At various computer speaker systems average price segment(in particular Microlab PRO2 and Thonet & Vander Dass) one common and very unpleasant drawback was noticed - when you plug something into a nearby outlet, loud annoying clicks are heard in the speakers. Which is especially not pleasing at night. It is convenient to turn the volume knob on computer speakers to a value close to the maximum in order to adjust it in the full range from the computer in the future. Which does not have the best effect on the volume of crackling noises. The speakers clicked especially loudly when the air purifier for soldering was turned off, but the reaction of the speakers to all sorts of small switching power supplies/chargers that were turned on and off from a nearby (and not only) outlet was also unpleasant. The identified problem is a consequence of the Chinese’s total savings on everything during design and production. The solution to the problem is to add to the scheme what was saved.
When examining the inside of the acoustics, it was noticed that there was no mains voltage interference filter. The amplifiers themselves in such devices are traditionally made on microcircuits with a built-in stabilizer, i.e., their entire power supply consists of a transformer, a diode bridge and a pair electrolytic capacitors(in my amplifiers their capacity is 4700 µF per arm).
To begin with, it was decided to install a surge protector. It will not solve the problem of clicks when turning on/off a fan in a nearby outlet (you can verify this by connecting the speakers to a high-quality external surge protector- the clicks do not disappear completely), but it certainly won’t be unnecessary, given the abundance of impulse noise in the socket. I didn’t bother too much with the filter and ordered one from China, like this one (you can look for similar filters on all sorts of aliexpress search query“EMI power amplifier filter”).
The filter is soldered into the power wire section. I didn’t attach it inside, I just placed it in a small box printed on a 3D printer so that it wouldn’t short out and electrocute someone..
The next simple and obvious way to improve power quality is to increase the capacity of the “electrolytes” to at least 10,000 - 15,000 uF. It should be taken into account that the starting currents when charging such containers will also increase, and diode bridge must have a good current reserve so that when turned on it does not get sick. Also, for better filtration, I added a choke to each arm (receiving a T-shaped LC filter). As a result, the following diagram was drawn:
And the boards are ordered:
Here, in each arm you can install up to five electrolytic capacitors with a capacity from 2200 μF to 4700 μF (with an operating voltage from 25 ... 63V) and a pair of non-polar capacitors. As the latter, I used Chinese 0.22 uF film ones, such as:
The board has input and output connectors, and both alternating voltage can be supplied to the input (in which case a diode bridge is installed) and already rectified voltage (if you plan to use the bridge already present in the amplifier).
The assembled board turned out like this:
Next, I removed the rectifier diodes from the amplifier board. In general, you can leave them, if it is not critical that a couple more volts of power will fall on them. Instead of diodes, I installed a couple more 100 µH chokes - they definitely won’t make things any worse. I secured the capacitor board in the case, the wires from the step-down transformer go to its input, the filter output goes to power the amplifier board. I also installed another FR157 diode at the filter outputs to shunt impulse noise (with cathodes to the positive), they made a significant contribution to suppressing clicks.
The result is that clicks when the neighbor’s fan is turned off began to occur less frequently, and their volume became noticeably less, they were no longer as annoying as initially. The loud sharp sound when turning on/off is no longer observed at all. Increasing the capacitance in the power supply filter results in smaller voltage drops, and at high volumes there should no longer be a feeling that the sound is failing.
Refinement of UMZCH with non-standard inclusion of op-amp
At one time, many radio amateurs repeated the AF power amplifier described by N. Troshin in the article “UMZCH with non-standard inclusion of an op-amp” (Radio. 1988. No. 6. pp. 55, 56). According to reviews from radio amateurs, the amplifier is simple, reliable in operation, easy to set up and has very good parameters. All this is true. However, it also has a small drawback. The thing is. that at low volume levels, the sound of a loudspeaker operating with this amplifier loses transparency and is noticeably distorted, especially when reproducing signals from acoustic instruments.
These distortions appear due to the low quiescent current of the output transistors.
