29.06.2020
Power supply circuit for a power amplifier. Switching power supply for amplifiers. Switching power supply circuit
Making a good power supply for a power amplifier (UPA) or other electronic device is a very responsible task. The quality and stability of the entire device depends on the power source.
In this publication I will tell you about making a simple transformer power supply for my homemade amplifier low frequency power "Phoenix P-400".
Like this, not complex block The power supply can be used to power a variety of low frequency power amplifier circuits.
Preface
For the future power supply unit (PSU) for the amplifier, I already had a toroidal core with a wound primary winding of ~220V, so the task of choosing “switching PSU or based on a network transformer” was not present.
Switching power supplies have small dimensions and weight, high output power and high efficiency. Power supply based on a network transformer - has heavy weight, is easy to manufacture and set up, and you don’t have to deal with dangerous voltages when setting up the circuit, which is especially important for beginners like me.
Toroidal transformer
Toroidal transformers, in comparison with transformers with armored cores made of W-shaped plates, have several advantages:
- less volume and weight;
- higher efficiency;
- better cooling for windings.
The primary winding already contained approximately 800 turns of 0.8 mm PELSHO wire; it was filled with paraffin and insulated with a layer of thin fluoroplastic tape.
By measuring the approximate dimensions of the transformer iron, you can calculate its overall power, so you can estimate whether the core is suitable for obtaining required power or not.
Rice. 1. Dimensions of the iron core for the toroidal transformer.
- Overall power (W) = Window area (cm 2) * Sectional area (cm 2)
- Window area = 3.14 * (d/2) 2
- Sectional area = h * ((D-d)/2)
For example, let's calculate a transformer with iron dimensions: D=14cm, d=5cm, h=5cm.
- Window area = 3.14 * (5cm/2) * (5cm/2) = 19.625 cm2
- Cross-sectional area = 5cm * ((14cm-5cm)/2) = 22.5 cm 2
- Overall power = 19.625 * 22.5 = 441 W.
The overall power of the transformer I used turned out to be clearly less than I expected - about 250 watts.
Selection of voltages for secondary windings
Knowing the required voltage at the output of the rectifier after the electrolytic capacitors, you can approximately calculate the required voltage at the output of the secondary winding of the transformer.
Numeric value DC voltage after the diode bridge and smoothing capacitors will increase by approximately 1.3..1.4 times compared to the alternating voltage supplied to the input of such a rectifier.
In my case, to power the UMZCH you need a bipolar DC voltage - 35 Volts on each arm. Accordingly, an alternating voltage must be present on each secondary winding: 35 Volts / 1.4 = ~25 Volts.
Using the same principle, I made an approximate calculation of the voltage values for the other secondary windings of the transformer.
Calculation of the number of turns and winding
To power the remaining electronic units of the amplifier, it was decided to wind several separate secondary windings. A wooden shuttle was made to wind the coils with enameled copper wire. It can also be made from fiberglass or plastic.
Rice. 2. Shuttle for winding a toroidal transformer.
Winding was done with enameled copper wire, which was available:
- for 4 power windings UMZCH - wire with a diameter of 1.5 mm;
- for other windings - 0.6 mm.
I selected the number of turns for the secondary windings experimentally, since I did not know the exact number of turns of the primary winding.
The essence of the method:
- We wind 20 turns of any wire;
- We connect the primary winding of the transformer to the ~220V network and measure the voltage on the wound 20 turns;
- We divide the required voltage by that obtained from 20 turns - we will find out how many times 20 turns are needed for winding.
For example: we need 25V, and from 20 turns we get 5V, 25V/5V=5 - we need to wind 20 turns 5 times, that is, 100 turns.
The calculation of the length of the required wire was done as follows: I wound 20 turns of wire, made a mark on it with a marker, reeled it off and measured its length. I divided the required number of turns by 20, multiplied the resulting value by the length of 20 turns of wire - I got approximately the required length of wire for winding. By adding 1-2 meters of reserve to the total length, you can wind the wire onto the shuttle and safely cut it off.
For example: you need 100 turns of wire, the length of 20 wound turns is 1.3 meters, we find out how many times 1.3 meters each need to be wound to get 100 turns - 100/20 = 5, we find out the total length of the wire (5 pieces of 1, 3m) - 1.3*5=6.5m. We add 1.5 m for reserve and get a length of 8 m.
For each subsequent winding, the measurement should be repeated, since with each new winding the wire length required by one turn will increase.
To wind each pair of 25 Volt windings, two wires were laid in parallel on the shuttle (for 2 windings). After winding, the end of the first winding is connected to the beginning of the second - we have two secondary windings for a bipolar rectifier with a connection in the middle.
After winding each pair of secondary windings to power the UMZCH circuits, they were insulated with thin fluoroplastic tape.
In this way, 6 secondary windings were wound: four for powering the UMZCH and two more for power supplies for the rest of the electronics.
Diagram of rectifiers and voltage stabilizers
Below is a schematic diagram of the power supply for my homemade power amplifier.
Rice. 2. Schematic diagram of the power supply for a homemade low-frequency power amplifier.
To power the LF power amplifier circuits, two bipolar rectifiers are used - A1.1 and A1.2. Rest electronic components The amplifier will be powered by voltage stabilizers A2.1 and A2.2.
Resistors R1 and R2 are needed to discharge electrolytic capacitors when the power lines are disconnected from the power amplifier circuits.
My UMZCH has 4 amplification channels, they can be turned on and off in pairs using switches that switch the power lines of the UMZCH scarf using electromagnetic relays.
Resistors R1 and R2 can be excluded from the circuit if the power supply is permanently connected to the UMZCH boards, in which case the electrolytic capacitors will be discharged through the UMZCH circuit.
KD213 diodes are designed for a maximum forward current of 10A, in my case this is enough. The D5 diode bridge is designed for a current of at least 2-3A, assembled from 4 diodes. C5 and C6 are capacitances, each of which consists of two capacitors of 10,000 μF at 63V.
Rice. 3. Schematic diagrams of DC voltage stabilizers on microcircuits L7805, L7812, LM317.
Explanation of names on the diagram:
- STAB - voltage stabilizer without adjustment, current no more than 1A;
- STAB+REG - voltage stabilizer with regulation, current no more than 1A;
- STAB+POW - adjustable voltage stabilizer, current approximately 2-3A.
When using LM317, 7805 and 7812 microcircuits, the output voltage of the stabilizer can be calculated using a simplified formula:
Uout = Vxx * (1 + R2/R1)
Vxx for microcircuits has the following meanings:
- LM317 - 1.25;
- 7805 - 5;
- 7812 - 12.
Calculation example for LM317: R1=240R, R2=1200R, Uout = 1.25*(1+1200/240) = 7.5V.
Design
This is how it was planned to use the voltage from the power supply:
- +36V, -36V - power amplifiers on TDA7250
- 12V - electronic volume controls, stereo processors, output power indicators, thermal control circuits, fans, backlighting;
- 5V - temperature indicators, microcontroller, digital control panel.
The voltage stabilizer chips and transistors were mounted on small heatsinks that I removed from non-working computer power supplies. The cases were attached to the radiators through insulating gaskets.
The printed circuit board was made of two parts, each of which contains a bipolar rectifier for the UMZCH circuit and the required set of voltage stabilizers.
Rice. 4. One half of the power supply board.
Rice. 5. The other half of the power supply board.
Rice. 6. Ready-made power supply components for a homemade power amplifier.
