Converting an ATX computer power supply into an adjustable power supply. The practice of converting computer power supplies into regulated laboratory ones. Radio engineering, electronics and circuits with your own hands Remaking the computer power supply

Or how to make a cheap power supply for a 100 W amplifier

How much will a 300 Watt ULF cost?

Depends on what for :)

Listen at home!

Bucks *** will be normal...

OMG! Is there any way to get it cheaper?

Mmmmm... We need to think...

And I remembered about a pulse power supply, powerful and reliable enough for ULF.

And I started thinking about how to remake it to suit our needs :)

After some negotiations, the person for whom all this was planned lowered the power level from 300 watts to 100-150 and agreed to take pity on the neighbors. Accordingly, a 200 W pulse generator will be more than enough.

As you know, an ATX format computer power supply gives us 12, 5 and 3.3 V. AT power supplies also had a voltage of “-5 V”. We don't need these tensions.

In the first power supply unit that came across, which was opened for rework, there was a PWM chip, beloved by the people - TL494.

This power supply was an ATX 200 W brand, I don’t remember which one. Not particularly important. Since my friend was “on fire,” the ULF cascade was simply purchased. It was a mono amplifier based on the TDA7294, which can output 100 W peak, which was quite satisfactory. The amplifier required bipolar power supply+-40V.

We remove everything superfluous and unnecessary in the decoupled (cold) part of the power supply, leaving the pulse shaper and the OS circuit. We install more powerful Schottky diodes and at a higher high voltage(in the converted power supply they were 100 V). We also put electrolytic capacitors in voltage exceeding the required voltage by 10-20 volts for reserve. Fortunately, there is a place to roam.

Look at the photo with caution: not all elements are worthy :)

Now the main “reworked part” is the transformer. There are two options:

• disassemble and rewind for specific voltages;
• solder the windings in series, adjusting the output voltage using PWM

I didn't bother and chose the second option.

We disassemble it and solder the windings in series, not forgetting to make a middle point:

To do this, the transformer leads were disconnected, ringed and twisted in series.

In order to see whether I made the wrong winding in a serial connection or not, I fired pulses with a generator and looked at what came out at the output with an oscilloscope.

At the end of these manipulations, I connected all the windings and made sure that from the middle point they have the same voltage.

We put it in place, calculate the OS circuit on the TL494 at 2.5V from the output with a voltage divider to the second leg and connect it in series through a 100W lamp. If everything works well, we add one more and then another hundred-watt lamp to the garland chain. For insurance against accidental parts flying :)

Lamp as a fuse

The lamp should blink and go out. It is highly advisable to have an oscilloscope to be able to see what is happening on the microcircuit and the drive transistors.

By the way, for those who don’t know how to use datasheets, let’s learn. Datasheets and Google help better than forums if you have developed the “Google” and “translator with an alternative point of view” skills.

I found an approximate power supply diagram on the Internet. The scheme is very simple (both schemes can be saved in good quality):

In the end it turned out something like this, but it's a very rough approximation and there are a lot of details missing!

The speaker design was coordinated and interfaced with the power supply and amplifier. It turned out simple and nice:

On the right - under the cut-off radiator for the video card and computer cooler there is an amplifier, on the left - its power supply. The power supply produced stabilized voltages of +-40 V on the positive side. The load was something like 3.8 Ohms (there are two speakers in the column). It fits compactly and works like a charm!

The presentation of the material is rather incomplete; I missed many points, since this happened several years ago. To help with repetition, I can recommend schemes from powerful car amplifiers low frequency - there are bipolar converters, usually on the same chip - tl494.

Photo of the happy owner of this device :)

He holds this column so symbolically, almost like an AK-47 assault rifle... Feels reliable and will soon join the army :)

We remind you that you can also find us in the VKontakte group, where every question will definitely be answered!

If you have an old computer power supply (ATX) at home, you shouldn’t throw it away. After all, it can be used to make an excellent power supply for home or laboratory purposes. Minimal modifications will be required and in the end you will get almost universal source supply with a number of fixed voltages.

