Switching voltage converters. Boost DC-DC converter

I came across a very interesting step-down voltage converter in the open spaces of Ali, with such a set of characteristics.

Here's what the seller stated:
1.Input voltage range:5-36VDC
2.Output voltage range:1.25-32VDC adjustable
3.Output current: 0-5A
4.Output power: 75W
5.High efficiency up to 96%
6.Built in thermal shutdown function
7.Built in current limit function
8.Built in output short protection function
9.L x W x H =68.2x38.8x15mm

The seller either did not mention the most interesting features of this converter or did not draw attention to them. And the features are very interesting.

1. Built-in input and output voltage voltmeter, ammeter and wattmeter, with reading calibration function. The calibration function for voltage and current operates independently. The actual accuracy of the readings after calibration is around ~0.05v. But more on that below.

2. This step-down converter can operate in both voltage stabilization mode and current stabilization mode. In fact, this is the smallest and cheapest laboratory power supply with a built-in multimeter. To which you just need to attach a battery crib to get a ready-made charger for any type of battery.

The idea was to use this converter as a powerful converter capable of utilizing the full power of a solar battery with a voltage of 6v. Since the solar battery is planned to be used far from civilization, where there is no extra multimeter with you, I really wanted to find a converter with a built-in voltmeter-ammeter.

Step-down converters with a current stabilization function that are not afraid of short circuits, with a built-in voltmeter-ammeter, are not at all a big offer. Closest competitors:

In general, we couldn’t find anything better, and this converter was purchased. A month later the package was waiting at the post office.

The first tests of this converter were disappointing. It turned out that although the converter itself starts to work at input voltages above 3.2v, there was a problem with the voltmeter. The voltmeter was lying by SEVERAL VOLT!!! Therefore, the first thing to do was calibration. But it turned out that calibration does not help. If you calibrate the voltmeter at 5v, then problems began with readings at 12v and vice versa.

Later, experiments showed that the voltmeter shows correct values ​​only if the input voltage is above 6.5v. When the input voltage dropped below 6.5v, the voltmeter began to lie. Moreover, absolutely all readings were distorted at low input voltage. Even the output voltage readings began to “float”, although in fact they were stable. It was extremely unpleasant to observe when, when the input voltage decreased from 6.5v to 4.2v, the built-in voltmeter began to show that the input voltage was increasing. Here is an example of numbers, input voltage and voltage on the built-in voltmeter.

6.74v – 6.6v
6.25v – 6.7v
5.95v – 6.7v
5.55v – 6.8v
5.07v – 7.2v
4.61v – 7.5v
4.33v – 7.8v

When the input voltage dropped below 4.2v, the voltmeter turned off completely.

A dispute was created, but the seller turned out to be normal and did not resist; he immediately returned 50% of the price.

If you forget about the voltmeter, or assume that the supply voltage will always be greater than 7v, then we can assume that the converter is working perfectly. But for my case, when the main operating voltage range was 4v-8v, this could be considered a complete fiasco.

But then autumn came, long gloomy evenings, and it became interesting to see if something could be done.

Photo of the main elements of the converter

It turned out that a number of important elements were hidden under the display, which I didn’t want to unsolder unless absolutely necessary. Therefore, it was not possible to draw a complete circuit of the converter. Moreover, despite its apparent simplicity, the scheme is not so simple. Having poked the working converter with a multimeter, it became clear that all the problems begin when a separate power bus, with a stabilized voltage of 5v for the voltmeter and other “brains,” begins to sag. The LM317 chip is responsible for stable 5v. And as soon as the voltage at its input begins to be insufficient to produce stable 5v, problems begin for the voltmeter.

The problem became clear, but its solution did not seem so simple. In theory, you need to replace LM317 with some kind of analogue that can not only lower the voltage, but also increase it. Analogue of SEPIC converter or similar. There are such chips, but they will definitely not be pin-compatible, they will definitely require additional wiring, and the prices for such chips are usually not reasonable. And then an idea came. What if you add a boost converter board in front of LM317. Moreover, the current consumed by the “brains” is very small. The MT3608 converter, reviews of which are either available, was ideal for such a board. Another undeniable advantage of the MT3608 is its price. Now on Ali the price of MT3608 starts at $0.35 and tends to become even cheaper.

In addition to the price, the good news is that for modification you need to make a minimum of changes on the board. It is enough to cut one track (1) and solder three wires to the MT3608 +Vin (2), -Vin (3) and +Vout (4).