Later, during debugging, I came to the conclusion that it would be much more convenient to make voltage stabilizers on separate boards. Nevertheless, the “all on one board” option is also not bad and is convenient in its own way.
Also, the rectifier for UMZCH (diagram in Figure 2) can be assembled by mounted mounting, and the stabilizer circuits (Figure 3) in the required quantity can be assembled on separate printed circuit boards.
The connection of the electronic components of the rectifier is shown in Figure 7.
Rice. 7. Connection diagram for assembling a bipolar rectifier -36V + 36V using wall-mounted installation.
Connections must be made using thick insulated copper conductors.
A diode bridge with 1000pF capacitors can be placed separately on the radiator. Installation of powerful KD213 diodes (tablets) on one common radiator must be done through insulating thermal pads (thermal rubber or mica), since one of the diode terminals has contact with its metal lining!
For the filtering circuit (electrolytic capacitors of 10000 µF, resistors and ceramic capacitors 0.1-0.33 µF) you can: a quick fix assemble a small panel - a printed circuit board (Figure 8).
Rice. 8. An example of a panel with slots made of fiberglass for mounting smoothing rectifier filters.
To make such a panel you will need a rectangular piece of fiberglass. Using a homemade cutter (Figure 9), made from a hacksaw blade for metal, we cut the copper foil along its entire length, then cut one of the resulting parts perpendicularly in half.
Rice. 9. A homemade cutter made from a hacksaw blade, made on a sharpening machine.
After this, we mark and drill holes for the parts and fastenings, clean the copper surface with fine sandpaper and tin it using flux and solder. We solder the parts and connect them to the circuit.
Conclusion
This simple power supply was made for a future homemade audio power amplifier. All that remains is to supplement it with a soft start and standby circuit.
UPD: Yuri Glushnev sent a printed circuit board for assembling two stabilizers with voltages +22V and +12V. It contains two STAB+POW circuits (Fig. 3) on LM317, 7812 microcircuits and TIP42 transistors.
Rice. 10. Printed circuit board for voltage stabilizers for +22V and +12V.
Download - (63 KB).
Another printed circuit board designed for the STAB+REG adjustable voltage regulator circuit based on LM317:
Rice. 11. Printed circuit board for an adjustable voltage stabilizer based on the LM317 chip.
Many people know how much I like to deal with different power supplies. This time I have a somewhat unusual power supply on my desk, at least I haven’t tested one yet. And by and large, I’ve never seen reviews of power supplies of this type before, although the thing is interesting in its own way and I’ve made similar power supplies myself before.
I decided to order it out of pure curiosity, I decided that it might be useful. However, more details in the review.
In general, it’s probably worth starting with a short lyrical introduction. Many years ago, I was quite keen on audio technology, and went through both completely homemade versions and “hybrids”, which used PAs with a power of up to 100 Watts from the store Young technician, and half-disassembled Radiotekhnika UKU 010, 101 and Odyssey 010, then there was Phoenix 200U 010S.
I even tried to assemble Sukhov’s UMZCH, but something didn’t work out then, I don’t even remember what exactly. 
The acoustics were also different, both homemade and ready-made, for example Romantika 50ac-105, Cleaver 150ac-009. 
But most of all I remember Amfiton 25AC 027, although they were slightly modified. Along with minor changes in the circuit and design, I replaced the original 50 GDN speakers with 75 GDN ones.
This and the previous photos are not mine, since my equipment was sold a long time ago, and then I switched to Sven IHOO 5.1, and then generally began to listen only to small computer speakers. Yes, this is such a regression. 
But then thoughts began to wander in my head, to do something, for example, a power amplifier, perhaps just like that, perhaps to do everything differently. But in the end I decided to order a power supply. Of course I can do it myself, moreover, in one of the reviews I not only did it, but also posted detailed instructions, but I’ll come back to this later, but for now I’ll move on to the review.
I'll start with a list of declared technical characteristics:
Supply voltage - 200-240 Volts
Output power - 500 Watt
Output voltages:
Basic - ±35 Volts
Auxiliary 1 - ± 15 Volt 1 Ampere
Auxiliary 2 - 12 Volt 0.5 Ampere, galvanically isolated from the rest.
Dimensions - 133 x 100 x 42 mm
The channels ± 15 and 12 Volts are stabilized, the main voltage ± 35 Volts is not stabilized. Here I will probably express my opinion.
I am often asked which power supply to buy for one or another amplifier. To which I usually answer - it’s easier to assemble it yourself based on the well-known IR2153 drivers and their analogues. The first question that follows after this is that they don’t have voltage stabilization.
Yes, personally, in my opinion, stabilizing the supply voltage of the UMZCH is not only unnecessary, but sometimes even harmful. The fact is that a stabilized power supply usually makes more noise at HF and, in addition, there may be problems with the stabilization circuits, because the power amplifier does not consume energy evenly, but in bursts. We listen to music, not just one frequency.
A power supply without stabilization usually has a slightly higher efficiency, since the transformer always operates in optimal mode, has no feedback and is therefore more similar to a regular transformer, but with lower active resistance of the windings.
Here we actually have an example of a power supply for power amplifiers.
The packaging is soft, but wrapped in such a way that it is unlikely to be damaged during delivery, although the confrontation between the post office and sellers will probably be eternal. 
Externally it looks beautiful, you can’t really complain. 
The size is relatively compact, especially when compared with a conventional transformer of the same power. 
More clear sizes are available on the product page in the store. 
1. There is a connector installed at the input of the power supply, which turned out to be quite convenient.
2. There is a fuse and a full-fledged input filter. But they forgot about the thermistor, which protects both the network and the diode bridge with capacitors from current surges, this is bad. Also in the area of the input filter there are contact pads that must be closed to transfer the power supply to a voltage of 110-115 Volts. Before turning on for the first time, it is better to check whether the sites are closed if your network is 220-230.
3. Diode bridge KBU810, everything would be fine, but it does not have a radiator, and at 500 watts it is already desirable.
4. The input filter capacitors have a declared capacitance of 470 µF, but the actual capacitance is about 460 µF. Since they are connected in series, the total input filter capacitance is 230 µF, not enough for an output power of 500 watts. By the way, the board requires the installation of one capacitor. But in any case, I would not recommend raising the container without installing a thermistor. Moreover, to the right of the fuse there is even a place for a thermistor, you just need to solder it and cut the track under it. 
The inverter uses IRF740 transistors, although they are far from new transistors, but I have also used them widely in similar applications before. Alternatively, IRF830.
The transistors are installed on separate radiators; this was done partly for a reason. The radiators are connected to the transistor body, not only at the mounting location of the transistor itself, but also the mounting pins of the radiator are connected on the board itself. In my opinion, this is a bad decision, since there will be excess radiation into the air at the conversion frequency; at least I would disconnect the lower transistor of the inverter (in the photo it is the farthest one) from the radiator, and the radiator from the circuit.
An unknown module controls the transistors, but judging by the presence of a power resistor, and just my experience, I think that I won’t be much mistaken if I say that there is a banal IR2153 inside. However, why to make such a module remains a mystery to me. 
The inverter is assembled using a half-bridge circuit, but the middle point is not the connection point of filtering electrolytic capacitors, but two film capacitors with a capacity of 1 μF (in the photo, two parallel to the transformer), and the primary winding is connected through a third capacitor, also with a capacity of 1 μF (in the photo, perpendicular to the transformer) .
