The principle of operation of a computer cooler with three wires. How to properly organize cooling in a gaming computer

In the process of resuscitation and modernization of the Solntsev amplifier, I had to get rid of a bulky power supply made on a TS-180 transformer. Was made impulse block power supply on IR2153 with a power of 200 W. However, during operation with a removed power of about 130 W, heating of the pulse transformer was detected. Not critical, but still present. In addition, the stabilizers L7815, L7915 warmed up quite noticeably. Installing large heatsinks was not allowed by tight mounting on the board.

To eliminate this effect, I decided to use a cooler. The choice settled on a small-sized fan with a power of 0.96 W with a power supply of 12 volts and a current consumption of 0.08 A. Since the transformer PSU for it will have unacceptable weight and size dimensions, I decided to assemble it with a quenching capacitor.

Scheme

A transformerless power supply in the general case is a symbiosis of a rectifier and a parametric stabilizer. Capacitor C1 for alternating current is a capacitive (reactive, i.e. not consuming energy) resistance Xc, the value of which is determined by the formula:

where f- network frequency (50 Hz); FROM- capacitance of the capacitor C1, F. Then the output current of the source can be approximately determined as follows:

where Uc- mains voltage (220 V).

With a current consumption of 0.08 A, the capacitance C1 should have a rating of 1.2 microfarads. Its increase will allow you to connect a load with a large current consumption. Approximately, you can focus on 0.06 A for each microfarad of capacitance C1. I had 2.2 microfarads at 400 volts at hand.

Resistor R1 is used to discharge the capacitor after the PSU is turned off. There are no special requirements for it. Nominal 330 kOhm - 1 Mohm. Power 0.5 - 2 W. In my case, 620 kOhm 2 watts.

Capacitor C2 serves to smooth out the ripples of the voltage rectified by the bridge. The rating is from 220 microfarads to 1000 microfarads with an operating voltage of at least 25 volts. I set 470 microfarads for a voltage of 25 volts.

As rectifier diodes, 1N4007 from a used energy-saving lamp were used.

A zener diode (12 Volts) serves to stabilize the output voltage, and by replacing it, you can achieve almost any required voltage at the PSU output.

When assembling the circuit, it should be borne in mind that the fan connection must be made correctly initially. An error in the wrong polarity of soldering the fan wires will cause the fan to fail. And the connection itself (soldering) should be done in advance, since the voltage on Idling at the fan connection points it can be 50-100 volts. If the polarity is unmistakable (red wire, this is a positive power bus), then when connected to a 220 V network, the fan will have approximately +12 volts.

The printed circuit board is made by the LUT method. Etching was carried out with hydrogen peroxide, citric acid and table salt at the rate of 50 ml of peroxide, 2 tsp. acid and a teaspoon of salt.

In addition, I give a diagram (maybe someone will need it) to adjust the fan speed.

In fact, this is a voltage regulator supplied to the fan motor. Changing the voltage leads to a change in the fan speed. A constant resistor R2 is specially introduced into the circuit, the purpose of which is to limit the minimum fan speed, so that even at the lowest speed, i.e. at the lowest voltage, to ensure its reliable start.

Instead of a preface While working on a computer based on the P166MMX, among other things, I found a non-working power supply fan. From the words of the owner, it turned out that the fan started knocking about a year ago - which was confirmed by physical damage to the blades and the inner surface of the case, the knock stopped almost immediately - along with the life of the fan itself, the owner himself immediately forgot about it. The power reserve of a conventional 200-watt power supply was enough to ensure the performance of the system unit without leaving the operating temperature. Since then, technology has not stood still, processor frequencies have increased by an order of magnitude, the total power consumption of system units has increased, and only the nameplate power of power supplies has not increased significantly, which means that the temperature conditions of operation of key elements are quite difficult, and a malfunction of the power supply fan can lead to irreparable consequences. The impetus for the development of the device described below was the installation of a second fan in a standard power supply unit, which is blown from the system unit and the operation of both fans at a supply voltage of 9V. If the operation of a regular power supply can be checked by substituting a palm under the blown air flow, then it is quite difficult to check the operation of the second one even visually. From this came the main "technical task" - to provide visual control of the fan operation mode. From the very beginning, cost characteristics were not brought to the fore, but in the end it turned out that the cost of the finished device did not exceed the cost of the fan itself. The occupied volume of the finished device, which in addition to signaling the fan operation mode in its final form, performs a number of functions - it provides the fan motor with a reduced supply voltage with filtering of impulse noise from it and a smooth start when turned on, does not exceed the volume of a matchbox.

