Switching power supply (60W) based on PWM UC3842. A simple switching power supply based on the UC3842 chip

Chip UC3842(UC3843)- is a PWM controller circuit with current and voltage feedback for controlling a key stage on an n-channel MOS transistor, ensuring the discharge of its input capacitance with a forced current of up to 0.7A. Chip SMPS the controller consists of a series of microcircuits UC384X (UC3843, UC3844, UC3845) PWM controllers. Core UC3842 specifically designed for long-term operation with a minimum number of external discrete components. PWM controller UC3842 It features precise duty cycle control, temperature compensation and is low cost. Feature UC3842 is the ability to operate within 100% duty cycle (for example UC3844 works with a fill factor of up to 50%.). Domestic analogue UC3842 is 1114EU7. Power supplies made on a microcircuit UC3842 are characterized by increased reliability and ease of execution.

Differences in supply voltage between UC3842 and UC3843:

UC3842_________| 16 Volt / 10 Volt
UC3843_________| 8.4 Volt / 7.6 Volt

Differences in pulse duty cycle:

UC3842, UC3843__| 0% / 98%

Tsokolevka UC3842(UC3843) shown in Fig. 1

The simplest connection diagram is shown in Fig. 2

PWM controller chips ka3842 or UC3842 (uc2842) is the most common when constructing power supplies for household and computer equipment; it is often used to control a key transistor in switching power supplies.

Operating principle of ka3842, UC3842, UC2842 microcircuits

The 3842 or 2842 chip is a PWM - pulse-width modulation (PWM) converter, mainly used to operate in DC-DC mode (converts a constant voltage of one value to a constant voltage of another) converter.

Let's consider the block diagram of microcircuits 3842 and 2842 series:
Pin 7 of the microcircuit is supplied with a supply voltage ranging from 16 Volts to 34. The microcircuit has a built-in Schmidt trigger (UVLO), which turns on the microcircuit if the supply voltage exceeds 16 Volts, and turns it off if the supply voltage for some reason falls below 10 Volts. The 3842 and 2842 series microcircuits also have overvoltage protection: if the supply voltage exceeds 34 Volts, the microcircuit will turn off. To stabilize the frequency of pulse generation, the microcircuit has its own 5-volt voltage stabilizer inside, the output of which is connected to pin 8 of the microcircuit. Pin 5 mass (ground). Pin 4 sets the pulse frequency. This is achieved by resistor R T and capacitor C T connected to 4 pins. - see typical connection diagram below.

Pin 6 – output of PWM pulses. 1 pin of the 3842 chip is used for feedback, if on 1 pin. lower the voltage below 1 Volt, then at the output (6 pins) of the microcircuit the pulse duration will decrease, thereby reducing the power of the PWM converter. Pin 2 of the microcircuit, like the first, serves to reduce the duration of the output pulses; if the voltage at pin 2 is higher than +2.5 Volts, then the pulse duration will decrease, which in turn will reduce the output power.

The microcircuit with the name UC3842, in addition to UNITRODE, is produced by ST and TEXAS INSTRUMENTS, analogues of this microcircuit are: DBL3842 by DAEWOO, SG3842 by MICROSEMI/LINFINITY, KIA3842 by KES, GL3842 by LG, as well as microcircuits from other companies with different letters (AS, MC, IP etc.) and digital index 3842.

Scheme of a switching power supply based on the UC3842 PWM controller

Schematic diagram of a 60-watt switching power supply based on a UC3842 PWM controller and a power switch based on a 3N80 field-effect transistor.

UC3842 PWM controller chip - full datasheet with the ability to download for free in pdf format or look in the online reference book on electronic components on the website

Circuits and printed circuit boards of power supplies based on UC3842 and UC3843 chips

Microcircuits for building switching power supplies of the UC384x series are comparable in popularity to the famous TL494. They are produced in eight-pin packages, and the printed circuit boards for such power supplies are very compact and single-sided. The circuitry for them has been debugged for a long time, all the features are known. Therefore, these microcircuits, along with TOPSwitch, can be recommended for use.

So, the first scheme is an 80W power supply. Source:

Actually, the diagram is practically from the datasheet.


click to enlarge
The printed circuit board is quite compact.