The solution is well-known and convenient in its own way, since it makes it very easy not only to increase the capacity of the input filter capacitor, but also to use one at 400 Volts, which can be useful when upgrading. 
The size of the transformer is very modest for the declared power of 500 watts. Of course, I will also test it under load, but I can already say that in my opinion its real long-term power is more than 300-350 watts.
On the store page, in the list of key features, it was indicated -
3. Transformers 0.1 mm * 100 multi-strand oxygen-free enameled wire, heat is very low, efficiency is more than 90%.Which in translation means - the transformer uses a winding of 100 pieces of oxygen-free wires with a diameter of 0.1 mm, heating is reduced and the efficiency is above 90%.
Well, I’ll check the efficiency later, but it’s a fact about the fact that the winding is multi-wire. Of course, I didn’t count them, but the tourniquet is pretty good and this option winding really has a positive effect on the quality of operation of the transformer in particular and the entire power supply unit in general.
They didn’t forget about the capacitor connecting the “hot” and “cold” sides of the power supply, and installed it of the correct (Y1) type. 
The output rectifier of the main channels uses diode assemblies MUR1620CTR and MUR1620CT (16 Amperes 200 Volts), and the manufacturer did not collectively farm “hybrid” options, but supplied, as expected, two complementary assemblies, one with a common cathode, and the other with a common anode. Both assemblies are mounted on separate heatsinks and, just like in the case of transistors, they are not isolated from the components. But in this case, the problem can only be in terms of electrical safety, although if the case is closed, then there is nothing wrong with that.
The output filter uses a pair of 1000 µF x 50 Volt capacitors, which in my opinion is not enough. 
In addition, to reduce ripple, a choke is installed between the capacitors, and the capacitors after it are additionally shunted with 100 nF ceramic.
In general, on the product page it was written -
1. All high-frequency low-impedance electrolytic capacitors specifications, low ripple.In translation, all capacitors have low impedance to reduce ripple. In general, this is how it is, Cheng-X is used, but this is essentially just a slightly improved version of ordinary Chinese capacitors and I would rather use my favorite Samwha RD or Capxon KF.
There are no discharge resistors parallel to the capacitors, although there is space on the board for them, so “surprises” may await you, since the charge lasts quite a long time. 
Additional power channels are connected to their own windings of the transformer, and the 12 Volt channel is galvanically isolated from the rest.
Each channel has independent voltage stabilization, chokes to reduce interference, and ceramic output capacitors. But you probably noticed that there are five diodes in the rectifier. The 12 Volt channel is powered by a half-wave rectifier. 
At the output, as well as at the input, there are terminal blocks, and they are of very good quality and design. 
On the product page there is a photo at the top where you can see everything at once. It was only later that I noticed that in all the photos in the store there were mounting stands; mine did not have them :( 
The printed circuit board is double-sided, the quality is very high, fiberglass is used, and not the usual getinax. A protective slot is made in one of the bottlenecks.
A pair of resistors were also found at the bottom, I assume that this is a primitive overload protection circuit, which is sometimes added to drivers on IR2153. But to be honest, I wouldn't count on it. 
Also at the bottom of the printed circuit board there are output markings and output voltage options for which these boards are manufactured. Two things intrigued me a little - two identical ± 70 Volt options and a custom option. 
Before moving on to the tests, I’ll tell you a little about my version of such a power supply.
About three and a half years ago I posted an regulated power supply unit, which used a power supply assembled in approximately the same way. 
When assembled it also looked pretty similar, sorry for the poor quality of the photo. 
If we remove from my version everything “unnecessary”, for example, a unit for adjusting fan speed depending on temperature, as well as a more powerful transistor driver and an additional power supply circuit from the inverter output, then we will get the circuit of the reviewed power supply.
In essence, this is the same power supply, only there are more output voltages. In general, the circuit design of this power supply is quite simple, only a banal self-oscillator is simpler. 
In addition, the reviewed power supply is equipped with a primitive output power limiting circuit; I suspect that it is implemented as shown in the selected section of the circuit. 
But let’s see what this circuit and its implementation in the reviewed power supply are capable of.
It should be noted here that since there is no stabilization of the main voltage, it directly depends on the voltage in the network.
With an input voltage of 223 Volts, the output is 35.2 in idle mode. The consumption is 3.3 watts. 
In this case, there is noticeable heating of the transistor driver power resistor. Its nominal value is 150 kOhm, which at 300 Volts gives a power dissipation of about 0.6 Watts. This resistor heats up regardless of the load on the power supply.
A slight heating of the transformer is also noticeable; the photo was taken approximately 15 minutes after switching on. 
For the load test, a structure was assembled consisting of two electronic loads, an oscilloscope and a multimeter.
The multimeter measured one power channel, the second channel was controlled by a voltmeter of the electronic load, which was connected with short wires. 
I won’t bore the reader with a large list of tests, so I’ll go straight to the oscillograms.
1, 2. Different output points of the power supply to the diode assemblies, and from at different times scans. The inverter operating frequency is 70 kHz.
3, 4. Ripple before and after the 12 Volt channel choke. After Krenka, everything is generally smooth, but there is a problem, the voltage at this point is only about 14.5 Volts without load on the main channels and 13.6-13.8 with load, which is not enough for a 12 Volt stabilizer. 
The load tests went like this:
First, I loaded one channel by 50%, then the second by 50%, then the load of the first was raised to 100%, and then the second. The result was four load modes - 25-50-75-100%.
First, the RF output, in my opinion, is very good, the ripple is minimal, and when installing an additional choke, it can be reduced to almost zero. 
But at a frequency of 100 Hz everything is quite sad, the input capacitance is too small, too small.
The total ripple swing at 500 watts of output power is about 4 volts. 
Load tests. Since the voltage sagged under load, I gradually increased the load current so that the output power roughly corresponded to the range 125-250-375-500 Watts.
1. First channel - 0 Watt, 42.4 Volts, second channel - 126 Watts, 33.75 Volts
2. The first channel - 125.6 Watts, 32.21 Volts, the second channel - 130 Watts, 32.32 Volts.
3. The first channel - 247.8 Watts, 29.86 Volts, the second channel - 127 Watts, 30.64 Volts.
4. The first channel is 236 Watts, 29.44 Volts, the second channel is 240 Watts, 29.58 Volts.
You probably noticed that in the first test the voltage of the unloaded channel is more than 40 Volts. This is due to voltage surges, and since there is no load at all, the voltage gradually rose, even a small load returned the voltage to normal. 
At the same time, consumption was measured, but since there is a relatively large error in measuring output power, I will also give the calculated efficiency values approximately.
1. 25% load, efficiency 89.3%
2. 50% load, efficiency 91.6%
3. 75% load, 90% efficiency
4. 476 Watt, about 95% load, efficiency 88%
5, 6. Just out of curiosity, I measured the power factor at 50 and 100% power.
In general, the results are approximately similar to the stated 90% 
Tests showed pretty good performance of the power supply and everything would have been great if not for the usual “fly in the ointment” in the form of heating. At the very beginning, I estimated the power of the power supply at approximately 300-350 Watts.
During the usual test with gradual warming up and intervals of 20 minutes, I found out that at a power of 250 watts the power supply behaves just fine, heating the components approximately as follows:
Diode bridge - 71
Transistors - 66
Transformer (magnetic core) - 72
Output diodes - 75
But when I raised the power to 75% (375 Watt), then after 10 minutes the picture was completely different
Diode bridge - 87
Transistors - 100
Transformer (magnetic core) - 78
Output diodes - 102
(more loaded channel)
After trying to figure out the problem, I found out that goes strong overheating of the transformer windings, as a result of this the magnetic circuit warms up, its saturation induction decreases and it begins to enter saturation, as a result, the heating of the transistors sharply increases (later I recorded the temperature up to 108 degrees), then I stopped the test. At the same time, “cold” tests with a power of 500 watts passed normally.
