, to disable any frequency correction in the path; the output signal level is adjusted by a trimming resistor to equalize the gain at a frequency of 1000 Hz in the TB mode and when bypassing.
The balance regulator is built into the second stage OOS and has no special features.

The low bias voltage of the OPA134 (in the author’s practice, at the output of the second stage is no more than 1 mV) makes it possible to exclude transition capacitors in the path, leaving only one at the input of the control unit, because the level of constant voltage at the output of the signal source is unknown. And, although at the output of the second stage the diagram shows capacitors of 4.7 μF + 2200 pF - with a zero offset level of about a millivolt or less - they can be safely eliminated by short-circuiting them. This will put an end to the debate about the effect of capacitors in the path on sound - the most radical method.

Design characteristics:
Rated input voltage, V 0.775
The overload capacity in the tone block bypass mode is at least 20 dB.
The minimum load resistance at which operation of the output stage is guaranteed in mode A is with a maximum peak-to-peak output voltage swing of 58V 1.5 kOhm.

When using a pre-amplifier only with CD players, it is permissible to reduce the buffer supply voltage to +\-15V because the output voltage range of such signal sources is obviously limited from above, this will not affect the parameters.
Setting up a pre-amplifier should begin by checking the DC modes of the output buffer transistors. Based on the voltage drop in the circuits of their emitters, the quiescent current is set - for the first stage it is about 20 mA, for the second - 20..25 mA. When using small heat sinks, which become mandatory at +\-30V, it is possible, depending on the temperature situation, to increase the quiescent current a little more.
It is best to select the quiescent current using resistors in the emitters of the first two buffer transistors. If the current is low, increase the resistance; if the current is high, decrease it. Both resistors need to be changed equally.
With the quiescent current set, we then set the TB regulators to the position corresponding to the flattest frequency response, and, by applying a 1000 Hz signal with a rated voltage of 0.775V to the input, we measure the voltage at the output of the second buffer. Then we turn on the bypass mode and use a trimming resistor to achieve the same amplitude as with the TB.
At the final stage, we connect the stereo balance control, check for the absence of various forms of instability (the author did not encounter such a problem) and conduct a listening session. Setting up Matyushkin's TB is well covered in the author's article and is not discussed here.
To power the preamplifier, a stabilized power supply is recommended, with independent windings for control unit and relay switching. Technically, the power requirements are nothing new.

The main thing is the low level of mid-range and high-frequency noise, the suppression of which by power supply is known for the op-amp. About the ripple level - it should not exceed 0.5 - 1 mV.

A complete set of boards consists of two PU channels, Matyushkin RT (one board for both channels) and a power supply. Printed circuit boards were designed by Vladimir Lepekhin.

Double Sided Pre-Amplifier PCB:

INCREASE

ENLARGE The circuit is stable. There is no noticeable voltage ripple at the output; measurements were taken on an oscilloscope in the 0.01 division/volt mode (for mine this is the minimum limit).

Double Sided Pre-Amplifier PCB:

Measurement results:

On OPA134 (only the first link of two), the power supply is single-stage, +\-15V:

Kni(1kHz)........................ -98dB (about 0.0003%)
Kim(50Hz+7kHz)................less than -98dB (about 0.0003%)

On OPA132 (both links), full version, two-stage power supply:

Kni (1kHz)........................ -100dB (about 0.00025%)
Kim (19kHz+20kHz)................... -96dB (about 0.0003%)

In the case of self-excitation of HF cascades, mica correction capacitors with a capacity of 100 to 470 pF should be soldered in parallel with resistors R28, R88 and their complementary ones in another channel.

This was discovered when using transistors BC546\BC556 + 2SA1837\2SC4793.