Cooling fans are now found in many household appliances, be it computers, stereos, home theaters. They do their job well, they cool the heating elements, but they make a heart-rending and very annoying noise. This is especially critical in music centers and home theaters, because the fan noise can interfere with enjoying your favorite music. Manufacturers often save money and connect cooling fans directly to the power supply, from which they always rotate at maximum speed, regardless of whether cooling is required at the moment or not. The solution to this problem can be quite simple - build your own automatic cooler speed controller. It will monitor the temperature of the radiator and only turn on the cooling if necessary, and if the temperature continues to rise, the regulator will increase the speed of the cooler up to the maximum. In addition to reducing noise, such a device will significantly increase the life of the fan itself. You can also use it, for example, when creating homemade powerful amplifiers, power supplies or other electronic devices.
Scheme
The circuit is extremely simple, contains only two transistors, a pair of resistors and a thermistor, but works great nonetheless. M1 in the diagram is a fan, the speed of which will be regulated. The circuit is designed to use standard 12-volt coolers. VT1 is a low-power n-p-n transistor, for example, KT3102B, BC547B, KT315B. Here it is desirable to use transistors with a gain of 300 or more. VT2 is a powerful n-p-n transistor, it is he who commutes the fan. You can use inexpensive domestic KT819, KT829, again, it is advisable to choose a transistor with a high gain. R1 is a thermistor (also called a thermistor), a key link in the circuit. It changes its resistance depending on the temperature. Any NTC thermistor with a resistance of 10-200 kOhm is suitable here, for example, the domestic MMT-4. The value of the trimmer resistor R2 depends on the choice of thermistor, it should be 1.5 - 2 times larger. This resistor sets the fan activation threshold.
Making a regulator
The circuit can be easily assembled by surface mounting, or you can make a printed circuit board, as I did. To connect the power wires and the fan itself, terminal blocks are provided on the board, and the thermistor is displayed on a pair of wires and is attached to the radiator. For greater thermal conductivity, you need to attach it using thermal paste. The board is made by the LUT method, below are several photos of the process.
Download the board:
(Downloads: 833)
After the board is made, parts are soldered into it, as usual, first small, then large. It is worth paying attention to the pinout of the transistors in order to solder them correctly. After completing the assembly, the board must be washed from the remnants of the flux, ring the tracks, make sure that the installation is correct.
Customization
Now you can connect a fan to the board and carefully apply power by setting the trimmer to the minimum position (VT1 base is pulled to ground). The fan should not rotate during this. Then, smoothly turning R2, you need to find a moment when the fan starts to rotate slightly at minimum speed and turn the trimmer back just a little bit so that it stops rotating. Now you can check the operation of the regulator - just put your finger on the thermistor and the fan will start rotating again. Thus, when the radiator temperature is equal to room temperature, the fan does not spin, but as soon as it rises a little, it will immediately begin to cool.The speed of a modern computer is achieved at a fairly high price - a power supply unit, processor, video card often need intensive cooling. Specialized cooling systems are expensive, so a home computer is usually equipped with multiple case fans and coolers (radiators with fans attached to them).
The result is an efficient and inexpensive, but often noisy cooling system. To reduce the noise level (provided that efficiency is maintained), a fan speed control system is needed. All kinds of exotic cooling systems will not be considered. The most common air cooling systems should be considered.
In order to reduce the noise during operation of the fans without reducing the cooling efficiency, it is advisable to adhere to the following principles:
- Large diameter fans work more efficiently than small ones.
- The maximum cooling efficiency is observed in coolers with heat pipes.
- Four-pin fans are preferred over three-pin fans.
There can be only two main reasons for excessive fan noise:
- Poor bearing lubrication. Eliminated by cleaning and new grease.
- The engine is turning too fast. If it is possible to reduce this speed while maintaining an acceptable level of cooling intensity, then this should be done. The most affordable and cheapest ways to control the rotation speed are discussed below.