Below are a couple of thermal photos, the first at 25% load power, the second at 75%, respectively, after half an hour (20+10 minutes). The temperature of the windings reached 146 degrees and there was a noticeable smell of overheated varnish. 
In general, I will now summarize some results, some of which are disappointing.
The overall workmanship is very good, but there are some design nuances, such as installing transistors without insulation from the heatsinks. Pleased with the large number of output voltages, for example 35 Volts to power the power amplifier, 15 for the pre-amplifier and independent 12 Volts for all kinds of service devices.
There are circuit defects, for example, the absence of a thermistor at the input and the low capacitance of the input capacitors.
In the specifications it was stated that additional 15 Volt channels can produce a current of up to 1 Ampere, in reality I would not expect more than 0.5 Ampere without additional cooling of the stabilizers. The 12 Volt channel most likely will not produce more than 200-300 mA at all.
But all these problems are either not critical or can be easily solved. The most complex problem- heating The power supply can supply up to 250-300 Watts for a long time, 500 Watts only for a relatively short time, or you will have to add active cooling.
Along the way, I had a small question for the respected public. There are thoughts about making your own amplifier, according to the reviews. But which one would be more interesting, a power amplifier, a preliminary amplifier, if a PA, then at what power, etc. Personally, I don’t really need it, but I’m in the mood to dig deeper. The reviewed power supply has little to do with this :)
That's all for me, I hope that the information was useful and, as usual, I look forward to questions in the comments.
The product was provided for writing a review by the store. The review was published in accordance with clause 18 of the Site Rules.
I'm planning to buy +38 Add to favorites I liked the review +115 +179
For the manufacture of power supplies for power amplifiers, low-frequency 50-Hz transformers are usually used. They are reliable, do not create high-frequency interference and are relatively simple to manufacture. But there are also disadvantages - dimensions and weight. Sometimes such shortcomings turn out to be decisive and we have to look for other solutions. Partially, the issue of overall dimensions (more precisely, only height) is solved by using a toroidal transformer. But such a transformer costs a lot of money due to the complexity of manufacturing. And yet it still has significant weight. A solution to this problem can be the use of a switching power supply.
But it has its own characteristics: difficulty in manufacturing or alteration. To adapt a computer power supply to power the PA, you need to resolder half of the board and, most likely, rewind the secondary winding of the transformer. But modern Chinese industry produces large quantities 12-volt Tashibra power supplies and the like, promising decent output power, 50, 100, 150 W and above. At the same time, the cost of such power supplies is ridiculous.
In the picture there are a couple of such blocks, above BUKO, below Ultralight, but essentially the same Tashibra. They have slight differences (perhaps they were made in different provinces of China): the Tashibra secondary winding has 5 turns, while the BUKO has 8 turns. In addition, the Ultralight has a slightly larger board, with space for installing additional parts. Despite this, they are remade identically. During the modification process, you must be extremely careful, since the board contains high voltage, after the diode bridge it is 300 volts. In addition, if you accidentally short-circuit the output, the transistors will burn out.
Now about the scheme.
The circuit of power supplies from 50 to 150 watts is the same, the only difference is in the power of the parts used.
What needs to be improved?
1. You need to solder the electrolytic capacitor after the diode bridge. The capacitor capacity should be as large as possible. For this modification, a 100 µF capacitor was used for a voltage of 400 volts.
2. It is necessary to replace current feedback with voltage feedback. What is it for? In order for the power supply to start without load.
3. If necessary, rewind the transformer.
4. It will be necessary to rectify the output AC voltage with a diode bridge. For these purposes, you can use domestic KD213 diodes, or imported high-frequency ones. Better, of course, than Schottky. It is also necessary to smooth out the ripple at the output with a capacitor.
Here is a diagram of the converted power supply.
The blue circle marks the current feedback coil. To turn it off, you must unsolder one end so as not to create a short-circuited winding. After this, you can safely close the coil contact pads on the board. After this, it is necessary to organize voltage feedback. To do this, take a piece of twisted pair wire and wind 2 turns onto the power transformer. Then the same wire is wound 3 turns onto the communication transformer T1. After this, a 2.4 - 2.7 Ohm resistor with a power of 5 - 10 Watts is soldered to the ends of this wire. A 12-volt light bulb is connected to the output of the converter, and a 220-volt, 150-watt light bulb is connected to the power wire. The first bulb is used as a load, and the second as a current limiter. We turn on the converter to the network. If the power light does not light up, then everything is fine with the converter and you can remove the light. We plug it back into the network, this time without it. If the 12-volt light bulb on the load does not light up, it means that the direction of winding the coupling coil on the T1 coupling transformer was not correct and it will need to be wound in the other direction. Don’t forget to discharge the mains capacitor with a 1 kOhm resistor after turning off the power.
The power supply for ULF is usually bipolar; in this case, you need to get 2 voltages of 30 volts each. The secondary winding of the power transformer has 5 turns. With an output voltage of 12 volts, this results in 2.4 volts per turn. To get 30 volts, you need to wind 30 Volts/2.4 Volts = 12.5 turns. Therefore, it is necessary to wind 2 coils of 12.5 turns each. To do this, you need to unsolder the transformer from the board, temporarily wind up two turns of voltage feedback and wind up the secondary winding. After this, the calculated two secondary windings are wound with a simple stranded wire. First one coil is wound, then the other. The two ends of different windings are connected - this will be the zero output.
If it is necessary to obtain a different voltage, more/fewer turns are wound.
The operating frequency of the power supply with the voltage coupling coil is about 30 kHz.
Then a diode bridge is assembled, electrolytes and ceramic capacitors in parallel with them are soldered to dampen high-frequency interference. Here are more options for connecting the secondary windings.
This article is devoted to the 2161 Second Edition (SE) series of switching power supplies based on the IR2161 controller.
- Short circuit and overload protection;
- Auto reset short circuit protection;
- Frequency modulation "dither" (to reduce EMI);
- Microcurrent startup (for initial startup of the controller, a current of no more than 300 μA is sufficient);
- Possibility of dimming (but we are not interested in this);
- Output voltage compensation (a kind of voltage stabilization);
- Soft start;
- Adaptive dead time A.D.T.;
- Compact body;
- Produced using lead-free technology (Leed-Free).
I will give some important ones for us specifications:
Maximum inflow/outflow current: +/-500mA
A sufficiently large current allows you to control powerful switches and build quite powerful switching power supplies based on this controller without the use of additional drivers;
Maximum current consumed by the controller: 10mA
Based on this value, the power circuits of the microcircuit are designed;
Minimum operating voltage of the controller: 10.5V
At a lower supply voltage, the controller switches to UVLO mode and the oscillation stops;
Minimum stabilization voltage of the zener diode built into the controller: 14.5V
The external zener diode must have a stabilization voltage no higher than this value to avoid damage to the microcircuit due to shunting excess current to the COM pin;
Voltage at the CS pin to trigger overload protection: 0.5V
The minimum voltage at the CS pin at which the overload protection is triggered;
Voltage at the CS pin for short circuit protection: 1V
The minimum voltage at the CS pin at which short circuit protection is triggered;
Operating frequency range: 34 - 70 kHz
The operating frequency is not directly set and depends only on the power consumed by the load;
Default dead time: 1µS
Used when it is impossible to work in adaptive dead time (ADT) mode, as well as when there is no load;
Operating frequency in soft start mode: 130 kHz
The frequency at which the controller operates in soft start mode;
The main attention should now be paid to what operating modes of the microcircuit exist and in what sequence they are located relative to each other. I will focus on describing the operating principle of each of the circuit blocks, and I will describe the sequence of their operation and the conditions for transition from one mode to another more briefly. I'll start with a description of each of the blocks of the diagram:
Under-voltage Lock-Out Mode (UVLO)- the mode in which the controller is when its supply voltage is below the minimum threshold value (approximately 10.5V).