Computer power supplies have a large load capacity, high stabilization and protection against short circuit.

Pinout of computer power supply outputs

The rework has begun

• - Screw terminals.
• - Resistors with a power of 10 W and a resistance of 10 Ohms (you can try 20 Ohms). We will use composites of two five-watt resistors.
• - Heat shrink tube.
• - A pair of LEDs with 330 Ohm quenching resistors.
• - Switches. One for networking, one for management

Computer power supply modification diagram

Let's get started

Watch a video of making a laboratory block with your own hands

Usually, ATX units assembled on TL494 (KA7500) chips are used to remake computer power supplies, but recently such units have not come across. They began to be collected for more specialized microcircuits, on which it is more difficult to adjust current and voltage from scratch. For this reason, an old 200W AT type unit that was available was taken for modification.

Remodeling stages

2. The self-starting parts of the primary circuit and the output voltage regulation circuit are soldered off the AT block board. All secondary rectifiers were also removed.

Special attention requires a shunt; the wires for adjustment and measurement must be connected directly to its terminals, since the voltage removed from it is small. In the diagram these connections are shown with purple arrows. The measured voltage for the control circuit is removed from the divider with correction to eliminate self-excitation in the control circuits.
The upper limit of the voltage setting is selected by resistors R38, R39 and R40. The upper limit of the current setting is selected by resistor R13.

3. A voltmeter-ammeter is used to measure current and voltage

4. The program for the microcontroller is written in SI (mikroC PRO for PIC) and provided with comments.

Construction and details

The drawings were made in the Frontplatten-Designer 1.0 program. The interstage transformer of the AT block is not modified. The output transformer of the AT block is also not modified, just the middle tap coming out of the coil is unsoldered from the board and isolated. The rectifier diodes were replaced with new ones indicated in the diagram.
The shunt was taken from a faulty tester and mounted on insulating stands on a radiator with diodes. The board for the voltmeter-ammeter is used from “Super simple ammeter and voltmeter on super affordable parts (auto range selection)” from Eddy71 with subsequent modification (paths were cut according to the diagram).

Observed features and disadvantages

Files

Not only for radio amateurs, but also simply in everyday life, you may need powerful block nutrition. So that there is up to 10A output current at a maximum voltage of up to 20 volts or more. Of course, the thought immediately goes to unnecessary computer blocks ATX power supply. Before you start remaking, find a diagram for your specific power supply.

Sequence of actions for converting an ATX power supply into a regulated laboratory one.

1. Remove jumper J13 (you can use wire cutters)

2. Remove diode D29 (you can just lift one leg)

3. The PS-ON jumper to ground is already installed.

5. Remove the 3.3-volt part: R32, Q5, R35, R34, IC2, C22, C21.

8. We change the bad ones: replace C11, C12 (preferably with larger capacity C11 - 1000uF, C12 - 470uF).

9. We change the inappropriate components: C16 (preferably 3300uF x 35V like mine, well, at least 2200uF x 35V is a must!) and resistor R27 - you no longer have it, and that’s great. I advise you to replace it with a more powerful one, for example 2W and take the resistance to 360-560 Ohms. We look at my board and repeat:

12. We separate the 15th and 16th legs of the microcircuit from “all the rest”, to do this we make 3 cuts in the existing tracks and restore the connection to the 14th leg with a jumper, as shown in the photo.

14. The core of cable No. 7 (power supply to the regulator) can be taken from the +17V power supply of the TL, in the area of ​​the jumper, more precisely from it J10/ Drill a hole into the track, clear the varnish and there. It is better to drill from the print side.

Actually, the idea of ​​​​making a laboratory power supply with adjustable output voltage and current from a computer one is not new. There are many options for such modifications on the Internet.

The advantages are obvious:

1. Such power supplies are literally “lying under your feet.”
2. They contain all the main components, and most importantly, ready-made pulse transformers.
3. They have excellent weight and size characteristics - such a transformer power supply would weigh more than 10 kg (this one is 1.3 kg in total).