Minor changes schematic diagram of the amplifier in question made it possible to increase the quiescent current of the output transistors, making it more stable and independent of temperature. For this purpose, diodes VD3-VD6 and resistor R7 were excluded from the amplifier circuit, and instead of them a complementary pair of low-power transistors VTT" and VT2" was introduced (see figure), which stabilizes the current through transistors VT2, VT3 and resistor R10. As a result, the quiescent current through transistors VT4 and VT5 is stabilized due to the resistors R11 and R12 included in their emitter circuits.
One of the newly introduced transistors VT" must have temperature contact with the output transistors. The easiest way for this purpose is to solder it to the terminals of these transistors. For example, the base terminal of transistor VT1" can be soldered directly to the base terminal of transistor VT4, and the terminals of its collector and emitter are thin connect the mounting wires respectively to the base of transistor VT2 and the emitter of VT2."
Another transistor VT2" should not have temperature contact with the output transistors of the UMZCH. To minimize the length of the connecting conductors, it should be installed on an additional getinaks plate next to the indicated transistors.
You can increase the stability of the UMZCH by connecting the circuit R1" and C1" to its output. The quiescent current of the output transistors is automatically set within 120... 150 mA. The UMZCH does not require additional adjustment.
When improving the stereo version of the UMZCH, it is useful to first modify one of the channels and. Only after assessing the sound quality do you begin to refine the second one.
It was very widely used by many music lovers and many still have it today.
However, even a superficial analysis of the circuit points to the ULF-50-8 module as the most weak link amplifier The module does not fully realize the capabilities of the pre-amplifier assembled on K157UD2 microcircuits.
I suggest option for remaking a complete amplifier Radiotekhnika U-101 in fact high quality amplifier for a household audio equipment complex. Distinctive features: high specifications and reliability with minimal intervention in the converted device, when all functionality original amplifier.
To replace the ULF-50-8 module, an UMZCH with low nonlinear distortion was selected. Its main technical characteristics:
nominal output power at a load with a resistance of 8 Ohms, 25 W;
harmonic coefficient in the frequency range 20-20000 Hz 0.03% (0.3% for ULF-50-8);
output voltage slew rate 40 V/µs.
Amplifier circuit for reworking Radio Engineering U101
The UMZCH consists of a two-stage voltage amplifier (op-amp DA1, DA2) and the power amplifier itself (VT1-VT4). Cascades on op-amps DA1, DA2 are powered from identical sources formed by elements VD1, VD2, R6, R7, C6, C7 and VD3, VD4, R14, R15, C13, C14. The midpoints of these power supplies are connected to a low-resistance voltage divider R5R12R20, connected to the output of the UMZCH, which ensures the supply of tracking potentials to the voltage amplifier stages. Circuits R16C8 and R19C10 filter the voltages supplying the first stages from nonlinear ripples generated by the signal in the power circuits of the output stage.
Problems with securing weekends UMZCH transistors did not occur on the Radiotekhnika radiator in accordance with the recommendations. The radiator has four metal plates insulated with mica spacers. There is enough space on each plate for three transistors, no modifications to the heatsink are required.
The only problem is the need to reduce printed circuit boards UMZCH, since there is not enough space in Radiotekhnika. Two UMZCHs should be assembled on two printed circuit boards, the width of which should not exceed 60 mm. The drawing of the voltage amplifier board (Fig. 3a) needs to be compacted in width to this size. This is not difficult to do if the K50-6 capacitors are replaced with K50-35 or other small ones. The drawing of the final stage board (Fig. 3.6c) will fit on a 60 mm wide board without changes (drawings are provided).
The first board is made 240 mm long and one voltage amplifier and two final stages are placed on it. Another voltage amplifier is placed on the second board.
The long board is attached to the Radiotekhnika radiator on 15 mm long stands so that it is placed vertically in the amplifier housing. The ULF-50-8 boards are first dismantled. The second voltage amplifier board is attached to a long board on 20 mm long stands on the side wall of the amplifier housing.