Fan speed control methods
Back to the table of contents
Method one: switching the BIOS to a function that regulates the operation of the fans
The functions Q-Fan control, Smart fan control, etc., supported by some of the motherboards, increase the fan speed when the load increases and decreases when it falls. It is necessary to pay attention to the way of such control of the fan speed using the example of Q-Fan control. You must follow the sequence of actions:
- Enter BIOS. Most often, for this you need to press the "Delete" key before booting the computer. If before loading at the bottom of the screen instead of the message "Press Del to enter Setup" there is a proposal to press another key, do it.
- Open the "Power" section.
- Go to the line "Hardware Monitor".
- Change to "Enabled" the value of the CPU Q-Fan control and Chassis Q-Fan Control functions on the right side of the screen.
- In the appeared lines CPU and Chassis Fan Profile, select one of three performance levels: enhanced (Perfomans), quiet (Silent) and optimal (Optimal).
- Press the F10 key to save the selected setting.
Back to the table of contents
In the foundation.
Features.
Axonometric ventilation diagram.Method two: fan speed control by switching method
Figure 1. Distribution of voltages across contacts.
For most fans, the nominal voltage is 12 V. When this voltage is reduced, the number of revolutions per unit of time decreases - the fan rotates more slowly and makes less noise. You can take advantage of this circumstance by switching the fan to several voltage ratings using an ordinary Molex connector.
The distribution of voltages on the contacts of this connector is shown in Fig. 1a. It turns out that three different voltage values \u200b\u200bcan be removed from it: 5 V, 7 V and 12 V.
To provide this method of changing the fan speed, you need:
- Open the case of a de-energized computer and remove the fan connector from its socket. The wires to the fan of the power supply are easier to remove from the board or just have a bite.
- Using a needle or awl, release the corresponding legs (most often the red wire is a plus, and the black wire is a minus) from the connector.
- Connect the fan wires to the contacts of the Molex connector to the required voltage (see Fig. 1b).
An engine with a nominal speed of rotation of 2000 rpm at a voltage of 7 V will give 1300 per minute, with a voltage of 5 V - 900 rpm. An engine rated at 3500 rpm - 2200 and 1600 rpm, respectively.
Figure 2. Diagram of serial connection of two identical fans.
A special case of this method is daisy chain connection of two identical fans with three-pin connectors. They each account for half the operating voltage, and both spin slower and produce less noise.
A diagram of such a connection is shown in Fig. 2. The left fan connector is connected to the motherboard as usual.
A jumper is installed on the right connector, which is fixed with electrical tape or tape.
Back to the table of contents
The third way: adjusting the fan speed by changing the supply current
To limit the fan speed, constant or variable resistors can be connected in series in its power supply circuit. The latter also allow you to smoothly change the rotation speed. When choosing such a design, one should not forget about its disadvantages:
- Resistors heat up, wasting electricity and contributing to the heating process of the entire structure.
- The characteristics of the electric motor in different modes can be very different, each of them requires resistors with different parameters.
- The power dissipation of the resistors must be large enough.
Figure 3. Electronic circuit for speed control.
It is more rational to use an electronic speed control circuit. Its simple version is shown in Fig. 3. This circuit is a stabilizer with adjustable output voltage. A voltage of 12 V is supplied to the input of the DA1 (KR142EN5A) microcircuit. A signal from its output is supplied to the 8-amplified output by the transistor VT1. The level of this signal can be regulated by the variable resistor R2. It is better to use a trimmer resistor as R1.
If the load current is not more than 0.2 A (one fan), the KR142EN5A microcircuit can be used without a heat sink. If it is present, the output current can reach 3 A. It is advisable to include a ceramic capacitor of small capacity at the input of the circuit.
Back to the table of contents
The fourth way: adjusting the fan speed using the reobass
Reobass is an electronic device that allows you to smoothly change the voltage supplied to the fans.
As a result, the speed of their rotation changes smoothly. The easiest way is to purchase a ready-made reobass. Usually fits into a 5.25 ”bay. There is, perhaps, only one drawback: the device is expensive.
The devices described in the previous section are actually reobases that allow only manual control. In addition, if a resistor is used as a regulator, the motor may not start because the current is limited at the moment of starting. Ideally, a full-fledged reobass should provide:
- Uninterrupted engine start.
- Control of the rotor speed not only in manual but also in automatic mode. With an increase in the temperature of the cooled device, the rotation speed should increase and vice versa.