Additionally, several layers of electrical tape were wound over the MT3608 inductor to align the height with the trimmer resistor. Plus, a jumper was added to the MT3608 board itself to expand the range of adjustments with the potentiometer, and a 10 uF ceramic capacitor was added to the output. The result looked like this:

The result exceeded all expectations:

1. The accuracy of voltmeter-ammeter readings has increased significantly at input voltages below 6.5v. Simply put, the voltmeter began to work as it should immediately. Taking into account the calibration, you can set the readings in the desired range around 0.05v. Although it should still be noted that if you accurately set the region to 5v, in the region of 12v the voltmeter will lie in the region of 0.3v.

2. The voltmeter now turns on at 1.9v. Now you can see on the built-in voltmeter the moment the power part of the converter is turned on when the input voltage increases above 3.2v.

3. Now, in the event of a source overload, this is when the converter tries to take more from the power source than it can give, the converter has become much more stable. When overloaded, the power section drops the input voltage to somewhere around 3.45v, which is quite enough to power the “brains” of the converter. The converter does not enter a kind of flickering mode when the voltage is not enough to start the “brains”.

This modification also has a couple of disadvantages:

1. The board has become higher, so in order not to damage the “sandwich”, screws were screwed in, allowing the board to be installed on a flat surface without risk.

2. The operating range of input voltages has decreased. Previously, the input voltage could reach 35v. Now the upper limit has been reduced to 20v due to the MT3608 input voltage limitation. But in my case this is absolutely not critical.

LM2596 reduces the input voltage (to 40 V) - the output is regulated, the current is 3 A. Ideal for LEDs in a car. Very cheap modules - about 40 rubles in China.

Texas Instruments produces high-quality, reliable, affordable and cheap, easy-to-use DC-DC controllers LM2596. Chinese factories produce ultra-cheap pulsed stepdown converters based on it: the price of a module for LM2596 is approximately 35 rubles (including delivery). I advise you to buy a batch of 10 pieces at once - there will always be a use for them, and the price will drop to 32 rubles, and less than 30 rubles when ordering 50 pieces. Read more about calculating the circuitry of the microcircuit, adjusting the current and voltage, its application and some of the disadvantages of the converter.

The typical method of use is a stabilized voltage source. It is easy to make a switching power supply based on this stabilizer; I use it as a simple and reliable laboratory power supply that can withstand short circuits. They are attractive due to the consistency of quality (they all seem to be made at the same factory - and it’s difficult to make mistakes in five parts), and full compliance with the datasheet and declared characteristics.

Another application is a pulse current stabilizer for power supply for high-power LEDs. The module on this chip will allow you to connect a 10-watt automotive LED matrix, additionally providing short-circuit protection.

I highly recommend buying a dozen of them - they will definitely come in handy. They are unique in their own way - input voltage is up to 40 volts, and only 5 external components are required. This is convenient - you can increase the voltage on the smart home power bus to 36 volts by reducing the cross-section of the cables. We install such a module at the points of consumption and configure it to the required 12, 9, 5 volts or as needed.

Let's take a closer look at them.

Chip characteristics:

• Input voltage - from 2.4 to 40 volts (up to 60 volts in the HV version)
• Output voltage - fixed or adjustable (from 1.2 to 37 volts)
• Output current - up to 3 amperes (with good cooling - up to 4.5A)
• Conversion frequency - 150 kHz
• Housing - TO220-5 (through-hole mounting) or D2PAK-5 (surface mounting)
• Efficiency - 70-75% at low voltages, up to 95% at high voltages

1. Stabilized voltage source
2. Converter circuit
3. Datasheet
4. USB charger based on LM2596
5. Current stabilizer
6. Use in homemade devices
7. Adjustment of output current and voltage
8. Improved analogues of LM2596

History - linear stabilizers

To begin with, I’ll explain why standard linear voltage converters like LM78XX (for example 7805) or LM317 are bad. Here is its simplified diagram.

The main element of such a converter is a powerful bipolar transistor, switched on in its “original” meaning - as a controlled resistor. This transistor is part of a Darlington pair (to increase the current transfer coefficient and reduce the power required to operate the circuit). The base current is set by the operational amplifier, which amplifies the difference between the output voltage and the one set by the ION (reference voltage source), i.e. it is connected according to the classical error amplifier circuit.

Thus, the converter simply turns on the resistor in series with the load, and controls its resistance so that, for example, exactly 5 volts are extinguished across the load. It is easy to calculate that when the voltage decreases from 12 volts to 5 (a very common case of using the 7805 chip), the input 12 volts are distributed between the stabilizer and the load in the ratio “7 volts on the stabilizer + 5 volts on the load.” At a current of half an ampere, 2.5 watts are released at the load, and at 7805 - as much as 3.5 watts.