PCB file: uc3842_pcb.lay6

In this circuit, the author decided not to use the input of the error amplifier due to its high input impedance in order to avoid interference. Instead, the feedback signal is connected to a comparator. The Schottky diode on the 6th pin of the microcircuit prevents possible voltage surges of negative polarity, which may be due to the characteristics of the microcircuit itself. To reduce inductive emissions in the transformer, its primary winding is sectioned and consists of two halves separated by a secondary one. The closest attention should be paid to inter-winding insulation. When using a core with a gap in the center core, external interference should be minimal. A current shunt with a resistance of 0.5 Ohm with the 4N60 transistor indicated in the diagram limits the power to around 75W. The snubber uses SMD resistors, which are connected in parallel and in series, because They generate noticeable power in the form of heat. This snubber can be replaced with a diode and a 200-volt zener diode (suppressor), but they say that this will increase the amount of impulse noise from the power supply. A space for an LED has been added on the printed circuit board, which is not reflected in the diagram. You should also add a load resistor parallel to the output, because At idle, the power supply can behave unpredictably. Most of the output elements on the board are installed vertically. The power supply to the microcircuit is removed during the reverse stroke, so when converting the unit into an adjustable one, you should change the phasing of the microcircuit's power winding and recalculate the number of its turns, as for a forward one.

The following schematic and PCB are from this source:

The dimensions of the board are a little larger, but there is room for a slightly larger mains electrolyte.

The scheme is almost similar to the previous one:

click to enlarge
A trim resistor is installed on the board to adjust the output voltage. Likewise, the chip is powered from the power winding in reverse, which can lead to problems with a wide range of power supply output voltage adjustments. To avoid this, you should also change the phasing of this winding and power the microcircuit in forward motion.

PCB file: uc3843_pcb.dip

The UC384x series microcircuits are interchangeable, but before replacing you need to check how the frequency is calculated for a specific microcircuit (the formulas are different) and what the maximum duty cycle is - they differ by half.

To calculate the transformer windings, you can use the Flyback 8.1 program. The number of turns of the microcircuit power winding in forward motion can be determined by the ratio of turns to volts.

The article will provide a description, operating principle and connection diagram of the UC3842. This is a microcircuit that is a pulse width controller. Scope of application - in DC-DC converters. Using one microcircuit, you can create a high-quality voltage converter that can be used in power supplies for various equipment.

Pin assignment of the microcircuit (brief overview)

First you need to consider the purpose of all the pins of the microcircuit. The description of the UC3842 looks like this:

1. The voltage necessary for feedback is supplied to the first pin of the microcircuit. For example, if you lower the voltage on it to 1 V or lower, the pulse time at pin 6 will begin to decrease significantly.
2. The second output is also necessary to create feedback. However, unlike the first one, a voltage of more than 2.5 V must be applied to it in order to reduce the pulse duration. This also reduces power.
3. If a voltage of more than 1 V is applied to the third pin, then pulses will stop appearing at the output of the microcircuit.
4. A variable resistor is connected to the fourth pin - with its help you can set the pulse frequency. An electrolytic capacitor is connected between this terminal and ground.
5. The fifth conclusion is general.
6. PWM pulses are removed from the sixth pin.
7. The seventh pin is intended for connecting power in the range of 16..34 V. Built-in overvoltage protection. Please note that the microcircuit will not work at voltages below 16 V.
8. To stabilize the pulse frequency, a special device is used that supplies +5 V to the eighth pin.

Before considering practical designs, you need to carefully study the description, operating principle and connection diagrams of the UC3842.

How does the microcircuit work?

Now we need to briefly consider the operation of the element. When a DC voltage of +5 V appears on the eighth leg, the OSC generator starts. A positive pulse of short length is supplied to the trigger inputs RS and S. Then, after a pulse is given, the trigger switches and zero appears at the output. As soon as the OSC pulse begins to fall, the voltage at the direct inputs of the element will be zero. But a logical one will appear at the inverting output.

This logic unit allows the transistor to turn on, so that electric current will begin to flow from the power source through the collector-emitter circuit to the sixth pin of the microcircuit. This shows that there will be an open pulse at the output. And it will stop only when a voltage of 1 V or higher is applied to the third pin.

Why do you need to check the microcircuit?

Many radio amateurs who design and install electrical circuits purchase parts in bulk. And it’s no secret that the most popular shopping places are Chinese online stores. The cost of products there is several times lower than on radio markets. But there are also a lot of defective products there. Therefore, you need to know how to test the UC3842 before starting to build the circuit. This will avoid frequent unsoldering of the board.

Where is the chip used?