True, they are not without their disadvantages:

1. Due to pulse conversion, the output voltage contains a rich spectrum of high-frequency interference, which makes them of limited applicability for powering radio stations.
2. They do not guarantee a low output voltage (less than 5 V) at low load currents.

And, nevertheless, such a power supply is perfect for powering automotive electronics at home, when testing and debugging electronic devices. And the presence of a current stabilization mode allows it to be used as a universal charger for a wide range of batteries!

Output voltage - from 1 to 20 V
Output current - up to 10 A
Weight 1.3 kg

First, let's figure out which power supplies are suitable for conversion. The best option for a laboratory power supply is the old AT or ATX power supply, assembled on a TL494 PWM controller (aka: μPC494, μA494, UTC51494, KA7500, IR3M02, MV3759, etc.) with a power of 200 - 250 W. Most of them are like this! Modern ATX12B, 350 - 450 W, of course, is also not a problem to remake, but they are still better suited for power supplies with a fixed output voltage (for example, 13.8 V).

To further understand the essence of the modification, consider the principle of operation of the power supply for a computer.

More or less standardized power supplies (PC/XT, AT, PS/2) for computers appeared in the early 80s thanks to IBM, and existed until 1996. Let's look at their operating principle according to the structural diagram:

AT power supply block diagram

The mains voltage is supplied to the power supply through an electromagnetic interference filter, which prevents the spread of high-frequency interference from the pulse converter to the supply network. It is followed by a rectifier and a smoothing filter, the output of which is constant voltage 310 V. This voltage is supplied to a half-bridge inverter, which converts it into square pulses and serves on primary winding step-down transformer T1.

Voltages from the secondary windings of the transformer are supplied to rectifiers and smoothing filters. As a result, at the output we obtain the necessary constant voltages.

When power is applied, at the initial moment, the inverter starts in self-generation mode, and after voltage appears on the secondary rectifiers, the PWM controller (TL494) is switched on, which synchronizes the operation of the inverter by supplying trigger pulses to the bases of the key transistors through the decoupling transformer T2.

The power supply uses pulse-width regulation of the output voltage. To increase the output voltage, the controller increases the duration (width) of the trigger pulses, and to decrease it, it decreases it.

Stabilization of the output voltage in such power supplies is often carried out by only one output voltage (+5 V, as the most important), sometimes by two (+5 and +12), but with priority +5 V. For this, the input of the controller comparator ( pin 1 of TL494, output voltage is supplied through the divider. The controller adjusts the width of the trigger pulses to maintain this voltage at the required level.

Also, the power supply has 2 types of protection system. The first - from exceeding the total power and short circuit, and the second, from overvoltage at the outputs. In case of overload, the circuit stops the pulse generator in the PWM controller (by supplying +5 V to pin 4 of the TL494).

In addition, the power supply contains a node (not shown in the diagram) that generates the POWER_GOOD signal (“voltage is normal”) at the output after the power supply reaches an operating mode that allows the processor to start in the computer.

The AT power supply (PC/XT, PS/2) has only 12 main wires for connecting to the motherboard (2 connectors of 6 pins each). In 1995, Intel was horrified to discover that existing power supplies could not handle the increased load and introduced a 20/24-pin connector standard. In addition, the power of the +3.3 V stabilizer on the motherboard to power the processor was also no longer enough, and it was moved to the power supply. Well, Microsoft introduced operating system Windows, Advanced Power Management (APM) power management modes... Thus, in 1996, the modern ATX power supply appeared.

Let's look at the differences between the ATX power supply and the old AT according to its structural diagram:

Block diagram of an ATX power supply

The Advanced Power Management (APM) mode required the abandonment of the mains switch and the introduction of a second pulse converter into the power supply - a standby voltage source of +5 V. This low-power power supply always works when the mains plug is plugged into the network. The primary voltage comes from the same rectifier and filter as the main inverter.

In addition, power to the PWM controller in ATX comes from the same standby source (not stabilized 12 - 22 V), and there is no autostart of the inverter. Therefore, the power supply starts only in the presence of start pulses from the controller. The main power supply is turned on by turning on the pulse generator of the PWM controller with the PS_ON signal (shorting it to ground) through a protection circuit.