It turns out that the “extra” 7 volts are simply extinguished on the stabilizer, turning into heat. Firstly, this causes problems with cooling, and secondly, it takes a lot of energy from the power source. When powered from an outlet, this is not very scary (although it still causes harm to the environment), but when powered by a battery or rechargeable battery, this cannot be ignored.

Another problem is that it is generally impossible to make a boost converter using this method. Often such a need arises, and attempts to solve this issue twenty or thirty years ago are amazing - how complex the synthesis and calculation of such circuits was. One of the simplest circuits of this kind is a push-pull 5V->15V converter.

It must be admitted that it provides galvanic isolation, but it does not use the transformer efficiently - only half of the primary winding is used at any time.

Let's forget this like a bad dream and move on to modern circuitry.

Voltage source

Scheme

The microcircuit is convenient to use as a step–down converter: a powerful bipolar switch is located inside, all that remains is to add the remaining components of the regulator - a fast diode, an inductance and an output capacitor, it is also possible to install an input capacitor - only 5 parts.

The LM2596ADJ version will also require an output voltage setting circuit, these are two resistors or one variable resistor.

Step-down voltage converter circuit based on LM2596:

The whole scheme together:

Here you can download datasheet for LM2596.

Operating principle: a powerful switch inside the device, controlled by a PWM signal, sends voltage pulses to the inductance. At point A, x% of the time there is full voltage, and (1-x)% of the time the voltage is zero. The LC filter smooths out these oscillations by highlighting a constant component equal to x * supply voltage. The diode completes the circuit when the transistor is turned off.

Detailed job description

Inductance resists the change in current through it. When voltage appears at point A, the inductor creates a large negative self-induction voltage, and the voltage across the load becomes equal to the difference between the supply voltage and the self-induction voltage. The inductance current and voltage across the load gradually increase.

After the voltage disappears at point A, the inductor strives to maintain the previous current flowing from the load and the capacitor, and shorts it through the diode to ground - it gradually drops. Thus, the load voltage is always less than the input voltage and depends on the duty cycle of the pulses.

Output voltage

The module is available in four versions: with a voltage of 3.3V (index –3.3), 5V (index –5.0), 12V (index –12) and an adjustable version LM2596ADJ. It makes sense to use the customized version everywhere, since it is available in large quantities in the warehouses of electronic companies and you are unlikely to encounter a shortage of it - and it only requires an additional two penny resistors. And of course, the 5 volt version is also popular.

The quantity in stock is in the last column.

You can set the output voltage in the form of a DIP switch, a good example of this is given here, or in the form of a rotary switch. In both cases, you will need a battery of precision resistors - but you can adjust the voltage without a voltmeter.

Frame

There are two housing options: the TO-263 planar mount housing (model LM2596S) and the TO-220 through-hole housing (model LM2596T). I prefer to use the planar version of the LM2596S, since in this case the heatsink is the board itself, and there is no need to buy an additional external heatsink. In addition, its mechanical resistance is much higher, unlike the TO-220, which must be screwed to something, even to a board - but then it is easier to install the planar version. I recommend using the LM2596T-ADJ chip in power supplies because it is easier to remove a large amount of heat from its case.

Input voltage ripple smoothing

Can be used as an effective “smart” stabilizer after current rectification. Since the microcircuit directly monitors the output voltage, fluctuations in the input voltage will cause an inversely proportional change in the conversion coefficient of the microcircuit, and the output voltage will remain normal.

It follows from this that when using the LM2596 as a step-down converter after a transformer and rectifier, the input capacitor (i.e. the one located immediately after the diode bridge) may have a small capacitance (about 50-100 μF).

Output capacitor

Due to the high conversion frequency, the output capacitor also does not have to have a large capacity. Even a powerful consumer will not have time to significantly reduce this capacitor in one cycle. Let's do the calculation: take a 100 µF capacitor, 5 V output voltage and a load consuming 3 amperes. Full charge of the capacitor q = C*U = 100e-6 µF * 5 V = 500e-6 µC.

In one conversion cycle, the load will take dq = I*t = 3 A * 6.7 μs = 20 μC from the capacitor (this is only 4% of the total charge of the capacitor), and immediately a new cycle will begin, and the converter will put a new portion of energy into the capacitor.

The most important thing is not to use tantalum capacitors as the input and output capacitors. They write right in the datasheets - “do not use in power circuits”, because they very poorly tolerate even short-term overvoltages, and do not like high pulse currents. Use regular aluminum electrolytic capacitors.