With minimal refinement of the circuit, the device can provide automatic speed control from temperature.

Inside the fan

The electrical circuits of all fans are approximately the same, two of their options can be found in the diagrams below from the Radio magazine:


In the same article ("Repair of fans of electronic devices" by R. Aleksandrov), you can also get acquainted with the principle of their operation.

Real fan circuits can differ only in the type of elements used and the degree of their integration. For the most part, "two-wire" fans are made similarly to the first scheme. "Three-wire" fans have an additional low-power transistor in their circuit, connected according to the "open (unconnected) collector" scheme - typical schemes the inclusion of such fans can be found, for example, in the "datasheet" on the monitoring chip of the W83781D motherboard.


This is how the board of one of these such fans looks like (view from both sides):


In the circuit of this fan, the Hall sensor is integrated with key transistors, the signal for the speed sensor is taken from a low-power transistor from the ZGA series.


A typical switching circuit will be kept in mind when developing a fan motor rotation sensor. Here is his diagram:


When the fan is running, both LEDs will glow, by selecting the resistance of the resistor R4 they achieve the same brightness of the glow, while when the engine is stopped, a change in the brightness of the glow should be noticeable. If the engine stops, only one of them will light up. When driving with interruptions, blinking of the LEDs will be noticeable. When connected to the gap between R2 and the base of the transistor capacitor with a capacity of about 50 microfarads, when the speed changes, the brightness of the LEDs will also change. When using a few more radio elements, it is possible to provide an emergency shutdown of the system unit when the fan exits the operating mode or the use of a spare one.

As a circuit for the rotation sensor of a "two-wire" fan, one could take this one (however, this circuit was also suitable for a "three-wire" fan).


In this case, the brightness of the LED glow would inversely depend on the current consumption of the fan - the maximum glow in the event of a break in the fan power circuit, the absence of glow in the event of a short circuit. Setting up such a device would be reduced to selecting the resistances of two resistors - by selecting R1 (~ 5 Ohm) we set the voltage drop across it at the rated current consumption of the fan in the region of 0.5-0.75V, by selecting R2 we achieve a noticeable change in the brightness of the LED when the engine stops. The circuit has the "right to life", but we will go the other way - we will turn the "two-wire" fan into a "three-wire" one, without changing anything in its circuit. To do this is easy enough. To remove a signal whose frequency is proportional to the speed of the fan impeller, a collector of any of the key transistors is suitable. In this case, the rotation sensor can be the first circuit with the resistor R1 removed from it without changing the parameters of the remaining elements of the circuit. It remains only to remove the impeller to access the circuit elements, find the collector of one of the transistors, solder and fix the wire and reassemble. At the same time, if the fan has already been in operation, carry out routine maintenance to remove dust and lubricate the shaft.


We find the necessary output of the transistor by checking the continuity of the outputs relative to the positive wire of the power supply of the circuit for the presence of a low-resistance circuit with a resistance of ~ 60 Ohms and solder the wire to it.


On this, the revision of two-wire fans can be considered complete. If you do not forget how to assemble it.

Noise control

A rare user, having installed a fan in the case, does not start the fight against noise. Moreover, as a rule, this consists in connecting the power supply of the engine between the wires + 12V and + 5V. As a rule, any arguments of opponents of such a connection are not taken into account by its supporters. I also decided to "invest my penny" in this dispute. To do this, I slightly modified the input circuits of the old Genius SM32x sound card and used it as an oscilloscope to measure ripples on both +12V and +5V power rails simultaneously using the Sony Sound Forge 7.0 audio editor.
The first "oscillogram" refers to the case of connecting the fan to the +12V and 0 buses.


The upper waveform refers to the +12V rail, the lower waveform to the +5V rail.

And here is what the oscillogram looks like when the fan is connected to the +12V and +5V buses.