The chip is often used to assemble power supplies for modern monitors. They are used in line scan TVs and monitors. It is used to control transistors operating in switch mode. But elements fail quite often. And the most common reason is a breakdown of the field switch controlled by the microcircuit. Therefore, when independently designing a power supply or repairing, it is necessary to diagnose the element.

What you need to diagnose faults

It should be noted that the UC3842 was used exclusively in converter technology. And for normal operation of the power supply, you need to make sure that the element is working. You will need the following devices for diagnostics:

1. Ohmmeter and voltmeter (the simplest digital multimeter will do).
2. Oscilloscope.
3. Source of current and voltage stabilized power supply. It is recommended to use adjustable ones with a maximum output voltage of 20..30 V.

If you do not have any measuring equipment, then the easiest way to diagnose is to check the output resistance and simulate the operation of the microcircuit when operating from an external power source.

Checking the output resistance

One of the main diagnostic methods is to measure the resistance value at the output. We can say that this is the most accurate way to determine breakdowns. Please note that in the event of a breakdown of the power transistor, a high-voltage pulse will be applied to the output stage of the element. For this reason, the microcircuit fails. At the output, the resistance will be infinitely large if the element is working properly.

Resistance is measured between terminals 5 (ground) and 6 (output). The measuring device (ohmmeter) is connected without special requirements - polarity does not matter. It is recommended to unsolder the microcircuit before starting diagnostics. During breakdown, the resistance will be equal to several ohms. If you measure resistance without soldering the microcircuit, the gate-source circuit may ring. And do not forget that in the power supply circuit on the UC3842 there is a constant resistor, which is connected between ground and output. If it is present, the element will have an output resistance. Therefore, if the output resistance is very low or equal to 0, then the microcircuit is faulty.

How to simulate the operation of a microcircuit

When simulating operation, there is no need to solder the microcircuit. But be sure to turn off the device before starting work. Checking the circuit on the UC3842 consists of applying voltage to it from an external source and evaluating the operation. The work procedure looks like this:

1. The power supply is disconnected from the AC mains.
2. A voltage greater than 16 V is supplied from an external source to the seventh pin of the microcircuit. At this moment, the microcircuit should start. Please note that the chip will not start working until the voltage is above 16 V.
3. Using an oscilloscope or voltmeter, you need to measure the voltage at the eighth pin. It should be +5 V.
4. Make sure the voltage on pin 8 is stable. If you reduce the power supply voltage below 16 V, then the current will disappear at the eighth pin.
5. Using an oscilloscope, measure the voltage at the fourth pin. If the element is working properly, the graph will show sawtooth-shaped pulses.
6. Change the voltage of the power supply - the frequency and amplitude of the signal at the fourth pin will remain unchanged.
7. Check with an oscilloscope whether there are rectangular pulses on the sixth leg.

Only if all the signals described above are present and behave as they should, can we talk about the serviceability of the microcircuit. But it is recommended to check the serviceability of the output circuits - diode, resistors, zener diode. With the help of these elements, signals are generated for current protection. They fail when broken.

Switching power supplies on a chip

For clarity, you need to consider the description of the operation of the power supply on the UC3842. It first began to be used in household appliances in the second half of the 90s. It has a clear advantage over all competitors - low cost. Moreover, reliability and efficiency are not inferior. To build a complete one, practically no additional components are required. Everything is done by the “internal” elements of the microcircuit.

The element can be made in one of two types of housing - SOIC-14 or SOIC-8. But you can often find modifications made in DIP-8 packages. It should be noted that the last numbers (8 and 14) indicate the number of pins of the microcircuit. True, there are not very many differences - if the element has 14 pins, pins are simply added for connecting ground, power and the output stage. Stabilized pulse-type power supplies with PWM modulation are built on the microcircuit. A MOS transistor is required to amplify the signal.

Turning on the chip

Now we need to consider the description, operating principle and connection circuits of the UC3842. Power supplies usually do not indicate the parameters of the microcircuit, so you need to refer to special literature - datasheets. Very often you can find circuits that are designed to be powered from an alternating current network of 110-120 V. But with just a few modifications you can increase the supply voltage to 220 V.

To do this, the following changes are made to the power supply circuit on the UC3842:

1. The diode assembly, which is located at the input of the power source, is replaced. It is necessary that the new diode bridge operates at a reverse voltage of 400 V or more.
2. The electrolytic capacitor is replaced, which is located in the power circuit and serves as a filter. Installed after the diode bridge. It is necessary to install a similar one, but with an operating voltage of 400 V and higher.
3. The nominal value in the power supply circuit increases to 80 kOhm.
4. Check whether the power transistor can operate at a voltage between drain and source of 600 V. BUZ90 transistors can be used.