Efficiency, efficiency and heat loss

The efficiency is not so high, since a bipolar transistor is used as a powerful switch - and it has a non-zero voltage drop, about 1.2V. Hence the drop in efficiency at low voltages.

As you can see, maximum efficiency is achieved when the difference between the input and output voltages is about 12 volts. That is, if you need to reduce the voltage by 12 volts, a minimal amount of energy will go into heat.

What is converter efficiency? This is a value that characterizes current losses - due to heat generation on a fully open powerful switch according to the Joule-Lenz law and to similar losses during transient processes - when the switch is, say, only half open. The effects of both mechanisms can be comparable in magnitude, so one should not forget about both loss paths. A small amount of power is also used to power the “brains” of the converter themselves.

Ideally, when converting voltage from U1 to U2 and output current I2, the output power is equal to P2 = U2*I2, the input power is equal to it (ideal case). This means that the input current will be I1 = U2/U1*I2.

In our case, the conversion has an efficiency below unity, so part of the energy will remain inside the device. For example, with efficiency η, the output power will be P_out = η*P_in, and losses P_loss = P_in-P_out = P_in*(1-η) = P_out*(1-η)/η. Of course, the converter will have to increase the input current to maintain the specified output current and voltage.

We can assume that when converting 12V -> 5V and an output current of 1A, the losses in the microcircuit will be 1.3 watts, and the input current will be 0.52A. In any case, this is better than any linear converter, which will give at least 7 watts of losses, and will consume 1 ampere from the input network (including for this useless thing) - twice as much.

By the way, the LM2577 microcircuit has a three times lower operating frequency, and its efficiency is slightly higher, since there are fewer losses in transient processes. However, it needs three times higher ratings of the inductor and output capacitor, which means extra money and board size.

Increasing output current

Despite the already fairly large output current of the microcircuit, sometimes even more current is required. How to get out of this situation?

1. Several converters can be parallelized. Of course, they must be set to exactly the same output voltage. In this case, you cannot get by with simple SMD resistors in the Feedback voltage setting circuit; you need to use either resistors with an accuracy of 1%, or manually set the voltage with a variable resistor.

USB charger for LM2596

You can make a very convenient travel USB charger. To do this, you need to set the regulator to a voltage of 5V, provide it with a USB port and provide power to the charger. I use a radio model lithium polymer battery purchased in China that provides 5 amp hours at 11.1 volts. This is a lot - enough to 8 times charge a regular smartphone (not taking into account efficiency). Taking into account the efficiency, it will be at least 6 times.

Don't forget to short the D+ and D- pins of the USB socket to tell the phone that it is connected to the charger and the current transferred is unlimited. Without this event, the phone will think that it is connected to the computer and will be charged with a current of 500 mA - for a very long time. Moreover, such a current may not even compensate for the current consumption of the phone, and the battery will not charge at all.

You can also provide a separate 12V input from a car battery with a cigarette lighter connector - and switch the sources with some kind of switch. I advise you to install an LED that will signal that the device is on, so as not to forget to turn off the battery after full charging - otherwise the losses in the converter will completely drain the backup battery in a few days.

This type of battery is not very suitable because it is designed for high currents - you can try to find a lower current battery, and it will be smaller and lighter.

Current stabilizer

Output current adjustment

Only available with adjustable output voltage version (LM2596ADJ). By the way, the Chinese also make this version of the board, with regulation of voltage, current and all kinds of indications - a ready-made current stabilizer module on LM2596 with short-circuit protection can be bought under the name xw026fr4.

If you do not want to use a ready-made module, and want to make this circuit yourself, there is nothing complicated, with one exception: the microcircuit does not have the ability to control current, but you can add it. I'll explain how to do this, and clarify the difficult points along the way.

Application

A current stabilizer is a thing needed to power powerful LEDs (by the way - my microcontroller project high power LED drivers), laser diodes, electroplating, battery charging. As with voltage stabilizers, there are two types of such devices - linear and pulsed.

The classic linear current stabilizer is the LM317, and it is quite good in its class - but its maximum current is 1.5A, which is not enough for many high-power LEDs. Even if you power this stabilizer with an external transistor, the losses on it are simply unacceptable. The whole world is making a fuss about the energy consumption of standby light bulbs, but here the LM317 works with an efficiency of 30% This is not our method.

But our microcircuit is a convenient driver for a pulse voltage converter that has many operating modes. Losses are minimal, since no linear operating modes of transistors are used, only key ones.