The UMZCH is connected to the ±26 V power supply of Radiotekhnika. The supply voltage ±30 V is not used. The output of the UMZCH is connected to the “Radio Engineering” protection board. The signal wires do not need to be confused (the same applies to the input wires).
Correctly assembled and connected UMZCHs begin to work immediately after turning on the power and do not require adjustment. I also recommend replacing the capacitors SZ, C4, C8, C9 on the Radiotekhnika rectifier board. They have probably already lost some of their capacity (dried out), so it is better to replace them with new ones with a capacity of 4000-5600 µF.
All functionality of Radiotekhnika is retained after the modification. The sound of the converted amplifier can be characterized by epithets: clean, transparent, rich with clear localization of sound sources. It's much better than the original amp and noticeably better than the AKAI FD-1 I have.
When operating the converted amplifier at sound speakers with LF heads of type 10GD-30 or 25GD-26, characteristic clicks are heard at high volumes. This occurs due to insufficient rigidity of the caps covering the magnetic gaps of the heads. The caps should be replaced with more rigid ones. The original amplifier has much worse performance, so no clicks were observed.
Y. M. Kogut, Lviv region.
Literature
1. Amplifier “Radio engineering U-101 stereo”. Manual.
2. Ageev A. UMZCH with low nonlinear distortions//Radio.-1987. -No. 2.-P.26-29.
Here are drawings of printed circuit boards and placement of elements published in Radio magazine, to which the author of the article refers.
RADIOAMATOR No. 10, 2001
Introductory part
Before starting the modernization, I’ll clarify some details:
- my knowledge of analog circuitry is limited;
- work to change the circuit design will be carried out in accordance with research work Self Douglas on the design of the UMZCH;
- all intermediate conclusions and adopted circuit design decisions are also based on this work, including the predicted numerical indicators of the UMZCH;
- my decision is not the ultimate truth, this modernization is one of the possible options;
- where necessary, I will indicate that the decision I made or the problem that arose is not completely clear to me;
- I will try to write the text as simply as possible.
Formulation of the problem
The ULF-50-8 block (hereinafter simply the ULF block) was a very common design in Soviet times and is a completely typical circuit design solution to the problem of constructing an UMZCH for that time. If you look at the indicators, for example, of an amplifier, then the harmonic coefficient is stated to be no more than 0.3%, which is actually a bit too much for transistor technology. I intend to focus on this indicator here, yes, I know that there are tube amplifiers that are specially designed with the desired distortion (the notorious “tube” sound, i.e., for the most part, the presence of a large second harmonic), I am also aware that what is and transistor amplifiers, designed with a similar approach to distortion. Based on the above, only the option will be considered when the UMZCH introduces as little distortion as possible into sound signal, if you need a “tube” sound, then just connect preamplifier with the necessary “distortions”. Let us designate the level of harmonic (non-harmonic too) distortion in the audio range as a number less than or comparable to 0.01%, which by the way, as I remember, is significantly less than the general requirements for Hi-Fi equipment of 0.1%. Douglas in his work achieves numbers of 0.001% or less; for the simplicity of the task, we do not need such values, and there are suspicions that it will be impossible to hear the difference. I will also explain that we will not consider the composition of spectral distortions, because with such small distortion values you can hear them to an ordinary person impossible, if possible at all.
Food for thought. I read somewhere in an old book from the 50s that an ordinary, untrained person begins to distinguish well harmonic distortions starting from 0.5%.
Now we can move on to analyzing the main complaints about the native ULF unit. As I saw, there are two options for such blocks. We will consider one of them (Fig. 1).
Rice. 1. Unch-50-8
So, there are two main problems in the block circuitry:
- Although the differential cascade is powered by a current source, the currents in the arms are not the same in practice (for example, the currents in the collectors of transistors VT2, VT4);
- the presence of the Early effect (in simple terms, the base current is not constant and changes itself due to changes in the collector voltage) in the voltage amplifier transistor we need VT10 c) the absence of a current source in the differential stage VT7, VT10, although there is a current mirror there (I’m not here I’m sure that in the indicated diagram this is not a current mirror, but in some sources there is a different switching on of the transistor VT3 and a current mirror is obtained, therefore it is not clear what and how, my bell tower says that there is a mirror).