A relatively simple diagram corresponding to these conditions is shown in Fig. 4. Having the appropriate skills, it is possible to make it with your own hands.
The fan supply voltage is changed in a pulse mode. Switching is carried out using powerful field-effect transistors, the channel resistance of which in the open state is close to zero. Therefore, starting the engines is easy. The highest speed will not be limited either.
The proposed scheme works as follows: at the initial moment, the cooler that cools the processor operates at the minimum speed, and when heated to a certain maximum allowable temperature, it switches to the maximum cooling mode. When the processor temperature drops, the reobass again switches the cooler to the minimum speed. The rest of the fans maintain the manually set mode.
Figure 4. Scheme of adjustment using the reobass.
The basis of the node that controls the operation of computer fans is the DA3 integral timer and the VT3 field-effect transistor. A pulse generator with a pulse repetition rate of 10-15 Hz is assembled on the basis of the timer. The duty cycle of these pulses can be changed using the R5 trimmer, which is part of the R5-C2 RC timing chain. Thanks to this, it is possible to smoothly change the rotation speed of the fans while maintaining the required current value at the time of start.
Capacitor C6 smooths the impulses so that the rotors of the motors rotate softer and without clicking. These fans are connected to the XP2 output.
The basis of a similar control unit for the processor cooler is the DA2 microcircuit and the VT2 field-effect transistor. The only difference is that when a voltage appears at the output of the operational amplifier DA1, thanks to the diodes VD5 and VD6, it is superimposed on the output voltage of the DA2 timer. As a result, VT2 opens completely and the cooler fan starts rotating as quickly as possible.
This is my first post, in the following I will talk about how to make video surveillance, a liquid cooling system, automated (programmable) lighting and much more tasty, we will solder, drill and flash the chips, but for now let's start with the simplest, but nevertheless , very effective reception: installation of a variable resistor.
The noise from the cooler depends on the number of revolutions, the shape of the blades, the type of bearings, etc. The higher the number of revolutions, the more efficient the cooling, and the more noise. 1600 rpm is not always and not everywhere needed. and if we lower them, then the temperature will rise by several degrees, which is not critical, and the noise may disappear altogether!
On modern motherboards, the control of the speed of the coolers that are powered by it is integrated. In BIOS, you can set a "reasonable" cut, which will change the speed of coolers depending on the temperature of the cooled chipset. But there is no such option on old and budget boards, and what about other coolers, for example, a power supply or case cooler? To do this, you can mount a variable resistor in the power supply circuit of the cooler, such systems are sold, but they cost incredible money, given that the cost of such a system is about 1.5 - 2 dollars! This system sells for $ 40:
You can make it yourself, using as a socket - a plug from your system unit (a plug in the basket where DVD / CD drives are inserted), and about other things you will learn from this post.
Because I broke off 1 blade from a cooler on a power supply unit, I bought a new one on ball bearings, it is much quieter than usual:
Now you need to find a wire with power, in the gap of which we mount a resistor. This cooler has 3 wires: black (GND), red (+ 12V) and yellow (tachometer contact).
We cut the red, clean and tinker.
Now we need a variable resistor with a resistance of 100 - 300 Ohm and with a power of 2-5 W... My cooler is rated for 0.18A and 1.7W. If the resistor is designed for less power than the power in the circuit, then it will heat up and eventually burn out. As exdeniz suggests, for our purposes the PPB-3A 3W 220 Ohm... Such as I have a variable resistor, 3 contacts. I will not go into details, just solder 1 wire to the middle contact and one extreme, and the second to the remaining extreme (You can find out the details with a multimeter / ohmmeter. Thanks guessss_who for the comment).
Now we mount the fan into the case and find a suitable place for attaching the resistor.
I decided to insert it like this:
The resistor has a nut for fastening to the plane. Please note that the case is metal and can close the contacts of the resistor and it will not work, so cut out an insulating gasket from plastic or cardboard. My contacts do not close, fortunately, so there are no gaskets in the photo.
Now the most important thing is the field test.
I turned on the system, opened the power supply case and found the hottest part with a pyrometer (this is an element, it looks like a transistor, which is cooled by a radiator). Then I closed it, unscrewed the resistor to maximum speed and waited 20-30 minutes ... The element has heated up to 26.3 ° C.