That's all the main differences.

How to choose a power supply for conversion?

As you know, power supplies are made in China. And this may entail the absence of some components that they considered “superfluous”:

1. There may be no EMI filter at the input. The most important thing in the filter is the inductor wound on a ferrite ring. Usually, it is clearly visible through the fan blades. Instead, there may be wire jumpers. The presence of a filter is an indirect sign of a high-quality power supply!

EMI filter elements

2. Also, you need to look at the size of the step-down transformer (the larger one). The maximum power of the power supply depends on it. Its height should be at least 3 cm. There are power supplies with a transformer less than 2 cm high. The power of these is 75 W, even if it says 200.

3. To check the functionality of the power supply, connect a load to it. I use car headlight bulbs with a power of 50 - 55 W and a voltage of 12 V. Be sure to connect one to the +5 V circuit (red wire), and the second to the +12 V circuit (yellow wire). Turn on the power supply. Disconnect the fan connector (or, if the Chinese saved on it, just stop it with your hand). The power supply should not beep.

After a minute, disconnect it from the network and feel with your hand the temperature of the radiators and group filtering choke in the secondary voltage filter. The throttle should be cold, and the radiators should be warm, but not hot!

I used a power supply from 1994 with a power of 230 W - they didn’t save money back then.

Reworking the power supply

You need to start by cleaning the power supply from dust. To do this, disconnect (unsolder) the network wires and wires to the 110/220 switch from the board - we will no longer need it, because in the 220 V position the switch is open. Remove the board from the case. Vacuum cleaner, hard brush, and go!

Next, you need to try to find an electrical schematic diagram your power supply, or at least as similar as possible to it (they do not differ significantly). It will help you navigate the values ​​of the “missing” components. I do not rule out that, like me, you will have to copy some components from the board.

Diagram of an electromagnetic interference filter, rectifier and primary voltage filter, and inverter after modification

The values ​​of the replaced components are highlighted in red in the diagram. Newly installed components have positional designations highlighted in red.

1. Check that all capacitors and inductors are present in the EMI filter. If missing, install them (I only had C2 missing). I also installed a second, additional noise filter, made in the form of a socket for connecting the power cord.

2. Look at the types of diodes used in the rectifier (D1 - D4). If there are diodes with a current of up to 1 A (for example, 1N4007) - replace them with at least 2 Amp ones, or install diode bridge. I had a 2-amp bridge.

3. In the vast majority of power supplies, capacitors with a capacity of no more than 200 μF (C5 - C6) are installed in the primary voltage filter. For giving back full power, replace them with capacitors with a capacity of 470 - 680 μF, suitable in size, with a voltage of at least 200 V. Preference should be given to the 105°C group.

4. Transistors in a half-bridge inverter (Q1, Q2) can be very diverse. In principle, most of them are not heated criminally. To reduce heating, they can be replaced with more powerful ones - for example, 2SC4706, installing them on a radiator through insulating gaskets. I went even further and replaced both radiators with more efficient ones.

5. While testing the power supply under maximum load, capacitor C7 heated up and burst (usually 1 µF 250 V). This capacitor should not get hot at all. I think it was faulty, but I still replaced it with 2.2 uF 400 V.

Now let's consider block diagram converted power supply:

Block diagram of a laboratory power supply

To modify, we will need to remove all secondary rectifiers except one (although replacing almost all components in it), redo the protection circuit, add a control circuit, a shunt and measuring instruments. Elements of the POWER_GOOG schema can be deleted. Now more details.

To remove the output voltage, the 12-volt winding of the step-down transformer T1 is used. But it is more convenient to install the rectifier and filter in place of the 5-volt one - there is more space for diodes and capacitors.

The secondary voltage rectifier and filter, after modification, should look like this:

Secondary voltage rectifier circuit after modification

1. Unsolder all elements of rectifiers and filters +5, +12 and -12 V. With the exception of damper circuits R1, C1, R2, C2 and R3, C3 and inductor L2. Subsequently, with an output voltage of about 20 V, I noticed resistor R1 heating up and replaced it with 22 Ohms.