The article is shown on UC3842. has a number of features that must be taken into account when designing and repairing power supplies.

Features of the microcircuit

If there is a short circuit in the secondary winding circuit, then when diodes or capacitors break down, the loss of electricity in the pulse transformer begins to increase. It may also turn out that there is not enough voltage for the normal functioning of the microcircuit. During operation, a characteristic “clanking” sound is heard, which comes from the pulse transformer.

Considering the description, operating principle and connection diagram of the UC3842, it is difficult to ignore the repair features. It is quite possible that the reason for the behavior of the transformer is not a breakdown in its winding, but a malfunction of the capacitor. This happens as a result of the failure of one or more diodes that are included in the power circuit. But if a breakdown of the field-effect transistor occurs, it is necessary to completely change the microcircuit.

UC3845
PRINCIPLE OF OPERATION

Frankly speaking, it was not possible to defeat the UC3845 the first time - self-confidence played a cruel joke. However, wise with experience, I decided to finally figure it out - the chip is not that big - only 8 legs. I would like to express special gratitude to my subscribers, who did not stand aside and gave some explanations; they even sent a rather detailed article by email and a piece of the model in Microcap. THANK YOU VERY MUCH .
Using the links and materials sent, I sat for an evening or two and, in general, all the puzzles fit together, although some cells turned out to be empty. But first things first...
It was not possible to assemble an analogue of the UC3845 using logic elements in Microcap 8 and 9 - the logic elements are strictly connected to a five-volt power supply, and these simulators have chronic difficulties with self-oscillation. Microcap 11 showed the same results:

There was only one option left - Multisim. Version 12 was even found with a localization. I haven't used Multisim for a VERY long time, so I had to tinker. The first thing that pleased me was that Multisim has a separate library for five-volt logic and a separate library for fifteen-volt logic. In general, with grief in half, it turned out to be a more or less workable option, showing signs of life, but it didn’t want to work exactly the way a real microcircuit behaves, no matter how much I tried to persuade it. Firstly, the models do not measure the level relative to real zero, so an additional source of negative bias voltage would have to be introduced. But in this case they would have to explain in some detail what it is and why, but I wanted to be as close as possible to the real microcircuit.

Having rummaged through the Internet, I found a ready-made scheme, but for Multisim 13. I downloaded option 14, opened the model and it even worked, but the joy did not last long. Despite the presence in the libraries themselves of both the twelfth and fourteenth Multisim of the UC3845 microcircuit itself and its analogues, it quickly became clear that the model of the microcircuit does not allow working out ALL options for switching on this microcircuit. In particular, limiting the current and adjusting the output voltage work quite reliably (though it often falls out of the simulation), but the microcircuit refused to accept the use of applying a ground error to the output of the amplifier.

In general, although the cart moved, it did not travel far. There was only one option left - printing out the datasheet on the UC3845 and a board with wiring. In order not to get carried away with simulating the load and simulating current limiting, I decided to build a microbooster and use it to check what actually happens to the microcircuit under one or another variant of inclusion and use.
First, a little explanation:
The UC3845 microcircuit really deserves the attention of designers of power supplies of various powers and purposes; it has a number of almost analogues. Almost because when replacing a chip on a board, you don’t need to change anything else, but changes in ambient temperature can cause problems. And some sub-options cannot be used as a direct replacement at all.

Based on the table above, it is clear that the UC3845 is far from the best version of this microcircuit, since its lower temperature limit is limited to zero degrees. The reason is quite simple - not everyone stores a welding machine in a heated room, and a situation is possible when you need to weld something in the off-season, but the welder either does not turn on or simply explodes. no, not to shreds, even pieces of power transistors are unlikely to fly out, but there will be no welding in any case, and the welder also needs repairs. Having skimmed through Ali, I came to the conclusion that the problem is completely solvable. Of course, UC3845 is more popular and there are more of them on sale, but UC2845 is also on sale:

UC2845 is of course somewhat more expensive, but in any case it is cheaper than ONE power transistor, so I personally ordered a dozen UC2845 despite the fact that there are still 8 pieces of UC3845 in stock. Well, as you wish.
Now we can talk about the microcircuit itself, or more precisely about the principle of its operation. The figure below shows the block diagram of UC3845, i.e. with an internal trigger that does not allow the duration of the control pulse to be more than 50% of the period:

By the way, if you click on the picture, it will open in a new tab. It’s not entirely convenient to jump between tabs, but in any case it’s more convenient than turning the mouse wheel back and forth, returning to the picture that went to the top.
The chip provides dual control of the supply voltage. COMP1 monitors the supply voltage as such and if it is less than the set value, it issues a command that turns the internal five-volt regulator off. If the supply voltage exceeds the switching threshold, the internal stabilizer is unlocked and the microcircuit starts. The second element supervising the power supply is element DD1, which, in cases where the reference voltage differs from the norm, produces a logical zero at its output. This zero goes to inverter DD3 and, transformed into a logical one, goes to logical OR DD4. In almost all block diagrams, this one simply has an inverse input, but I took the inverter outside of this logical element - it’s easier to understand the principle of operation.
The OR logic element works on the principle of determining the presence of a logical one at any of its inputs. That is why it is called OR - if there is a logical one at input 1, OR at input 2, OR at input 3, OR at input 4, then the output of the element will be a logical one.
When a logical one appears at the first input of this adder of all control signals, a logical one will appear at its direct output, and a logical zero will appear at its inverse output. Accordingly, the upper driver transistor will be closed, and the lower one will open, thereby closing the power transistor.
The microcircuit will be in this state until the reference power analyzer gives permission to operate and a logical unit appears at its output, which, after the inverter DD3, unlocks the output element DD4.
Let's say our power supply is normal and the microcircuit starts working. The master oscillator begins to generate control pulses. The frequency of these pulses depends on the values ​​of the frequency-setting resistor and capacitor. There is a slight discrepancy here. The difference doesn’t seem to be big, but nevertheless it exists and there is a possibility of getting something that is not exactly what you wanted, namely a very hot device when a “faster” microcircuit from one manufacturer is replaced with a slower one. The most beautiful picture of the dependence of frequency on the resistance of the resistor and capacitance of the capacitor is from Texas Instruments:

Things are a little different for other manufacturers:

Dependence of frequency on RC ratings of a Fairchild microcircuit

Dependence of frequency on RC ratings of a microcircuit from STMicroelectronics

Dependence of frequency on RC ratings of a microcircuit from UNISONIC TECHNOLOGIES CO

The clock generator produces fairly short pulses in the form of a logical unit. These impulses are divided into three blocks:
1. The same final adder DD4
2. D-trigger DD2
3. RS trigger on DD5
The DD2 trigger is available only in microcircuits of the 44 and 45 subseries. It is this that prevents the duration of the control pulse from becoming longer than 50% of the period, since with each arriving edge of a logical unit from the clock generator it changes its state to the opposite. By doing this, it divides the frequency into two, forming zeros and ones of equal duration.
This happens in a rather primitive way - with each edge arriving at clock input C, the trigger writes to itself the information located at the information input D, and input D is connected to the inverse output of the microcircuit. Due to the internal delay, the inverted information is recorded. For example, the inverting output has a logical zero level. When the edge of the pulse arrives at input C, the trigger manages to record this zero before zero appears at its direct output. Well, if the direct output is zero, then the inverse output will be a logical one. With the arrival of the next edge of the clock pulse, the trigger already writes a logical unit into itself, which will appear at the output after some nanoseconds. Writing a logical one leads to the appearance of a logical zero at the inverse output of the trigger and the process will begin to repeat from the next edge of the clock pulse.