Soft Start Mode- operating mode in which the controller oscillator operates at an increased frequency for a short time. When the oscillator is turned on, its operating frequency is initially very high (about 130 kHz). This causes the converter output voltage to be lower because the power supply transformer has a fixed inductance which will have a higher impedance at higher frequency and thus reduces the voltage on the primary winding. Reduced voltage will naturally result in reduced current in the load. As the CSD capacitor charges from 0 to 5V, the oscillation frequency will gradually decrease from 130 kHz to the operating frequency. The duration of the soft start sweep will depend on the capacitance of the CSD capacitor. However, since the CSD capacitor also sets the shutdown delay time and participates in the operation of the voltage compensation unit, its capacitance must be strictly 100nF.
Soft start problem. I would like to be completely honest and mention the fact that if there are high-capacity filter capacitors at the output of the power supply, soft start most often does not work and the SMPS starts immediately at the operating frequency, bypassing the soft start mode. This happens due to the fact that at the moment of start, the discharged capacitors in the secondary circuit have a very low intrinsic resistance and a very high current is required to charge them. This current causes a short-circuit protection to operate briefly, after which the controller immediately restarts and goes into RUN mode, bypassing the soft start mode. You can combat this by increasing the inductance of the chokes in the secondary circuit, located immediately after the rectifier. Chokes with high inductance extend the charging process of the output filter capacitors; in other words, the capacitors are charged with a smaller current, but longer in time. Less charging current does not trigger protection at start and allows soft start to perform its functions normally. Just in case, regarding this issue, I contacted the manufacturer’s technical support, to which I received the following answer:
"A typical halogen converter has an AC output without rectifiers or output capacitors. Soft starting works by reducing the frequency. To achieve soft starting, the transformer needs to have significant leakage. However, this should be possible in your case. Try placing the inductor on the secondary side of the bridges diodes to the capacitor.
Best wishes.
Infineon Technologies
Steve Rhyme, Support Engineer"
My assumptions about the reason for the unstable operation of soft start turned out to be correct and, moreover, they even offered me the same method of dealing with this problem. And again, to be completely honest, it should be added that the use of coils with increased inductance, relative to those usually used at the output of the SMPS, improves the situation, but does not completely eliminate the problem. However, this problem can be tolerated given that there is a thermistor at the SMPS input that limits the inrush current.
Run Mode, operating mode. When the soft start is completed, the system enters voltage compensated operating mode. This function provides some stabilization of the converter output voltage. Voltage compensation occurs by changing the operating frequency of the converter (increasing the frequency reduces the output voltage), although the accuracy of this type of “stabilization” is not high, it is nonlinear and depends on many parameters and, therefore, is not easy to predict. IR2161 monitors the load current through a current resistor (RCS). The peak current is detected and amplified in the controller and then applied to the CSD pin. The voltage on the CSD capacitor, in operating mode (voltage compensation mode), will vary from 0 (at minimum load) to 5V (at maximum load). In this case, the generator frequency will vary from 34 kHz (Vcsd = 5V) to 70 kHz (Vcsd = 0V).
It is also possible to attach feedback to the IR2161, which will allow you to organize almost complete stabilization of the output voltage and will allow you to much more accurately monitor and maintain the required voltage at the output: 
We will not consider this scheme in detail within the framework of this article.
Shut Down Mode, shutdown mode. The IR2161 contains a two-position automatic shutdown system that detects both short circuit and overload conditions of the inverter. The voltage at the CS pin is used to determine these conditions. If the output of the converter is shorted, a very large current will flow through the switches and the system must shut down within a few periods of time on the grid, otherwise the transistors will be quickly destroyed due to thermal runaway of the junction. The CS pin has a turn-off delay to prevent nuisance tripping, either due to inrush current at turn-on or due to transient currents. Lower threshold (when Vcs > 0.5< 1 В), имеет намного long delay before switching off the power supply. The delay for overload shutdown is approximately 0.5 seconds. Both shutdown modes (overload and short circuit) have an automatic reset, which allows the controller to resume operation approximately 1 second after the overload or short circuit is eliminated. This means that if the fault is corrected, the inverter can continue to operate normally. The oscillator operates at its minimum operating frequency (34 kHz) when the CSD capacitor is switched to the shutdown circuit. In soft start mode or running mode, if the overload threshold is exceeded (Vcs > 0.5V), the IR2161 quickly charges the CSD to 5V. When the voltage at the CS pin is greater than 0.5V and when the short circuit threshold of 1V is exceeded, the CSD will charge from 5V to the controller supply voltage (10-15V) in 50ms. When the overload threshold voltage Vcs is more than 0.5V but less than 1V, the CSD is charged from 5V to the supply voltage in approximately 0.5 sec. It should be remembered and taken into account the fact that high-frequency pulses with a 50% duty cycle and a sinusoidal envelope appear at the CS pin - this means that only at the peak of the network voltage the CSD capacitor will be charged in stages, in each half-cycle. When the voltage on the CSD capacitor reaches the supply voltage, the CSD is discharged to 2.4V and the converter starts again. If the fault is still present, the CSD starts charging again. If the fault disappears, the CSD will discharge to 2.4V, and then the system will automatically return to the voltage compensation operating mode.
STANDBY mode, standby mode- the mode in which the controller is in the case of insufficient supply voltage, while it consumes no more than 300 μA. In this case, the oscillator is naturally turned off and the SMPS does not work; there is no voltage at its output.
Blocks Fault Timing Mode, Delay and Fault Mode, although shown in the block diagram, are not essentially operating modes of the controller; rather, they can be attributed to transition stages (Delay and Fault Mode) or conditions for transition from one mode to another (Fault Timing Mode).