Solving these problems will significantly reduce distortions in the UMZCH (tens of times). You can read more details about the causes and consequences of these problems in Self Douglas’s book on our disk; if possible, I will give his considerations somewhere when I make circuit design decisions.
A solution to these problems is shown in Fig. 2. I’ll say right away that this is only one of the options, there are others.
Rice. 2. Upgrade option ULF-50-8
Now let's try to figure out what, how and why was changed. All changes, unless specifically stated, I have marked in different colors.
Green Zone"
To equalize the currents in the diff arms. cascade VT2,VT4 a current mirror was introduced. According to Douglas, this solution sharply reduces the second harmonic in the distortion spectrum and subsequent harmonics as well. An additional advantage of this solution is the stabilization of the DC component at the output of the UMZCH. I achieved output values of 20 mV without any additional effort. If the DC voltage at the output of a normally operating unit is more than 50 mV, then perhaps one of the transistors in the current mirror is very different in coefficient. h21e, it’s worth picking them up and replacing them.
The part numbers are as follows:
R5, R8 from 100 to 200 Ohm, power > 0.125 W.
VT5, VT6 any n-p-n with Uke > 50 V and h21e > 40.
For ease of installation on the original board of the unit, I took the KT961B.
Turquoise "zone"
Here we use a classic cascode circuit on VT7,VT10. The current generator is standard on VT8.
Food for thought. An amplifier that is highly praised for good sound, a cascode circuit is also used in the voltage amplifier. True, there is no current mirror in the diff. cascade, but there is a tuning resistor that sets the level to a constant state. at the output of the UMZCH (evens out the currents of the cascade arms).
I changed the value of resistor R16 so that the quiescent current in the cascade was equal to the quiescent current in the original circuit in Fig. 1 (the value can be set by connecting resistors in parallel, I think you can plug in 160 Ohms without problems).
The part numbers are as follows:
VD7, VD8 - any diodes, for example the KD522 or 1N4148 series.
VD9 – 4.7-5.1 V zener diode and starting stabilization current from 5 mA. I don’t know exactly the lower limit of this voltage; experience in simulating this circuit in Multisim has shown that the min. The stabilization voltage value is somewhere in this area. If anyone knows, tell me.
R10 – 30-51 kOhm or so, not very important, power from 0.125 W.
VT7, VT10 - high voltage, for example, KT940A, KT961A.
C4 – Miller capacitance, it’s hard to say which is better, my version of 33-47-51 pF is quite workable.
Red "zone"
In this “zone” all changes are practically cosmetic. Resistors in the emitters of the output transistors are 0.5 Ohms - a bit too much, you can reduce them to 0.33 Ohms by soldering in parallel (from the side of the board tracks) another 1 Ohm resistor. Douglas recommends even less, but the board itself does not allow you to run up too much. Zobel output circuit (R41, C13) - replaced with a more conventional and familiar one.
Unreported changes
I recommend capacitor C3 in the circuit feedback increase it at least twice and put a non-polar electrolyte there. A more acceptable option is to increase the capacitance to 470-500 µF and connect in parallel to it any film or ceramic capacitor with a capacitance of 0.1 µF.
That's all the changes. Now some of my practical comments:
- all of these changes can be mounted on the board of the ULF unit, with difficulty, but it fits;
- The print quality is usually so-so, so be careful with soldering, the tracks come off quickly;
- there is a problematic track in the wiring of the block, this is the track between the emitters of the output transistors and the output of the block, which is too thin for the currents flowing through it, it needs to be strengthened (you can use an MGTF wire);
- similarly, it is better to strengthen the tracks in the relay protection board;
- The quiescent currents of the output stage should not be exceeded (the range is indicated) and output transistors, for example KT805, are not selected according to similar characteristics at the factory, so their quiescent currents can vary greatly.