2. Cut the tracks leading from the 5-volt taps of the winding of transformer T1 to the +5 V rectifier diode assembly, while maintaining its connection to the -5 V rectifier diodes (we will need it later).

3. In place of the diode assembly of the +5 V rectifier (D3), install an assembly on Schottky diodes for a current of 2x30 A and a reverse voltage of at least 100 V, for example, 63CPQ100, 60CPQ150. (The standard 5-volt diode assembly has a reverse voltage of only 40 V, and the standard diodes in the 12 V rectifier are designed for too low a current - they cannot be used.) This assembly practically does not heat up during operation.

4. Connect the terminals of the 12-volt winding with the installed diode assembly using thick wire jumpers. The damper circuits R1, C1 connected to this winding are saved.

5. In the filter, instead of the standard ones, install electrolytic capacitors (C5, C6) with a capacity of 1000 - 2200 μF for a voltage of at least 25 V. And also add ceramic capacitors C4 and C7. Install a 100 Ohm load resistor with a power of 2 W instead of the standard one.

6. If, when checking the power supply under load, the group filtering choke (L1) did not heat up, then just rewind it. Wind all the windings from it, counting the turns. (Usually, 5 V windings contain 10 turns, and 12 V windings contain 20 turns.) Wind a new winding with two wires folded together with a diameter of 1.0 - 1.3 mm (similar to a standard 5 volt one) and a number of turns of 25-27.

If the inductor was heated, then its core is damaged (powdered iron has such a problem - it “sinters”), then you will have to look for a new core made of powdered iron (not ferrite!). I had to buy a white ring core with a slightly larger diameter and wind a new winding. It doesn't heat up at all.

7. Choke L2 remains standard, from a 5-volt filter.

8. To power the fan, a 5-volt winding is used, and the rectifier wiring is -5 V, which we convert to +12. The standard diodes are used, from the -5 V rectifier (D1, D2), they must be soldered with reverse polarity. The choke is no longer needed - solder the jumper. And in place of the standard filter capacitor, install a capacitor with a capacity of 470 uF 16 V, of course, with reverse polarity. Throw a jumper from the filter output (formerly -5 V) to the fan connector. Directly near the connector, install a ceramic capacitor C9. My fan voltage is +11.8 V, and at low load currents it decreases.

9. In the power supply circuit of the PWM controller (Vcc), it is necessary to increase the capacitance of filter capacitors C10 and C11. The voltage from capacitor C10 (Vdd) is used to power digital ammeter and a voltmeter.

The protection circuit for exceeding the total power remains unchanged. Only the output overvoltage protection circuit is changed. Here is the final diagram:

Scheme of the protection unit after modification

When the load on the inverter increases above the permissible limit, the pulse width at the middle terminal of the isolation transformer T2 increases. Diode D1 detects them, and the negative voltage across capacitor C1 increases. Having reached a certain level (approximately -11 V), it opens transistor Q2 through resistor R3. A voltage of +5 V will flow through an open transistor to pin 4 of the controller and stop the operation of its pulse generator. In your power supply, such protection may be organized differently. In any case, you don't need to touch it.

All diodes and resistors suitable from the secondary rectifiers to the Q1 base are unsoldered from the circuit, and a zener diode D3 is installed at a voltage of 22 V, for example, KS522A, and resistor R8.

In the event of an emergency increase in voltage at the output of the power supply above 22 V, the zener diode will break through and open transistor Q1. This, in turn, will open transistor Q2, through which a voltage of +5 V will be supplied to pin 4 of the controller, and will stop the operation of its pulse generator.

All that remains is to assemble the control circuit and connect it to the PWM controller.

The control circuit consists of two amplifiers (current and voltage), which are connected to the standard inputs of the controller error comparators. He has 2 of them (pins 1 and 16 of TL494) and they work via OR. This allows you to obtain both voltage and current stabilization. Final control unit diagram:

Control unit diagram

The operational amplifier DA1.1 is used to assemble a differential amplifier in the voltage measurement circuit. The gain is selected in such a way that when the output voltage of the power supply changes from 0 to 20 V (taking into account the voltage drop across the shunt R7), the signal at its output changes within 0...5 V. The gain depends on the ratio of the resistances of resistors R2/R1 =R4/R3.