If the + 12V bus calmly endured such a connection, then pay attention to the pulses that appeared on the + 5V bus in positive values. These pulses are nothing more than switching noise of the key transistors of the motor control circuit and impulse noise of its coils. These interferences are quite strong - when measuring the peak value using an S1-55 oscilloscope, a value of more than 0.2V was obtained for the switching noise of this fan - when using a processor cooler to cool an integrated 4-channel power amplifier with a total power of 120W, powered through an integrated stabilizer KR142EN8 background was removed only when a capacitor with a capacity of at least 1000 microfarads was connected. It is this capacitance value that is also recommended for the fan motor supply voltage reduction circuit, which will be discussed below. And now let's find out how the performance of the cooler decreases when the power is reduced. To do this, we will remove the dependences of the impeller rotation speed on the motor supply voltage for different fans (all of them are shown in the first photo), the frequency / voltage dependence for the "two-wire" fans that were under alteration was similar to the dependence for the third fan with a nominal speed of 2400 rpm. /min




We see that the rotational speed depends linearly on the supply voltage up to the border of the working section of the supply voltage. However, the dependence of the passing air volume on the rotational speed can be taken as a quadratic one - based on this, it can be understood that the slower the engine, the less performance we will lose with the same decrease in the supply voltage compared to faster ones. With a decrease in the supply voltage, in my opinion, it is enough to stop at the border of 8-9 volts - firstly, it is here that there is a sharp decrease in acoustic noise from the rotating impeller, and, secondly, the performance drop is not so noticeable. Since, in addition to reducing acoustic noise, we are also pursuing the task of reducing impulse noise, and we have to connect a capacitor in parallel with the supply terminals of the fan motor large capacity, then it is necessary to somehow limit the initial starting current, the value of which will be the sum of the capacitor charge current and the starting current of the motor itself - the measured values ​​​​of the starting current for different fans gave its value not less than twice the value of the rated current. The best solution to this problem should be recognized as the use of a powerful field-effect MOSFET transistor - due to the large input resistance of the gate, you can limit yourself to small capacitors in time-setting circuits - up to 100 μF.


The final edition was the following scheme, the setting of which is to select the capacitance C1, in which there is a smooth increase in the current consumption when turned on. Depending on the type of field effect transistor, you can get an output voltage in the range of 9.5-8.5 V. I chose the IRFZ24N (in terms of price / technical characteristics) - with it, the output voltage at an input voltage of 12V is 8.8V. This circuit can be slightly modified - the gate voltage can be supplied from the middle output of the potentiometer connected to the supply wires, by shunting one of the arms of this potentiometer with a thermistor, you can get a voltage directly or inversely proportional to the temperature change at the output. In addition, if necessary, increase the output voltage, you can shunt the drain and source terminals with a resistor with a resistance of about 50 ohms.
The final device looks like this:


The field-effect transistor is mounted on a copper flange soldered to the contact pad from a similar case, before soldering which it is necessary to chamfer along its contour. The temperature regime of the transistor under load in "one fan" with such cooling is 40 degrees. Mounting is done on a double-sided board using surface-mount radio elements (from old ISA device boards). Board fastening - in place. The LEDs are placed on the front panel.

Automatic activation of the standby fan

Consider the complete scheme of the resulting device.


We see that if we exclude the resistor R1 from the circuit, then it is possible to open the VT2 key using a circuit that would work according to the following algorithm - there is a signal to open the key when the other fan motor stops, there is no signal - during normal operation of the fan motor. We implement this algorithm using the simplest fan sensor status detector.


In the presence of rotation, the capacitor C2 is recharged, which causes the appearance of an alternating component on the resistor R6, the positive half-wave of which opens the transistor VT2 and recharges the capacitor C3, which does not allow the transistor VT2 to close during the negative half-wave, which through the VD3 diode "sits down" to circuit zero. For a more accurate operation of the detector in place of this diode, it is better to use diodes with a low forward voltage, for example, germanium type D9. I used a D18 diode. In the absence of rotation, the capacitor C3 is discharged through resistors R6 and R7, as well as through the emitter junction VT2. In this case, the voltage on the VT2 collector rises, which leads to the opening of the field-effect transistor and the supply voltage to the backup fan.
By selecting the capacitance of the capacitor C3, it is possible to ensure the "testing" of the operation of the backup fan at the first switch-on during the charging time of this capacitor.
When the main fan is replaced with a serviceable backup fan, it stops again.

Here is a complete diagram of such a device:



And here is his appearance assembled:


Two fan sensor boards are installed on the cross-board, on which the detector is located. Fans are connected to standard three-prong fan plugs. Power can be supplied, for example, through a standard fan connector (as in the picture). Instead of pairs of LEDs, two-anode two-color LEDs can be used.