Now I’ll describe how does it all work together:
When power is applied, the controller starts in UVLO mode. As soon as the controller supply voltage exceeds the minimum voltage value required for stable operation, the controller switches to soft start mode, the oscillator starts at a frequency of 130 kHz. The CSD capacitor charges smoothly up to 5V. As the external capacitors charge, the operating frequency of the oscillator decreases to the operating frequency. Thus, the controller switches to RUN mode. As soon as the controller enters RUN mode, the CSD capacitor is instantly discharged to ground potential and is connected by an internal switch to the voltage compensation circuit. If the SMPS is started not at idle, but under load, there will be a potential at the CS pin proportional to the load value, which, through the internal circuits of the controller, will affect the voltage compensation unit and will not allow the CSD capacitor, after the completion of the soft start, to completely discharge. Thanks to this, the start will not occur at the maximum frequency of the operating range, but at a frequency corresponding to the load value at the output of the SMPS. After switching to RUN mode, the controller works according to the situation: either it remains working in this mode until you get tired and unplug the power supply from the outlet, or... In case of overheating, the controller goes into FAULT mode, the oscillator stops working . After the chip cools down, a restart occurs. In the event of an overload or short circuit, the controller goes into Fault Timing mode, and the external capacitor CSD is instantly disconnected from the voltage compensation unit and connected to the shutdown unit (the CSD capacitor in this case sets the controller shutdown delay time). The operating frequency is instantly reduced to the minimum. In case of overload (when the voltage at the CS pin > 0.5< 1 В), контроллер переходит в режим SHUTDOWN и выключается, но происходит это не мгновенно, а только в том случае, если перегрузка продолжается дольше половины секунды. Если перегрузки носят импульсный характер с продолжительностью импульса не более 0,5 сек, то контроллер будет просто работать на минимально возможно частоте, постоянно переключаясь между режимами RUN, Fault Timing, Delay, RUN (при этом будут отчетливо слышны щелчки). Когда напряжение на выводе CS превышает 1В, срабатывает защита от короткого замыкания. При устранении перегрузки или короткого замыкания, контроллер переходит в режим STANDBY и при наличии благоприятных условий для перезапуска, минуя режим софт-старта, переходит в режим RUN.
Now that you understand how the IR2161 works (I hope so), I will tell you about the switching power supplies themselves based on it. I want to immediately warn you that if you decide to assemble a switching power supply based on this controller, then you should assemble the SMPS guided by the latest, most advanced circuit on the corresponding one printed circuit board. Therefore, the list of radioelements at the bottom of the article will be given only for latest version power supply. All intermediate editions of the IIP are shown only to demonstrate the process of improving the device.
And the first IIP that will be discussed is conventionally named by me 2161 SE 2.
The main and key difference of the 2161 SE 2 is the presence of a controller self-supply circuit, which made it possible to get rid of boiling quenching resistors and, accordingly, increase the efficiency by several percent. Other equally significant improvements were also made: optimization of the printed circuit board layout, more output terminals were added for connecting the load, and a varistor was added.
The SMPS diagram is shown in the image below:
The self-powering circuit is built on VD1, VD2, VD3 and C8. Due to the fact that the self-supply circuit is connected not to a low-frequency 220V network (with a frequency of 50Hz), but to the primary winding of a high-frequency transformer, the capacity of the self-supply quenching capacitor (C8) is only 330pF. If self-supply was organized from a low-frequency network of 50 Hz, then the capacity of the quenching capacitor would have to be increased 1000 times, of course, such a capacitor would take up much more space on the printed circuit board. The described method of self-powering is no less effective than self-powering from a separate winding of a transformer, but it is much simpler. Zener diode VD1 is necessary to facilitate the operation of the built-in zener diode of the controller, which is not capable of dissipating significant power and without installing an external zener diode can simply be broken, which will lead to a complete loss of functionality of the microcircuit. The stabilization voltage VD1 should be in the range of 12 - 14V and should not exceed the stabilization voltage of the controller's built-in zener diode, which is approximately 14.5V. As VD1, you can use a zener diode with a stabilization voltage of 13V (for example, 1N4743 or BZX55-C13), or use several zener diodes connected in series, which is what I did. I connected two zener diodes in series: one of them was 8.2V, the other was 5.1V, which ultimately gave a resulting voltage of 13.3V. With this approach to powering the IR2161, the controller’s supply voltage does not sags and is practically independent of the load size connected to the SMPS output. In this scheme, R1 is only needed to start the controller, so to speak, for the initial kick. R1 gets a little warm, but not nearly as much as it was in the first version of this power supply. Using a high-resistance resistor R1 gives another interesting feature: the voltage at the output of the SMPS does not appear immediately after being plugged into the network, but after 1-2 seconds, when the C3 is charged to the minimum voltage of 2161 (approximately 10.5V).
Starting with this SMPS and all subsequent ones, a varistor is used at the SMPS input; it is designed to protect the SMPS from exceeding the input voltage above the permissible value (in this case - 275V), and also very effectively suppresses high-voltage interference by preventing them from entering the SMPS input from network and without releasing interference from the SMPS back into the network.
In the rectifier of the secondary power supply of the power supply, I used SF54 diodes (200V, 5A) two in parallel. The diodes are located on two floors, the leads of the diodes should be as long as possible - this is necessary for better heat dissipation (the leads are a kind of radiator for the diode) and better air circulation around the diodes.
The transformer in my case is made on a core from a computer power supply - ER35/21/11. The primary winding has 46 turns in three 0.5mm wires, two secondary windings have 12 turns in three 0.5mm wires. The input and output chokes are also taken from the computer power supply.
The described power supply is capable of delivering 250W to the load for a long time (without operating time limitation), and 350W for a short time (no more than a minute). When using this SMPS in dynamic load mode (for example, to power an audio frequency power amplifier of class B or AB), it is possible to power an UMZCH with a total output power of 300W (2x150W in stereo mode) from this switching power supply.
Oscillogram on the primary winding of the transformer (without snubber, R5 = 0.15 Ohm, 190W output):
As can be seen from the oscillogram, with an output power of 190 W, the operating frequency of the SMPS is reduced to 38 kHz; at idle, the SMPS operates at a frequency of 78 kHz:
From the oscillograms, in addition, it is clearly visible that there are no outliers on the graph, and this undoubtedly characterizes this SMPS positively.
At the output of the power supply, in one of the arms you can see the following picture:
The ripple has a frequency of 100Hz and a ripple voltage of approximately 0.7V, which is comparable to the ripple at the output of a classic, linear, non-stabilized power supply. For comparison, here is an oscillogram taken when operating at the same output power for a classic power supply (capacitor capacity 15000 μF in the arm):
As can be seen from the oscillograms, the supply voltage ripple at the output of a switching power supply is lower than that of a classic power supply of the same power (0.7V for an SMPS, versus 1V for a classic unit). But unlike a classic power supply, a small high-frequency noise is noticeable at the output of the SMPS. However, there is no significant high-frequency interference or emissions. The ripple frequency of the supply voltage at the output is 100Hz and is caused by the voltage ripple in the primary circuit of the SMPS along the +310V bus. To further reduce ripple at the SMPS output, it is necessary to increase the capacitance of capacitor C9 in the primary circuit of the power supply or the capacitance of the capacitors in the secondary circuit of the power supply (the former is more effective), and to reduce high-frequency interference, use chokes with higher inductance at the SMPS output.
The PCB looks like this:
The following SMPS diagram that will be discussed is 2161 SE 3:
The finished power supply assembled according to this diagram looks like this:
In the scheme fundamental differences from SE 2 - no, the differences mainly concern the printed circuit board. The circuit added only snubbers in the secondary windings of the transformer - R7, C22 and R8, C23. The values of the gate resistors have been increased from 22 Ohm to 51 Ohm. The value of capacitor C4 has been reduced from 220 µF to 47 µF. Resistor R1 is assembled from four 0.5W resistors, which made it possible to reduce the heating of this resistor and make the design slightly cheaper because In my area, four half-watt resistors are cheaper than one two-watt one. But the opportunity to install one two-watt resistor remains. In addition, the value of the self-feeding capacitor was increased to 470pF, there was no particular point in this, but it was done as an experiment, the flight was normal. MUR1560 diodes in a TO-220 package are used as rectifier diodes in the secondary circuit. Optimized and reduced printed circuit board. The dimensions of the SE 2 printed circuit board are 153x88, while the SE 3 printed circuit board has dimensions of 134x88. The PCB looks like this:
The transformer is made on a core from a computer power supply - ER35/21/11. The primary winding has 45 turns in three 0.5mm wires, two secondary windings have 12 turns in four 0.5mm wires. The input and output chokes are also taken from the computer power supply.