Then I set the resistor to half, the noise is no longer heard waited again 30 minutes ... The element has heated up to 26.7 ° C.
Again I lower the speed to a minimum (~ 100 Ohm), wait 30 minutes, I don't hear any noise from the cooler at all ... The element has heated up to 28.1 ° C.
I do not know what kind of element it is and what its operating temperature is, but I think that it will withstand another 5-10 degrees. But if we take into account that there was no noise on the "half" of the resistor, then we don't need anything else! \u003d)
Now you can make such a panel as I gave at the beginning of the article and it will cost you a penny.
Thank you.
UPD: Thanks to the gentlemen from the comments, for the reminder about watts.
UPD: If you are interested in the topic and you know what a soldering iron is, then you can easily assemble an analog reobass. As fleshy tells us, the article Analog Reobass describes this wonderful device. Even if you have never soldered boards, you can still assemble a reobass. There is a lot of text in the article, which I do not understand, but the main thing: Composition, Scheme, Motage ( this paragraph has links to all necessary articles on soldering).
First, the thermostat. When choosing a circuit, factors such as its simplicity, the availability of elements (radio components) necessary for the assembly, especially those used as temperature sensors, the manufacturability of assembly and installation in the PSU case, were taken into account.
According to these criteria, the most successful, in our opinion, turned out to be V. Portunov's scheme. It reduces wear on the fan and reduces the noise generated by it. The diagram of this automatic fan speed controller is shown in Fig. 1. The temperature sensor is the VD1-VD4 diodes, connected in the opposite direction to the base circuit of the composite transistor VT1, VT2. The choice of diodes as a sensor determined the dependence of their reverse current on temperature, which is more pronounced than the analogous dependence of the resistance of thermistors. In addition, the glass case of these diodes makes it possible to do without any dielectric spacers when installing the power supply transistors on the heat sink. An important role was played by the prevalence of diodes and their availability for radio amateurs.
Resistor R1 excludes the possibility of failure of the VTI, VT2 transistors in the event of thermal breakdown of the diodes (for example, when the fan motor is jammed). Its resistance is selected based on the maximum allowable base current value VT1. Resistor R2 determines the threshold for the regulator.
Fig. 1
It should be noted that the number of temperature sensor diodes depends on the static current transfer ratio of the composite transistor VT1, VT2. If the fan impeller is stationary at the indicated resistance R2, room temperature and power on, the number of diodes should be increased. It is necessary to ensure that after the supply voltage is applied, it confidently begins to rotate at a low frequency. Naturally, if the rotational speed is too high with four sensor diodes, the number of diodes should be reduced.
The device is mounted in the power supply housing. The same terminals of the VD1-VD4 diodes are soldered together, placing their cases in the same plane close to each other.The resulting block is glued with BF-2 glue (or any other heat-resistant, for example, epoxy) to the heat sink of high-voltage transistors on the reverse side. Transistor VT2 with resistors R1, R2 and transistor VT1 soldered to its terminals (Fig. 2) is installed with the emitter lead into the "+12 V fan" hole of the power supply board (earlier the red wire from the fan was connected there). The adjustment of the device is reduced to the selection of the resistor R2 after 2 .. 3 minutes after turning on the PC and warming up the power supply transistors. Temporarily replacing R2 with a variable (100-150 kOhm), select such a resistance so that at rated load the heat sinks of the power supply transistors heat up to no more than 40 ºС.
To avoid electric shock (the heatsinks carry high voltages!), You can only "measure" the temperature by touch by turning off the computer.
A simple and reliable scheme was proposed by I. Lavrushov (UA6HJQ). The principle of its operation is the same as in the previous circuit, however, an NTC thermistor is used as a temperature sensor (the nominal value of 10 kΩ is not critical). The transistor in the circuit is selected as KT503. As determined empirically, its operation is more stable than other types of transistors. It is advisable to use a multiturn trimmer resistor, which will allow you to more accurately adjust the temperature threshold of the transistor operation and, accordingly, the fan speed. The thermistor is glued to the 12V diode assembly. If not, it can be replaced with two diodes. More powerful fans with a consumption current of more than 100 mA should be connected through a composite transistor circuit (the second KT815 transistor).