Literature on the topic


  1. Magazine "Radio" №12, 2001 "Repair of fans of electronic devices", R. Aleksandrov, pp. 33-35.
  2. Magazine "Radio" №2, 2002 "Audible fan failure alarm", D. Frolov, p.34

The operation of most PC electronic components is accompanied by increased heat generation. Most effective way cooling is active (forced, fan). But does everyone know how to properly connect a cooler to a computer power supply? Here we will deal with this in detail.

In principle, the work is simple - you just need to install the cooler in place and connect its wires of a certain color to the necessary contacts of the computer's power supply. But there are a number of nuances, without taking into account which correct connection not to do.

Firstly, there are computer fans with different connector designs on sale. They can have from 2 to 4 contacts. But the outputs of the PC power supply to which the connection is made are always four.

Secondly, the cooler wires can have one of two color marking options.

Thirdly, laptop processors require special temperature regime. Therefore, their fans turn on only periodically, as needed. With desktop computers, it's different. The task of the cooler is to provide continuous cooling of their electronics, that is, we are talking about its constant operation. And here such an indicator as the "noisiness" of the fan already comes to the fore. That is why it is desirable to reduce the nominal voltage supplying the cooler (standard +12 V) at least a little. This will not significantly affect the cooling efficiency of the system unit, but the user's comfort will be ensured.

Connection order

Power off the computer

Simply turning off the PC with a button the best solution. It must be completely isolated from the mains, that is, unplug it from the socket or put the switch in the “off” position.

Fix the cooler in place

To do this, you need to dismantle the side cover, install the fan in the place intended for it and fix it with bolts. It is necessary to pay attention to the indicator of the direction of rotation of its impeller (arrow on the end part of the cooler). Depending on how the fan is positioned, the airflow can be directed either into the computer (pull-in) or out of it. And this directly affects the cooling efficiency of the system unit electronics. In order not to be mistaken, it is advisable to replace the cooler "one to one", therefore it is not advisable to remove the faulty one before purchasing a new one.

Connecting to the power supply

The author does not know which fan the reader will install to replace the failed one. It can be a used product from another computer or purchased, but they all happen various modifications. Therefore, only possible options are considered below.

The photo shows the pinout of the cooler connectors depending on the number of contacts. If their number does not match the conclusions of the computer's power supply, you will have to use adapters. In brackets - the color designation of the conductors according to the second option.

Wire marking

  • +12 V - Kr (Zhl).
  • -12 V is always black.
  • Tachometer line - Zhl (Zel).
  • Speed ​​control - blue.

Computer power supply pinout
Cooler connector pinout

If the fan is quite noisy, then it can be powered not with 12 V, but with a seven (connection to the extreme terminals) or five (to red). The ground wire, as noted above, is always black.

Some articles give recommendations for changing the speed of rotation of the impeller using limiting resistors. Their power is about 1.2 - 2 W, and the dimensions are appropriate. Already - not very convenient. In general, this is understandable. But what are the criteria for choosing the resistance value, if the user with electronic equipment is at best only “you”? And at worst, no way.

The author advises not to experiment and, if desired, include a diode in the circuit. Regardless of the type, it will necessarily provide a certain voltage drop of the order of 0.6 to 0.85 volts. If you want to reduce the rating even more, you can use 2 - 3 semiconductors in series. To do this, you do not need to engage in engineering calculations or consult with a specialist.

Computer fan device. How to get a tachosignal if it is not brought out

The fan in a modern computer is perhaps the most massive device. Where are they not installed? Power supply, processor cooler, video card cooler, often used for additional cooling of the hard drive, actually 1-2 pieces are mounted in the case. A total of at least 4 pieces.

Wouldn't you like to see how it's made? Fan, so to speak, on the inside?

For experiments, let's take a pair of the cheapest 80 mm fans on sleeve bearings (sleeve bearing), the first is an ordinary two-wire fan with a molex connector, the price is 25-35 rubles, the second is 1.5-2 times more expensive, three-wire, with a tachometer. At the same time, let's see how justified such a big difference in price.

The process of disassembling the fan is simple:
- remove the branded sticker,
- take out the rubber plug-seal
- carefully remove (with a needle or something thin and sharp) a split fluoroplastic washer from the rotor shaft.
On this First stage disassembly is completed - you can remove the fan impeller.