Please note: for correct voltage measurement, resistors R1 and R3 are connected with separate thin wires directly to the output voltage connecting terminals.

The operational amplifier DA1.2 is used to assemble an amplifier in the current measurement circuit. It amplifies the magnitude of the voltage drop across shunt R7. The gain is selected in such a way that when the load current of the power supply changes from 0 to 10 A, the signal at its output changes within 0...5 V. The gain depends on the ratio of the resistances of resistors R6/R5.

As a current sensor (R7), I used a standard 75SHIP1500.5 measuring shunt with a fairly low resistance of 1.5 mOhm. Therefore, in the measurement circuit I also included the connecting wires that connect the shunt. This made it possible to eliminate the differential amplifier and reduce the number of wires. Resistor R5 is connected directly to ground near the op-amp, and the non-inverting input (pin 5) is connected to the same wire (from R3) going to the negative terminal.

Measuring shunt 75SHIP1500.5

When using a shunt with a different resistance and with a different length of connecting wires, you will need to select resistor R5 so that maximum current stabilization corresponded to 10 A.

Signals from both amplifiers (voltage and current) are supplied to the inputs of the error comparators of the PWM controller (pins 1 and 16 of DA2). To set the required voltage and current values, the inverting inputs of these comparators (pins 2 and 15 of DA2) are connected to adjustable dividers reference voltage(variable resistors R8, R10). The +5 V voltage for these dividers is taken from the internal reference voltage source of the PWM controller (pin 14 of DA2).

Resistors R9, R11 limit the lower adjustment threshold. Capacitors C2, C3 eliminate possible “noise” when turning the variable resistor motor. Resistors R14, R15 are also installed in case of a “break” of the variable resistor motor.

A comparator is assembled on the operational amplifier DA1.4 to indicate the transition of the power supply to current stabilization mode (LED1).

In the circuit I used quad operational amplifier LM324A, but you can use others that work in wide range supply voltages, for example, two dual LM358. Power to it (Vcc) is supplied from the power circuit of the PWM controller (from pin 12 of DA2) which varies within 5...25 V, depending on the output voltage of the power supply.

Adjustment elements R8 - R11, as well as capacitors C2 and C3, are located on a small board screwed to the front panel of the power supply. All other elements of the circuit are located in free space printed circuit board power supply.

To connect amplifiers to a PWM controller (DA2), you must first unsolder all standard components going to pins 1, 2, 3, 15 and 16 from it.

To measure and display the output voltage and current, I used ready-made digital voltmeter and an ammeter, connected according to the circuit according to the instructions attached to them. Power is supplied to them from capacitor C10 (see diagram of secondary rectifiers). If you have an ATX power supply with a standby power supply at your disposal, then supply power to the meters (Vdd) from this source - it has an unstabilized voltage output of +12 - 22 V.

To connect these devices, it is convenient to use the connectors for Floppy drives available on the standard cables of the AT power supply.

Please note that the measuring leads of the voltmeter are connected with separate thin wires directly to the output terminals of the power supply. And the measuring leads of the ammeter go directly to the measuring contacts of the shunt.

Part of the standard metal case (bottom and side wall) of the power supply in my design serves as a chassis for the board and for the shunt.

To reduce the level of high-frequency interference, ceramic capacitors with a capacity of 1 μF are located directly at the output terminals (C6, C7 in the control unit diagram).

For my power supply, I used a ready-made case with a carrying handle. A Ø50 mm fan is used for cooling. It drives air inside the housing. To do this, the necessary hole was cut in the case opposite the radiators, and on the opposite side and back wall, holes were drilled for air outlet. Design ideas depend only on your taste.

If you intend to use such a power supply for radio stations, then I strongly recommend keeping the standard one in the design metal case- it perfectly shields and reduces the level of electromagnetic interference emitted by the inverter.