Distortion Measurements
Formally, such a modification, even on such transistors, should easily lead to a reduction in the level of harmonic distortion to 0.01% or comparable values. I haven't taken any measurements yet - I'm too lazy. Now, if someone runs them, it will be cool (suddenly they say that I lied with three boxes!), but I can say the following, in Multisim 13, using analogues of domestic transistors, I received 25 W at 4 Ohms of harmonic distortion at a frequency of 20 kHz 0.022%.
P.S. I know that there is a whole thread on the Vegalab forum, this modernization works quite well, but in the one presented there will certainly be less distortion.
It often happens that they will ask you to look at the amplifier to improve the sound a little.....The job is done over the weekend, but there are no traces left.
Then you remember - what was it?
Therefore, at least for myself (maybe it will be useful to someone), in each message I will describe a little what was done in a particular amplifier and, if possible, attach photos.
I will also provide links here if any amplifier is described in another blog or forum.
This has been asked more than once, so in the introduction I’ll say right away: I won’t Please include oscillograms, amplifier spectra and mention about adjusting the quiescent current of the mind. This is all done using the usual measurement and adjustment methods... specialists know how and what result needs to be achieved in order for the amplifier to play well and they know well what the wrong bias setting can lead toand temperature compensation quiescent currents.
But it’s better for non-specialists not to know this.
http://www.electronics.ru/files/article_pdf/3/article_3164_325.pdf
http://www.rlocman.ru/shem/schematics.html?di=278201
Disclaimer: the advisor does not bear any responsibility for his advice.
E.O.S. AE-90F LE
The amplifier is made to last.
The noise and interference at the output are excellent (with the exception of the rear right channel - here you can feel the proximity of the power supply,... but everything is within normal limits).
The opamps are at the input OPA2604, and the filters are AD712JN.
But I didn’t like the sound of the piano, violins, drums.
I put the socket on the preliminary stage and plugged in all the op-amps that I had (listened to it in full band, without filters).
The power supply for the preliminary stages and filters is bipolar 14 volts.... it’s a pity that you can’t try 8066.
In general, my result: if for mid, then AD823, for HF, LM4562 or from the LME49XXX series works well.
I settled on LM4562 and LME49722..... most likely I’ll leave it that way... I won’t do anything else there. It played more smoothly.. now there is no sharpness, the resolution has increased.. the stage has gone deeper, the aftersound has become better..
For power supply, I additionally blocked the op-amp with 0.1 µF NPO ceramics, since I don’t see what’s on the back side.
In theory, you need to think about improving the UM driver... but with filters, I haven’t decided what to do yet.
Genesis STEREO 100
Well-known amplifier, quite good.
The task is to make a direct (irrevocable).
Will work for mids.
I’m checking... it works, but in one channel on a rectangle (already at medium power), excitation is at the tops of the pulses, and at maximum power(at 4 Ohms) - an obvious drawdown in amplitude..... understandable, we'll look into it.
I will also redo the well-known feature of the amplifier (who is in the know), that for powering the preliminary stages in the negative arm there is a 79L12, and in the positive arm there is an ordinary parametric stabilizer.... and so on, but without fanaticism.
At the beginning I dealt with the excitation/drawdown... most likely one or two of the output 142/147 leaked into the mind. Fortunately, they are inexpensive, so I selected the transistors and replaced all six (three TIP142T and three TIP147T). There is no more excitement, the mind works stably.
I replaced the power supply capacitors at the input and after the inductor and the resistor, blocked the primary and secondary with ceramics.
I unsolder the op-amp that worked for filtering, and unsolder all the elements that are somehow connected to the input op-amp, cut a couple of tracks, and change the input to LME49740. I installed non-polar Nichicon coupling capacitors... tried polar ones, heard the color, so ES passed the best in terms of neutrality.
I replace 79L12 with 7912 (it’s clear that this little Elka can heat up to 150 degrees, but it working temperature for 80 I like it... it’s clear that the task of extinguishing from 33 to 12 volts at a current of 18 mA is not fun...), then I unsolder the 12 volt zener diode and install a 7812 stabilizer. Before the stabilizers I install a low-pass filter (RC chain 27 Ohm and 47 uF * 35), after stabilizers 100.0 * 16 Silmik2, everything is blocked by ceramics and film 0.1.