It is for this reason that the UC3844 and UC3845 microcircuits have an output frequency that is 2 times less than that of the UC3842 and UC3843 - it is shared by the trigger.
When the first pulse enters the unit setting input of the RS trigger DD5, it switches the trigger to a state where its direct output is logical one, and its inverse output is zero. And until one appears at input R, trigger DD5 will be in this state.
Suppose we do not have any control signals from the outside, then at the output of the error amplifier OP1 a voltage will appear close to the reference voltage - there is no feedback, the inverting input is in the air, and the non-inverting input is supplied with a reference voltage of 2.5 volts.
Here I’ll make a reservation right away - I personally was somewhat confused by this error amplifier, but after studying the datasheet more carefully and thanks to poking the noses of subscribers, it turned out that the output of this amplifier is not entirely traditional. In the output stage OP1 there is only one transistor connecting the output to the common wire. A positive voltage is generated by a current generator when this transistor is slightly open or completely closed.
From the output of OP1, the voltage passes through a kind of limiter and voltage divider 2R-R. In addition, this same bus has a voltage limit of 1 volt, so that under any conditions more than one volt does not reach the inverting input OP2.
OP2 is essentially a comparator that compares the voltages at its inputs, but the comparator is also tricky - a conventional operational amplifier cannot compare such low voltages - from actual zero to one volt. A conventional op-amp needs either a higher input voltage or a negative side of the supply voltage, i.e. bipolar voltage. The same comparator quite easily copes with the analysis of these voltages, it is possible that there are some biasing elements inside, but we don’t really care about the circuit diagram.
In general, OP2 compares the voltage coming from the output of the error amplifier, or more precisely, the remaining voltage that is obtained after passing through the divider with the voltage at the third pin of the microcircuit (DIP-8 package is meant).
But at this moment in time, we have nothing at all on the third pin, and a positive voltage is applied to the inverting input. Naturally, the comparator will invert it and form a clear logical zero at its output, which will not affect the state of the RS trigger DD5 in any way.
As a result of what is happening, we have a logical zero at the first input from the top, DD4, since our power supply is normal, at the second input we have short pulses from the clock generator, at the third input we have pulses from the D-flip-flop DD2, which have the same duration of zero and one . At and at the fourth input we have a logical zero from the RS trigger DD5. As a result, the output of the logic element will completely repeat the pulses generated by the D-trigger DD2. Therefore, as soon as a logical one appears at the direct output of DD4, transistor VT2 will open. At the same time, the inverse output will have a logical zero and transistor VT1 will be closed. As soon as a logical zero appears at the DD4 output, VT2 closes, and the inverse output of DD4 opens VT1, which will be the reason for opening the power transistor.
The current that VT1 and VT2 can withstand is one ampere, therefore this microcircuit can successfully control relatively powerful MOSFET transistors without additional drivers.
In order to understand exactly how the processes occurring in the power supply are regulated, the simplest booster was assembled, since it requires the least number of winding parts. The first GREEN ring that came to hand was taken and 30 turns were wound on it. The quantity was not calculated at all, just one layer of winding was wound and nothing more. I wasn’t worried about consumption - the microcircuit operates in a wide range of frequencies and if you start with frequencies under 100 kHz, then this will be quite enough to prevent the core from entering saturation.

The result was the following booster circuit:

All external elements have the prefix out, meaning that they are OUTSIDE microcircuit details.
I’ll immediately describe what’s on this diagram and why.
VT1 - the base is essentially in the air, the ends are soldered on the board for putting on jumpers, i.e. the base is connected either to ground or to a saw generated by the chip itself. There is no resistor Rout 9 on the board - I even missed its necessity.
Optocoupler Uout 1 uses the error amplifier OP1 to adjust the output voltage, the degree of influence is regulated by resistor Rout 2. Optocoupler Uout 2 controls the output voltage bypassing the error amplifier, the degree of influence is regulated by resistor Rout 4. Rout 14 is a current measuring resistor, specially taken at 2 Ohms so as not to remove the power transistor. Rout 13 - adjusting the current limit threshold. Well, Rout 8 - adjusting the clock frequency of the controller itself.

The power transistor is something that was soldered out of a car converter that was once being repaired - one arm flared up, I changed all the transistors (why ALL the answer is HERE), and this is, so to speak, a surrender. So I don’t know what it is - the inscription is very worn, in general it’s something like 40-50 amperes.
Rout 15 type load - 2 W at 150 Ohm, but 2 W turned out to be not enough. You need to either increase the resistance or increase the power of the resistor - it starts to stink if it works for 5-10 minutes.
VDout 1 - to exclude the influence of the main power on the operation of the controller (HER104 seems to have been a hit), VDout 2 - HER308, well, so that it doesn’t immediately go off if something goes wrong.
I realized the need for resistor R9 when the board was already soldered. In principle, this resistor will still need to be selected, but this is purely optional for those who REALLY want to get rid of the relay method of stabilization at idle. More on this a little later, but for now I stuck this resistor on the side of the tracks:

First start - engines ALL interlinear connectors must be connected to ground, i.e. they do not affect the circuit. The Rout 8 engine is installed so that the resistance of this resistor is 2-3 kOhm, since the capacitor is 2.2 nF, the frequency should be about 300-odd kHz, therefore at the output of the UC3845 we will get somewhere around 150 kHz.