The very first inclusion of this SMPS in the network showed that the snubbers in the secondary circuit of the power supply were clearly superfluous; they were immediately soldered off and were not used further. Later the snubber of the primary winding was also soldered off, as it turned out it did much more harm than good.
It was possible to extract 300-350W of power from this power supply for a long time; for a short time (no more than a minute) this SMPS can supply up to 500W; after a minute of operation in this mode, the overall radiator heats up to 60 degrees.
Look at the oscillograms:
Everything is still beautiful, the rectangle is almost perfectly rectangular, there are no outliers. With snubbers, oddly enough, everything was not so beautiful.
The following diagram is the final and most advanced 2161 SE 4:
When assembled, the device according to this diagram looks like this:
Like last time, there were no major changes in the scheme. Perhaps the most noticeable difference is that the snubbers have disappeared, both in the primary circuit and in the secondary ones. Because, as my experiments have shown, due to the peculiarities of the IR2161 controller, snubbers only interfere with its operation and are simply contraindicated. Other changes were also made. The values of the gate resistors (R3 and R4) have been reduced from 51 to 33 Ohms. In series with the self-feeding capacitor C7, a resistor R2 is added to protect against overcurrents when charging capacitors C3 and C4. Resistor R1 still consists of four half-watt resistors, and resistor R6 is now hidden under the board and consists of three SMD resistors of the 2512 format. Three resistors provide the required resistance, but it is not necessary to use exactly three resistors; depending on the required power, you can use one, two or three resistors are acceptable. Thermistor RT1 has been moved from the SMPS to the +310V target. The remaining measurements concern only the layout of the printed circuit board and it looks like this:
A safety gap has been added to the printed circuit board between the primary and secondary circuits, and a through cut has been made in the board at the narrowest point.
The transformer is exactly the same as in the previous power supply: it is made on a core from a computer power supply - ER35/21/11. The primary winding has 45 turns in three 0.5mm wires, two secondary windings have 12 turns in four 0.5mm wires. The input and output chokes are also taken from the computer power supply.
The output power of the power supply remained the same - 300-350W in long-term mode and 500W in short-term mode (no more than a minute). From this SMPS you can power a UMZCH with a total output power of up to 400W (2x200W in stereo mode).
Now let's look at the oscillograms on the primary winding of the transformer of this switching power supply:
Everything is still beautiful: the rectangle is rectangular, there are no outliers.
At the output of one of the arms of the power supply, at idle, you can observe the following picture:
As you can see, the output contains negligible high-frequency noise with a voltage of no more than 8 mV (0.008 V).
Under load, at the output, you can observe the already well-known ripples with a frequency of 100 Hz:
With an output power of 250W, the ripple voltage at the output of the SMPS is 1.2V, which, considering the lower capacitance of the capacitors in the secondary circuit (2000uF in the shoulder, versus 3200uF for SE2) and the high output power at which the measurements were made, looks very good. The high-frequency component at a given output power (250W) is also insignificant, has a more ordered character and does not exceed 0.2V, which is a good result.
Setting the protection threshold. The threshold at which the protection will operate is set by resistor RCS (R5 - in SE 2, R6 - in SE 3 and SE 4).
This resistor can be either output or SMD format 2512. RCS can be composed of several resistors connected in parallel.
The RCS denomination is calculated using the formula: Rcs = 32 / Pnom. Where, Pnom is the output power of the SMPS, above which the overload protection will operate.
Example: let's say that we need the overload protection to be triggered when the output power exceeds 275W. We calculate the resistor value: Rcs=32/275=0.116 Ohm. You can use either one 0.1 Ohm resistor, or two 0.22 Ohm resistors connected in parallel (which will result in 0.11 Ohm), or three 0.33 Ohm resistors, also connected in parallel (which will result in 0.11 Ohm) .
Now it’s time to touch on the topic that interests people the most - calculation of a transformer for a switching power supply. Due to your numerous requests, I will finally tell you in detail how to do this.
First of all, we need a core with a frame, or just a core if it is a ring-shaped core (shape R).
Cores and frames can be of completely different configurations and can be used in any way. I used an ER35 frame core from a computer power supply. The most important thing is that the core does not have a gap; cores with a gap cannot be used.
By default, immediately after starting the program, you will see similar numbers.
Starting the calculation, the first thing we will do is select the shape and dimensions of the core in the upper right corner of the program window. In my case, the shape is ER, and the sizes are 35/21/11.
The dimensions of the core can be measured independently; how to do this can be easily understood from the following illustration:
Next, select the core material. It’s good if you know what material your core is made of, if not, then it’s okay, just choose the default option - N87 Epcos. In our conditions, the choice of material will not have a significant impact on the final result.
The next step is to select the converter circuit; ours is half-bridge:
In the next part of the program - “supply voltage”, select “variable” and indicate 230V in all three windows.
In the “converter characteristics” part, we indicate the bipolar output voltage we need (voltage of one arm) and the required output power of the SMPS, as well as the diameter of the wire with which you want to wind the secondary and primary windings. In addition, the type of rectifier used is selected - “bipolar with a midpoint”. There we also check the box “use the desired diameters” and under “stabilization of outputs” select “no”. Select the type of cooling: active with a fan or passive without it. You should end up with something like this:
The actual values of the output voltages will be greater than what you indicate in the program when calculating. In this case, with a voltage of 2x45V specified in the program, the output of a real SMPS will be approximately 2x52V, so when calculating, I recommend specifying a voltage that is 3-5V less than required. Or indicate the required output voltage, but wind one turn less than indicated in the program calculation results. The output power should not exceed 350W (for 2161 SE 4). The diameter of the wire for winding, you can use any one you have, you need to measure and indicate its diameter. You should not wind the windings with a wire with a diameter of more than 0.8 mm; it is better to wind the windings using several (two, three or more) thin wires than one thick wire.
After all this, click on the “calculate” button and get the result, in my case it turned out like this:
We focus our attention on the points highlighted in red. The primary winding in my case will consist of 41 turns, wound in two wires with a diameter of 0.5 mm each. The secondary winding consists of two halves of 14 turns, wound in three wires with a diameter of 0.5 mm each.
After receiving all the necessary calculation data, we proceed directly to winding the transformer.
It seems to me that there is nothing complicated here. I'll tell you how I do it. First, the entire primary winding is wound. One of the ends of the wire(s) is stripped and soldered to the corresponding terminal of the transformer frame. After which the winding begins. The first layer is wound and then a thin layer of insulation is applied. After which the second layer is wound and a thin layer of insulation is applied again and thus the entire required number of turns of the primary winding is wound. It is best to wind the windings turn to turn, but you can also do it askew or just “anyhow”, this will not play a noticeable role. After the required number of turns have been wound, the end of the wire(s) is cut off, the end of the wire is stripped and soldered to another corresponding terminal of the transformer. After winding the primary winding, a thick layer of insulation is applied to it. It is best to use a special Mylar tape as insulation:
The same tape is used to insulate the windings of pulse transformers of computer power supplies. This tape conducts heat well and has high heat resistance. From available materials, it is recommended to use: FUM tape, masking tape, paper plaster or a baking sleeve cut into long strips. It is strictly forbidden to use PVC and fabric insulating tape, stationery tape, or fabric plaster to insulate windings.