At the bottom of the board, I connect the outputs of the preliminary op-amp to the inputs of the PA with wires (six new PA transistors are visible on the right):
General view of the amplifier board:
Well, what can I say... the sound has improved in better side in all respects.
I liked it - it sounds clear and clean.
Genesis Five Channel Amplifier.
Task: all channels in direct (without fanaticism and irrevocably).
It’s a bit out of scale for a weekend design… but I’ll write it here nonetheless.
The amplifier is working.
Let's open it.
The board is beautiful, it has passive and active radio elements installed.
Two power supplies, not stabilized, from one SG3525.
One power supply works for the sub channel, the second power supply works for 4 channels - this is good.
It is immediately clear that one of the input capacitors for power supply 1000.0*16 has swelled, but the electrolyte has not leaked out yet... this is also good.
I check for functionality - it plays... one might say that it enhances it. After listening, I was left with a painful impression - there are problems in the sound in the form of slurred bass, no high frequencies and no soapy mids (channels 3 and 4) and sudden movements of the KIZ to the left and up in all channels... the sub channel seems to be fine, but somehow dry.
Instrumental check:
In channels 1-4 at the output you can see rapid jumps in interference, and in channel 4 the output is constant at 0.2 volts.
The power supply for the preliminary stages of the subwoofer is bipolar 12 volts, and 1-4 channels are bipolar 18 volts. True, positive 18 volts shows 19.5 volts and seems to jump even higher...
Okay, we'll figure it out.
I'll start describing the modification in direct from the sub channel.
The sub channel in the drain provides the ability to remotely control the volume of the sub, so the amplifier has THAT2181LC with a buffer on a single op-amp 071. We unsolder them (red daws) and bypass them with jumpers. At the input of the sub channel it is 074. We only need two op-amps - one input and a second level regulator, there is no desire to make such an adapter, so I installed a quad OPA4134 on the adapter, .... and connected two unused op-amps as usual: negative and output, and positive to ground. I change the input coupling capacitor to Wima 0.47 µF.
Also, in the part of the board where the preliminary cascades of 1-4 channels are located, we solder the adder of the sub channel, which consists of an assembly of 100 kOhm resistors and a dual op-amp 072. We also cut off the track on the edge of the board along which the summed signal goes to the input of the sub channel in order to exclude mutual crosstalk from/to sub.
Now we take on channels 1-4. General principle the same - we unsolder the op-amps that are used for the filter, get rid of all unnecessary connections in the form of resistors and capacitors, leave only the input op-amp and one op-amp for the Level regulator.
The input op amps in the amplifier are used in L housing.
It takes a long time to look for such an adapter, and I’m afraid a standard DIP adapter won’t fit, you only need SMD, since the installation is very tight due to the resistor assemblies. Easier to make from a universal holey scarf. This is how it all turned out:
This is what part of the modified input board for channels 1-4 looks like:
The replaced input coupling capacitors Elna Cerafine 10*16, op-amps LME49740 and LME49860 are marked with red checkmarks.
In the quad op-amp 49740, only two op-amps are used, the other two are neutralized.
I replace electrolytes in the power supply of the preliminary stages:
I examine stabilizers 7818 and 7918 in more detail. I discover a problem in the form of an alternating contact (most likely a crack in the track) of the middle terminal (negative) of the stabilizer 7818 with the ground. I solder the middle pin to the ground plane of the board. Turned it on - now the positive and negative power supply of the preliminary stages is the same and amounts to 18 volts (exactly +18.3 and -18.4 volts).
This is what the back side of the board looks like (already washed from the factory flux):
We fiddle a little with the power supply, along the way correcting the detected defects in the form of a swollen input capacitor:
To prevent the signal from channels 1 and 2 from passing into channels 3 and 4 accidentally when disconnecting the interblocks of channels 3 and 4, you need to cut the tracks on the board indicated by red arrows (pre-output 1, 2 channels), since the tracks go to the red points (input 3 ,4 channels).