We check the frequency at the output of the microcircuit itself - this is more accurate, since the signal is not cluttered by shock processes from the inductor. To confirm the differences between the generation frequency and the conversion frequency, we turn the yellow ray to pin 4 and see that the frequency is 2 times higher. The operating frequency itself turned out to be 146 kHz:

Now we increase the voltage on the optocoupler LED Uout 1 in order to control the change in stabilization modes. Here it should be recalled that the resistor Rout 13 slider is in the lower position in the diagram. A common wire is also supplied to the VT1 base, i.e. Absolutely nothing happens at pin 3 and comparator OP2 does not respond to the non-inverting input.
By gradually increasing the voltage on the optocoupler LED, it becomes obvious that control pulses simply begin to disappear. By changing the scan this becomes most clear. This happens because OP2 only monitors what is happening at its inverting input and as soon as the output voltage of OP1 drops below the threshold value, OP2 forms a logical one at its output, which sets trigger DD5 to zero. Naturally, but a logical one appears at the inverse output of the trigger, which blocks the final adder DD4. Thus the microcircuit stops completely.

But the booster is loaded, therefore the output voltage begins to decrease, the Uout 1 LED begins to decrease brightness, the Uout 1 transistor closes and OP1 begins to increase its output voltage and as soon as it passes the OP2 response threshold, the microcircuit starts again.
In this way, the output voltage is stabilized in relay mode, i.e. the microcircuit generates control pulses in batches.
By applying voltage to the LED of the optocoupler Uout 2, the transistor of this optocoupler opens slightly, entailing a decrease in the voltage supplied to the comparator OP2, i.e. the adjustment processes are repeated, but OP1 no longer takes part in them, i.e. the circuit is less sensitive to changes in output voltage. Thanks to this, the control pulse packets have a more stable duration and the picture seems more pleasant (even the oscilloscope is synchronized):

We remove the voltage from the Uout 2 LED and, just in case, check for the presence of a saw on the upper terminal of R15 (yellow beam):

The amplitude is slightly more than a volt and this amplitude may not be enough, because there are voltage dividers on the circuit. Just in case, we unscrew the slider of the tuning resistor R13 to the upper position and control what is happening at the third pin of the microcircuit. In principle, hopes were fully justified - the amplitude is not enough to start limiting the current (yellow ray):

Well, if there is not enough current through the inductor, it means either many turns or a high frequency. Rewinding is too lazy, because the board has a trimming resistor Rout8 to adjust the frequency. We rotate its regulator until the required voltage amplitude is obtained at pin 3 of the controller.
In theory, as soon as the threshold is reached, that is, as soon as the voltage amplitude at pin 3 becomes not much more than one volt, the duration of the control pulse will begin to be limited, since the controller is already beginning to think that the current is too high and it will turn off the power transistor.
Actually, this begins to happen at a frequency of about 47 kHz, and further decreases in frequency had virtually no effect on the duration of the control pulse.

A distinctive feature of the UC3845 is that it controls the flow through the power transistor at almost every cycle of operation, and not the average value, as for example the TL494 does, and if the power supply is designed correctly, then it will never be possible to damage the power transistor...
Now we raise the frequency until the current limitation ceases to have an effect, however, we will make a reserve - we set it to exactly 100 kHz. The blue ray still shows control pulses, but we put the yellow one on the LED of the optocoupler Uout 1 and begin to rotate the trimmer resistor knob. For some time, the oscillogram looks the same as during the first experiment, however, a difference also appears; after passing the control threshold, the duration of the pulses begins to decrease, i.e., real regulation occurs through pulse-width modulation. And this is just one of the tricks of this microcircuit - as a reference saw for comparison, it uses a saw that is formed on the current-limiting resistor R14 and thus creates a stabilized voltage at the output:

The same thing happens when the voltage on the optocoupler Uout 2 increases, although in my version it was not possible to get the same short pulses as the first time - the brightness of the optocoupler LED was not enough, and I was too lazy to reduce the resistor Rout 3.
In any case, PWM stabilization occurs and is quite stable, but only in the presence of a load, i.e. the appearance of a saw, even of no great significance, at pin 3 of the controller. Without this saw, stabilization will be carried out in relay mode.
Now we switch the base of the transistor to pin 4, thereby forcibly feeding the saw to pin 3. There is not a big stumble here - for this feint you will have to select a Rout 9 resistor, since the amplitude of the dust and the level of the constant component turned out to be somewhat too large for me.