Fig. 3
The diagrams of the other two, relatively simple and inexpensive regulators of the rotational speed of the PSU cooling fans, are often given on the Internet (CQHAM.ru). Their feature is that the TL431 integral stabilizer is used as a threshold element. It is quite easy to "get" this microcircuit when disassembling the old power supply units of the ATX PC.
The author of the first scheme (Fig. 4) Ivan Shor (RA3WDK). With the repetition, the expediency of using a multi-turn of the same rating as a trimmer resistor R1 was revealed. The thermistor is attached to the radiator of the cooled diode assembly (or to its case) through KPT-80 thermal grease.
Fig. 4
A similar scheme, but on two KT503 connected in parallel (instead of one KT815) was used by Alexander (RX3DUR). At the values \u200b\u200bof the parts indicated in the diagram (Fig. 5), 7V is supplied to the fan, increasing when the thermistor heats up. KT503 transistors can be replaced with imported 2SC945, all resistors with a power of 0.25W.
A more complex circuit of the cooling fan speed regulator is described in. For a long time it has been successfully used in another power supply unit. In contrast to the prototype, it uses "television" transistors. I will refer readers to the article on our website "Another universal power supply unit" and the archive, which presents a variant of the printed circuit board (Fig. 5 in the archive) and a journal source. The role of the radiator of the regulated transistor T2 on it is played by the free foil section left on the front side of the board. This scheme allows, in addition to automatically increasing the fan speed when the radiator of cooled transistors of a power supply unit or diode assembly heats up, set the minimum threshold speed manually, up to the maximum.
Fig. 6
Proportional control is the key to silence!
What is the challenge for our management system? Yes, so that the propellers do not rotate in vain, so that the dependence of the rotation speed on the temperature. The hotter the device, the faster the fan rotates. Is it logical? It is logical! We will decide on that.
Of course, it is possible to bother with microcontrollers, in which it will be even easier, but absolutely not necessary. In my opinion, it is easier to make an analog control system - there will be no need to bother with programming in assembly language.
It will be both cheaper and easier to set up and configure, and most importantly anyone can expand and build on the system to their liking by adding channels and sensors. All you need is just a few resistors, one microcircuit and a thermal sensor. Well, also straight arms and some soldering skill.
Shawl top view
Bottom view
Structure:
- Chip resistors size 1206. Well, or just buy in a store - the average price of one resistor is 30 kopecks. In the end, no one bothers you to slightly tweak the board, so that in place of the chip resistors you can solder ordinary ones, with legs, and they are in bulk in any old transistor TV.
- Multi-turn variable resistor approx. 15kΩ.
- You will also need a chip capacitor of size 1206 at 470nf (0.47uF)
- Any electrolytic capacitor with a voltage of 16 volts and above and a capacity in the region of 10-100 μF.
- Screw terminal blocks at will - you can just solder the wires to the board, but I put the terminal block, purely for aesthetic reasons - the device should look solid.
- We will use a powerful MOSFET transistor as a power element that will control the power supply of the cooler. For example IRF630 or IRF530 it can sometimes be ripped out of old power supplies from a computer. Of course, for a tiny propeller, its power is excessive, but you never know, what if you want to put something more powerful there?
- We will feel the temperature with a precision sensor LM335Z, it costs no more than ten rubles and does not represent a deficit, and you can replace it with some kind of thermistor on occasion, since it is also not uncommon.
- The main part on which everything is based is the microcircuit, which is four operational amplifiers in one package - the LM324N is a very popular piece. It has a bunch of analogues (LM124N, LM224N, 1401UD2A), the main thing is to make sure that it is in a DIP package (so long, with fourteen legs, as in the pictures).
Wonderful mode - PWM
PWM signal generation
To make the fan rotate more slowly, it is enough to reduce its voltage. In the simplest reobasses, this is done by means of a variable resistor, which is placed in series with the motor. As a result, part of the voltage will drop across the resistor, and less will fall on the motor as a result - a decrease in speed. Where is the bastard, don't you notice? Yes, the ambush is that the energy released on the resistor is converted not into anything, but into ordinary heat. Do you need a heater inside your computer? Obviously not! Therefore, we will go in a more cunning way - we will apply pulse width modulation aka PWMor PWM... Sounds scary, but don't be afraid, everything is simple here. Imagine that the engine is a massive cart. You can push it with your foot continuously, which is tantamount to direct inclusion. And you can move the kicks - this will be PWM... The longer the kick, the more you accelerate the cart.