What we see:
1. the motor windings together with the magnetic circuit are fixedly mounted on the fan casing;
2. Inside the impeller there is an annular magnet with a yoke closing the magnetic flux.

This design of the engine is called with an external rotor.

The usual brushes for commutator DC motors are nowhere to be seen. How does the current switch in the windings so that the rotor rotates? To switch the current in the windings, a special microcircuit based on a Hall sensor is used. The Hall sensor is made of a semiconductor material that is sensitive to a magnetic field.

To rotate the rotor, it is necessary to switch the stator windings strictly at a certain moment and in a given sequence.
The position of the rotor (impeller with a ring magnet) is determined by the Hall sensor, which also controls the switches located in the microcircuit. The ring magnet has 4 poles - N-S-N-S, so when the poles pass by the Hall sensor, it generates two pulses per revolution of the rotor. At the outputs of the microcircuit switching the windings, two antiphase sequences of pulses are formed. The signal from any of these outputs can be used to generate a tachosignal - this was done in the microcircuits of earlier designs. At present, microcircuits are also produced with the output of a tachosignal.

Let's take a closer look at the fan boards.
The following figure on the left shows a fan board with a tachometer and next to it in the figure on the right is its diagram:


Everything is very simple - the microcircuit switches the windings, and has a built-in tachometer output. The tachometer output is an open collector npn transistor. The concept of "open collector" means that it is not connected anywhere, it hangs in the air. Such an output is usually used to match voltage levels. Read more about the output of the tachometer and its practical use in the next article.

The following figures show the board and fan circuit without a tachometer output. Empty places for installing elements and a fat jumper instead of a diode are striking.


A simple analysis shows that if you install 2 resistors and 1 transistor in empty places, we get a fan with a tachometer output. It is also desirable to install the diode in its intended place, and remove the jumper - this will reduce the level of interference in the +12 V circuit (although the speed of rotation of the impeller will slightly decrease). After all these changes, the board and circuit will look like in the following figures:


The values ​​of the resistors R1, R2 may need to be clarified for a particular fan. Transistor VT1 can be used in almost any low power n-p-n type.

But even if the board did not provide space for the installation of these elements, they can always be added by surface mounting.

In the fans of the described design, a tacho signal is always implicitly present - this is a signal for switching the windings. Therefore, it is enough to add a few penny details and get a tachometer signal outside the fan. The price of all these additional elements is approximately 1.5-2 rubles at retail, and for mass production - 50 kopecks. Draw your own conclusions about the validity of a 1.5-2-fold difference in the prices of fans with and without a tachometer output.

When assembling a power supply in a case from a computer PSU, I decided to use a cooler from a PC for cooling. There were no suitable windings on the transformer, it was not possible to wind it up, so I decided to connect it separately. Near the back wall and the installed transformer, there was an empty place with two racks, and it was planned to install the cooler power supply scarf there. The power supply circuit of the cooler itself is a standard transformerless one with a quenching capacitor.

Ballast capacitor C1 (non-polar, film or metal-paper, for a voltage of at least 400 V, and for reliability it is better for all 630 V) at a voltage of 220 V passes a current of the order of 0.07 A for each microfarad of its capacitance. The exact formula is "didn't know, didn't know and forgot", but for practical application this figure is quite enough (resistor R1 serves exclusively to discharge the capacitor after turning it off). In fact, such reactance is an alternator (the capacitor simply will not let more current through). It happens that it can provide up to 0.14 A. If more is needed, the capacitance C1 increases.

The voltage is rectified by the diode bridge VD1 and smoothed out by the capacitor C2 for a voltage of at least 16 V. The zener diode VD2 serves to protect C2 from breakdown if something happens to the cooler. The speed of revolutions is regulated by the current shunt R2, which "sucks" part of the current into itself parallel to the cooler. R3 can be set if you do not need to reduce the speed to zero. Denominations to select "in place". The power allocated to R2, R3 at a current of 0.14 A will not exceed 1.7 W.

As for the design, I have a power supply for 0-30 V 3A, and an additional 12 W power supply for a 6 V soldering iron. There are two windings for 26 V and 6 V 3 A, so that a good transformer simply does not lie around, I decided to attach it, and a mini soldering iron became more often needed. In the archive on the forum there is a description of the scheme, a print of a simple scarf and a photo. Specially for the site radio circuits- Igoran.

Discuss the article CONNECTING A COMPUTER COOLER TO 220V