However, now the principle of operation itself is more interesting, so we check it by lowering the Rout 13 trimmer engine to the ground and begin to rotate Rout 1.
There are changes in the duration of the control pulse, but they are not as significant as we would like - the large constant component has a strong effect. If you want to use this inclusion option, you need to think more carefully about how to organize it correctly. Well, the picture on the oscilloscope is as follows:

With a further increase in voltage on the optocoupler LED, a breakdown occurs in the relay mode of operation.
Now you can check the load capacity of the booster. To do this, we introduce a limitation on the output voltage, i.e. Apply a small voltage to the Uout 1 LED and reduce the operating frequency. The sociogram clearly shows that the yellow ray does not reach the level of one volt, i.e. There is no current limit. The limitation is provided only by adjusting the output voltage.
In parallel with the load resistor Rour 15, we install another 100 Ohm resistor and the oscillogram clearly shows an increase in the duration of the control pulse, which leads to an increase in the time of energy accumulation in the inductor and its subsequent release to the load:

It is also not difficult to notice that by increasing the load, the voltage amplitude at pin 3 also increases, since the current flowing through the power transistor increases.
It remains to see what happens at the drain in stabilization mode and in its complete absence. We turn a blue beam onto the drain of the transistor and remove the feedback voltage from the LED. The oscillogram is very unstable, since the oscilloscope cannot determine which edge it should synchronize with - after the pulse there is quite a decent “chatter” of self-induction. The result is the following picture.

The voltage on the load resistor also changes, but I won’t make a GIF - the page is already quite “heavy” in terms of traffic, so I declare with full responsibility that the voltage on the load is equal to the voltage of the maximum value in the picture above minus 0.5 volts.

LET'S SUM IT UP

UC3845 is a universal self-clocking driver for single-ended voltage converters, can work in both flyback and forward converters.
Can operate in relay mode, can operate in full-fledged PWM voltage stabilizer mode with current limitation. It is precisely a limitation, since during an overload the microcircuit goes into current stabilization mode, the value of which is determined by the circuit designer. Just in case, a small sign showing the dependence of the maximum current on the value of the current-limiting resistor:

For full PWM voltage regulation, the IC requires a load because it uses a ramp voltage to compare with the controlled voltage.
Voltage stabilization can be organized in three ways, but one of them requires an additional transistor and several resistors, and this conflicts with the formula LESS PARTS - MORE RELIABILITY, so two methods can be considered basic:
Using an integrated error amplifier. In this case, the feedback optocoupler transistor is connected by the collector to a reference voltage of 5 volts (pin 8), and the emitter supplies voltage to the inverting input of this amplifier through the OS resistor. This method is recommended for more experienced designers, since if the gain of the error amplifier is high, it may become excited.
Without using an integrated error amplifier. In this case, the collector of the regulating optocoupler is connected directly to the output of the error amplifier (pin 1), and the emitter is connected to the common wire. The input of the error amplifier is also connected to the common wire.
The operating principle of PWM is based on monitoring the average output voltage and maximum current. In other words, if our load decreases, the output voltage increases, and the saw amplitude at the current-measuring resistor drops and the pulse duration decreases until the lost balance between voltage and current is restored. As the load increases, the controlled voltage decreases and the current increases, which leads to an increase in the duration of the control pulses.

It is quite easy to organize a current stabilizer on a microcircuit, and the control of the flowing current is controlled at each cycle, which completely eliminates overloading of the power stage with the correct choice of the power transistor and the current-limiting, or more precisely, measuring resistor installed at the source of the field-effect transistor. It is this fact that has made the UC3845 the most popular when designing household welding machines.
UC3845 has quite serious “rake” - the manufacturer does not recommend using the microcircuit at temperatures below zero, so in the manufacture of welding machines it would be more logical to use UC2845 or UC1845, but the latter are in some shortage. UC2845 is slightly more expensive than UC3845, not as catastrophically as domestic sellers indicated (prices in rubles as of March 1, 2017).

The frequency of the XX44 and XX45 microcircuits is 2 times less than the clock frequency, and the coefficient of filling cannot exceed 50%, then it is most favorable for converters with a transformer. But the XX42 and XX43 microcircuits are best suited for PWM stabilizers, since the duration of the control pulse can reach 100%.

Now, having understood the operating principle of this PWM controller, we can return to designing a welding machine based on it...