It was originally intended for voltage stabilization circuits, but several elements turn it into a current stabilizer. The fact is that the microcircuit relies entirely on the “Feedback” signal as feedback, but what to feed it is up to us.

In the standard switching circuit, voltage is supplied to this leg from a resistive output voltage divider. 1.2V is a balance; if Feedback is less, the driver increases the duty cycle of the pulses; if it is more, it decreases it. But you can apply voltage to this input from a current shunt!

Shunt

For example, at a current of 3A you need to take a shunt with a nominal value of no more than 0.1 Ohm. At such a resistance, this current will release about 1 W, so that’s a lot. It is better to parallel three such shunts, obtaining a resistance of 0.033 Ohm, a voltage drop of 0.1 V and a heat release of 0.3 W.

However, the Feedback input requires a voltage of 1.2V - and we only have 0.1V. It is irrational to install a higher resistance (the heat will be released 150 times more), so all that remains is to somehow increase this voltage. This is done using an operational amplifier.

Non-inverting op-amp amplifier

Classic scheme, what could be simpler?

We unite

Now we combine a conventional voltage converter circuit and an amplifier using an LM358 op-amp, to the input of which we connect a current shunt.

A powerful 0.033 Ohm resistor is a shunt. It can be made from three 0.1 Ohm resistors connected in parallel, and to increase the permissible power dissipation, use SMD resistors in a 1206 package, place them with a small gap (not close together) and try to leave as much copper layer around the resistors and under them as possible. A small capacitor is connected to the Feedback output to eliminate a possible transition to oscillator mode.

We regulate both current and voltage

Let's connect both signals to the Feedback input - both current and voltage. To combine these signals, we will use the usual wiring diagram “AND” on diodes. If the current signal is higher than the voltage signal, it will dominate and vice versa.

A few words about the applicability of the scheme

You cannot adjust the output voltage. Although it is impossible to regulate both the output current and voltage at the same time - they are proportional to each other, with a coefficient of "load resistance". And if the power supply implements a scenario like “constant output voltage, but when the current exceeds, we begin to reduce the voltage,” i.e. CC/CV is already a charger.

The maximum supply voltage for the circuit is 30V, as this is the limit for the LM358. You can extend this limit to 40V (or 60V with the LM2596-HV version) if you power the op-amp from a zener diode.

In the latter option, it is necessary to use a diode assembly as summing diodes, since both diodes in it are made within the same technological process and on the same silicon wafer. The spread of their parameters will be much less than the spread of parameters of individual discrete diodes - thanks to this we will obtain high accuracy of tracking values.

You also need to carefully ensure that the op-amp circuit does not get excited and go into lasing mode. To do this, try to reduce the length of all conductors, and especially the track connected to pin 2 of the LM2596. Do not place the op amp near this track, but place the SS36 diode and filter capacitor closer to the LM2596 body, and ensure a minimum area of ​​the ground loop connected to these elements - it is necessary to ensure a minimum length of the return current path “LM2596 -> VD/C -> LM2596”.

Application of LM2596 in devices and independent board layout

I spoke in detail about the use of microcircuits in my devices not in the form of a finished module in another article, which covers: the choice of diode, capacitors, inductor parameters, and also talked about the correct wiring and a few additional tricks.

Opportunities for further development

Improved analogues of LM2596

The easiest way after this chip is to switch to LM2678. In essence, this is the same stepdown converter, only with a field-effect transistor, thanks to which the efficiency rises to 92%. True, it has 7 legs instead of 5, and it is not pin-to-pin compatible. However, this chip is very similar and will be a simple and convenient option with improved efficiency.

L5973D– a rather old chip, providing up to 2.5A, and a slightly higher efficiency. It also has almost twice the conversion frequency (250 kHz) - therefore, lower inductor and capacitor ratings are required. However, I saw what happens to it if you put it directly into the car network - quite often it knocks out interference.

ST1S10- highly efficient (90% efficiency) DC–DC stepdown converter.

• Requires 5–6 external components;

ST1S14- high-voltage (up to 48 volts) controller. High operating frequency (850 kHz), output current up to 4A, Power Good output, high efficiency (no worse than 85%) and a protection circuit against excess load current make it probably the best converter for powering a server from a 36-volt source.

If maximum efficiency is required, you will have to turn to non-integrated stepdown DC–DC controllers. The problem with integrated controllers is that they never have cool power transistors - the typical channel resistance is no higher than 200 mOhm. However, if you take a controller without a built-in transistor, you can choose any transistor, even AUIRFS8409–7P with a channel resistance of half a milliohm

DC-DC converters with external transistor

Next part