And further.
It is obvious only in this type of amplifiers: the rectifier diodes of the sub channel are not pressed at all (there is no clamping screw provided) to the radiator... when disassembling it is clear that the diodes touch the radiator only with the edge of the case - a 1mm strip.
Therefore, during assembly, you need to apply a thick layer of thermal paste to these diodes.
*******************************************
Now the amplifier sounded smooth and clear.
Second trip.
A year later the amplifier returned.Somehow it was specifically filled with water while it was on......
After the inspection, the fuses burned out and the terminals inside the fuse holders burned out, the board under one power supply transistor of four channels burned out and all the base resistors burned out, the legs of the capacitors at the input (after the chokes) burned out, in the first channel the printed chokes in the bases of one arm burned out, 22 ohm ones burned out resistors to RCA grounds.... etc. and so on.
After restoration and switching on, there is a constant in the sub channel at the output and the output transistors get very hot.
Total: 16 transistors, not counting capacitors, burnt-out board, etc.:
This is how the board burned out (this is after restoration):
Modification of fuse holders:
I tried it - everything works, although from experience - perhaps the defects will not end there... who knows what other transistors leaked as a result of the overload.
E.O.S. AE-620T
To understand the incomprehensible aspects of the amplifier's operation and make a complete directive.
Let's look: before the clip, the peak is 68 volts across the channels, 136 volts in the bridge (4 ohm load, supply 12.2 volts, 36.2 volts on the secondary).
Flat - the bass rise remains working.
We turn on L.P - it works, but if we set the phase adjustment to closer to zero, the output power drops three times.
At the outputs there is a constant voltage of approximately 22 millivolts, the noise is approximately 10 mV and purely - interference from the power supply is not visible in the noise.
The power supply is not stabilized, with an input of 12 volts - plus/minus 35.7 volts on the secondary, with an input of 14.7 volts - 43.3 volts on the secondary.
For the preliminary power stages there is a parametric stabilizer on zener diodes.... with inputs above 14 volts, the negative voltage of the pre-stage jumps out beyond 15.6 volts, naturally the stub does not heat up like a child... (the value of the ballast resistors of 390 Ohms is shown on the board, and cost 270 ohms.....)
There are four resistors on the board, soldered on. Why and for what - there is no time to figure it out.
Input op-amp - AD712.... in filters - JRC5532.
At idle it doesn’t heat up at all... at 12.2 volts the current is 0.62 A.
I change the input op-amp to 5532, unsolder all the op-amps of the filters and the phase shifter, leaving only the outermost op-amp to invert the signal so that the amplifier can work as a bridge.
I connect them with jumpers (having previously trimmed the tracks so that the signal does not go where it is not needed), I block the power supply op-amp with 0.1 µF capacitors.
I change the separators to Nichicon Muse ES 22.0 * 50 volts, in the pre-power supply I set them to 1000.0 * 25 volts (they were 470.0).
So, since the load of the parametric stabilizer has decreased, I recalculate the ballast resistor and set it to five watts at 560 Ohms.
But the negative zener diode still goes beyond the stabilization zone and gets hot. I change the negative zener diode (at the same time and the positive one... just in case) - and everything becomes normal - with an input of 14.7 volts (secondary 43.3) on the zener diodes of 15.2 volts (and with an input of 12 volts (secondary 35, 7) - on zener diodes of 14.8 volts)
The output noise has decreased to 5 mV.
Now the amplifier works direct and can be bridged... just right for bass.
I smiled: when unwinding the amplifier, under the screws of the clamping strips of the powerful transistors there were two washers and one Grover... but under one screw there were three washers, and there was no Grover Washer. I think these collectors are not good...
When I unscrewed and removed the board from the case, the grower found itself pressed under one of the output transistors.
So everything is fine, the Grover washer was found.
It’s good that the insulating gasket is quite thick and was not pressed onto the body by the Grover washer.