After the primary winding is wound and insulated, we proceed to winding the secondary winding. Some people wind two halves of the winding at once and then separate them, but I wind the halves of the secondary winding one by one. The secondary winding is wound in the same way as the primary. First, we strip and solder one end of the wire(s) to the corresponding terminal of the transformer frame, wind the required number of turns, applying insulation after each layer. Having wound the required number of turns of one half of the secondary winding, we strip and solder the end of the wire to the corresponding terminal of the frame and apply a thin layer of insulation. We solder the beginning of the wire of the next half of the winding to the same terminal as the end of the previous half of the winding. We wind in the same direction, the same number of turns as the previous half of the winding, applying insulation after each layer. Having wound the required number of turns, solder the end of the wire to the corresponding terminal of the frame and apply a thin layer of insulation. There is no need to apply a thick layer of insulation after winding the secondary winding. At this point, the winding can be considered complete.
After winding is completed, it is necessary to insert the core into the frame and glue the core halves together. For gluing, I use one-second super glue. The adhesive layer should be minimal so as not to create a gap between the parts of the core. If you have a ring core (shape R), then naturally you won’t have to glue anything, but the winding process will be less convenient and will take away more strength and nerves. In addition, the ring core is less convenient due to the fact that you will have to create and mold the transformer leads yourself, as well as think about attaching the finished transformer to the printed circuit board.
Upon completion of winding and assembly of the transformer, you should get something like this:
For convenience of narration, I will also add here the SMPS 2161 SE 4 diagram for a brief description talk about the element base and possible replacements.
Let's go in order - from entrance to exit. By entry mains voltage meets fuse F1, the fuse can have a rating from 3.15A to 5A. Varistor RV1 must be designed for 275V, such a varistor will be marked 07K431, but it is also possible to use variators 10K431 or 14K431. It is also possible to use a varistor with a higher threshold voltage, but the effectiveness of protection and noise suppression will be noticeably lower. Capacitors C1 and C2 can be either regular film capacitors (such as CL-21 or CBB-21) or noise-suppressing type (for example X2) for a voltage of 275V. We unsolder the dual inductor L1 from a computer power supply or other faulty equipment. The inductor can be made independently by winding 20-30 turns on a small ring core, with a wire with a diameter of 0.5 - 0.8 mm. The VDS1 diode bridge can be any for a current from 6 to 8A, for example, indicated in the diagram - KBU08 (8A) or RS607 (6A). Any slow or fast diode with a current from 0.1 to 1A and a reverse voltage of at least 400V is suitable as VD4. R1 can consist of either four half-watt resistors of 82 kOhm, or be one two-watt resistor with the same resistance. Zener diode VD1 must have a stabilization voltage in the range of 13 - 14V; it is allowed to use either one zener diode or serial connection two zener diodes with lower voltage. C3 and C5 can be either film or ceramic. C4 should have a capacitance of no more than 47 µF, voltage 16-25V. Diodes VD2, VD3, VD5 must be very fast, for example - HER108 or SF18. C6 can be either film or ceramic. Capacitor C7 must be designed for a voltage of at least 1000V. C9 can be either film or ceramic. The R6 rating must be calculated for the required output power, as described above. As R6, you can use either SMD resistors of the 2512 format or output one- or two-watt resistors; in any case, the resistor(s) are installed under the board. Capacitor C8 must be film (type CL-21 or CBB-21) and have an allowable operating voltage of at least 400V. C10 is an electrolytic capacitor with a voltage of at least 400V; the magnitude of low-frequency ripples at the output of the SMPS depends on its capacitance. RT1 is a thermistor, you can buy it, or you can unsolder it from a computer power supply, its resistance should be from 10 to 20 Ohms and the permissible current should be at least 3A. Both the IRF740 indicated in the diagram and other transistors with similar parameters, for example, IRF840, 2SK3568, STP10NK60, STP8NK80, 8N60, 10N60, can be used as transistors VT1 and VT2. Capacitors C11 and C13 must be film (type CL-21 or CBB-21) with a permissible voltage of at least 400V, their capacitance must not exceed the 0.47 μF indicated in the diagram. C12 and C14 are ceramic, high-voltage capacitors for a voltage of at least 1000V. The VDS2 diode bridge consists of four diodes connected by a bridge. As VDS2 diodes, it is necessary to use very fast and powerful diodes, for example, such as - MUR1520 (15A, 200V), MUR1560 (15A, 600V), MUR820 (8A, 200V), MUR860 (8A, 600V), BYW29 (8A, 200V) , 8ETH06 (8A, 600V), 15ETH06 (15A, 600V). Chokes L2 and L3 are soldered from the computer power supply or made independently. They can be wound either on individual ferrite rods or on a common ring core. Each of the chokes should contain from 5 to 30 turns (more is better), with a wire with a diameter of 1 - 1.5 mm. Capacitors C15, C17, C18, C20 must be film (type CL-21 or CBB-21) with a permissible voltage of 63V or more, the capacitance can be any, the larger their capacitance, the better, the stronger the suppression of high-frequency interference. Each of the capacitors designated in the diagram as C16 and C19 consists of two 1000uF 50V electrolytic capacitors. In your case you may need to use higher voltage capacitors.
And as a final touch, I’ll show you a photo that shows the evolution of the switching power supplies I created. Each subsequent SMPS is smaller, more powerful and better quality than the previous one: 
That's all! Thank you for your attention!
List of radioelements
| Designation | Type | Denomination | Quantity | Note | Shop | My notepad | |
|---|---|---|---|---|---|---|---|
| Switching Power Supply 2161 SE 4 | |||||||
| R1 | Resistor | 82 kOhm | 4 | 0.5W | To notepad | ||
| R2 | Resistor | 4.7 Ohm | 1 | 0.25W | To notepad | ||
| R3, R4 | Resistor | 33 Ohm | 2 | 0.25W | To notepad | ||
| R5 | Resistor | ||||||
A switching power supply, providing bipolar voltage +/-50V with a power of up to 300 W, is intended for use, or high-power laboratory power supplies (). This one is relatively simple circuit The pulse power supply is assembled mainly from radioelements taken from old AT/ATX power supplies.
Schematic diagram of the converter 220/2x50V
Scheme of a homemade pulse power supply for UMZCH The inverter transformer was wound on an ETD39 ferrite core. The winding data is practically the same, only the output windings are slightly wound to accommodate the increase in voltage. The key transistors are powerful IRFP450. The driver is the popular TL494 chip. Power is supplied through a special stabilizer. In it, the starting resistor with the rectified mains voltage charges the power capacitor, on which, when the voltage reaches the threshold, the stabilizer turns on, starting the driver. It will be powered only when energy is accumulated on the capacitor, and after the converter starts, the additional winding of the transformer will take over the driver power. The operating principle of this launch option has been known for a long time and is used in the popular m/s UC384x.
Printed circuit board Power cascade
Another feature of the power supply circuit design is control field effect transistors. Here the lower IRFP450 circuit is controlled directly from the driver output, and the upper one is controlled using a small transformer.
In addition, the system was equipped with current protection, monitoring the current of the lower field worker using its resistance Rdson.
PSU test results
Finished power supply - board with parts In practice, it was possible to obtain about 100-150 output power from 4 ohm speakers. The voltage +/-50V is set by resistor P1 10k. Of course, it can take any value, depending on the application ULF circuits. The system currently operates as a .