GENESIS Profile Four Ultra
Task: do it in full direct.
The amplifier will operate in two-way (more precisely: mids and mid/high frequencies passively), so pay attention to the midbass region.
Let's look: the power supply is unstabilized, with an input of 11.7 - secondary 23 and 32 volts, with an input of 14.5 - secondary 28 and 40 volts.
Constant on channels 1 and 3 at 40-45 mV, on channel 2 210 mV, on channel 3 130 mV.
Quite large pulsations with the power supply frequency at the PA outputs (gain at maximum) - 1-1.5 volts.
The power supply of the preliminary stages is surprising, but very clean, but the bipolar 37 volts, which goes to the pre-output stages of the PA - in the negative voltage there is an obvious saw of 3 volts, in the positive voltage a little less - 1 volt....
1. solder 4 resistors of 100 kOhm each
2. solder two 2.55 kOhm resistors
3. solder two 150 nF capacitors
4. Cut 2 tracks on the back of the board that lead to the HI Level IN connector
5. Place two jumpers (indicated in blue)
6. Place decoupling capacitors at the input of Silmic II - 10 µF * 25V
7. change two capacitors 100 uF * 16v
8. cut two tracks that go to 3.4 channels (marked with red circles 1 and 2)
9. Change the pre-chip to OPA1654
10. Solder the TL074 chip, which is located on the filters
11. We change four capacitors in the PA (marked with red daws), the nominal value is the same 10 μF * 16V.
The constant at the output of channels 1 and 2 became 36 mV and 102 mV.
The ripples at the output of PAs 1 and 2 channels have become two times less.
We turn it on, warm it up and then listen and compare:
Very clearly the depth of the stage has become twice as deep, clarity has appeared in the midrange and high frequencies.
We are trying to fight the “saw” in the power supply to the pre-output cascades of the PA.
We solder two 100 µF*50V capacitors.
We measure - one is practically dead and only a third of the capacity remains in it, the second is half... naturally, the ESR shows 1.1 Ohm and 0.5 Ohm, respectively.
We install new capacitors of the same rating (low-impedance, with an ESR below 0.1 Ohm) and look at the subject of the saw - now the “saw” is only 0.2-0.3 volts in amplitude.
We continue in the same spirit:
... "blow off" the extra microcircuits (marked with red circles) and resolder the input microcircuit in channels 3 and 4 on the OPA1654.
We change the capacitors in the output stages of the PA of channels 3 and 4:
We solder 0.22 uF, replace the capacitors in the power filter of the preliminary stages with 220.0 * 16v.....
The yellow lines draw the path of the jumpers, which are soldered on the back side of the board...
As a result, the preliminary stages on the board after modification look like this:
The jumpers on the back of the board and the cut tracks (1-6) look like this:
The constant at the output of channels 3 and 4 became 35 mV and 60 mV.
The noise at the PA inputs has decreased by at least two times.
Now a little about the power supply.
It is made according to the usual scheme on SG3525.
In stock, the power supply operates at a frequency of 41 kHz (period 24.4 microseconds).
We change the resistor on leg 6 SG3525 with a nominal value of 1.8 kOhm to 1.5 kOhm.... the frequency became 47 kHz.
As a result of all the above modifications (gain at maximum), the interference (pulsations of the power supply) at the output of the PA of all channels decreased by five times.
Notes for yourself:
- I would also like to replace 79M12 and 78M12 with 15 volt ones (they are in the D-PAK package, but can also be soldered into TO-220).
- to reduce ripple at 37 volts, replace two 10 uF * 50 volt pieces (located at the edges of the PA drivers) with something with low ESR ... you can use 47 uF * 50 volts.
There was an opportunity to look into the amplifier again.
I changed the capacitors 10 uF * 50 volts to 100.0 * 50 volts. Steel ripples are at the level of 10 mV.... in principle, so little is not necessary, that is, it needs to be changed, but it is enough to set 47.0 * 50 volts... but it must be 105 degrees.
Loading...