When PWM power supply to the engine is not a constant voltage, but rectangular pulses, as if you turn the power on and off, only quickly, dozens of times per second. But the motor has not weak inertia, and also the inductance of the windings, so these pulses seem to be summed up with each other - they are integrated. Those. the larger the total area under the pulses per unit of time, the greater the equivalent voltage goes to the motor. You feed narrow, like needles, impulses - the engine barely rotates, and if you feed wide, practically without gaps, then this is tantamount to direct inclusion. Turning the engine on and off will be ours MOSFET transistor, and the circuit will form pulses.
Saw + Straight \u003d?
Such a clever control signal is easy to obtain. To do this, we need to comparator drive the signal sawtooth forms and compare him with any permanent tension. Look at the picture. Let's say our saw goes to negative output comparator, and constant voltage to positive. The comparator adds these two signals, determines which of them is greater, and then issues a verdict: if the voltage at the negative input is greater than the positive, then the output will be zero volts, and if the positive is greater than negative, then the output will be supply voltage, that is about 12 volts. The saw runs continuously, it does not change its shape over time, such a signal is called a reference signal.
But the constant voltage can move up or down, increasing or decreasing depending on the temperature of the sensor. The higher the temperature of the sensor, the more voltage comes out of it., which means that the pressure at the constant input becomes higher and, accordingly, the pulses at the output of the comparator become wider, forcing the fan to spin faster. This will continue until the DC voltage blocks the saw, causing the engine to run at full speed. If the temperature is low, then the voltage at the sensor output is low and the constant will go below the lowest saw tooth, which will cause any impulses to stop at all and the engine will stop altogether. Loaded, right? ;) Nothing, it is useful for brains to work.
Temperature mathematics
Regulation
As a sensor we use LM335Z... Essentially it is thermostabilitron... The trick of the zener diode is that a strictly defined voltage drops out on it, like on a limiting valve. Well, for a thermostabilitron, this voltage depends on the temperature. Have LM335th dependence looks like 10mV * 1 degree Kelvin... Those. counting is carried out from absolute zero. Zero Celsius is equal to two hundred seventy-three degrees Kelvin. So, in order to get the voltage output from the sensor, say at plus twenty-five degrees Celsius, we need to add two hundred seventy-three to twenty-five and multiply the resulting sum by ten millivolts.
(25 + 273) * 0.01 \u003d 2.98V
At other temperatures, the voltage will not change much, for the same 10 millivolts per degree... This is another setup:
The voltage from the sensor does not change much, by some tenths of a volt, but it must be compared with a saw in which the height of the teeth reaches as much as ten volts. To get a constant component directly from the sensor for such a voltage, you need to heat it up to a thousand degrees - a rare mess. How then to be?
Since our temperature is still unlikely to drop below twenty-five degrees, everything below us is not of interest, which means that we can select only the very top from the output voltage from the sensor, where all the changes take place. How? Yes, just subtract two whole ninety-eight hundredths of a volt from the output signal. And multiply the remaining crumbs by gainlet's say thirty.
We will get exactly about 10 volts at fifty degrees, and down to zero at lower temperatures. Thus, we get a kind of temperature "window" from twenty-five to fifty degrees, within which the regulator operates. Below twenty-five - the engine is off, above fifty - directly on. Between these values, the fan speed is proportional to the temperature. The window width depends on the gain. The larger it is, the narrower the window. the limit 10 volts, after which the constant component on the comparator will be higher than the saw and the motor will turn on directly, will come earlier.
But after all, we do not use either a microcontroller or computer tools, how are we going to do all these calculations? And the same operational amplifier. It is not for nothing that it is called operational, its original purpose is mathematical operations. All analog computers are built on them - awesome machines, by the way.
To subtract one voltage from the other, you need to feed them to different inputs of the operational amplifier. We supply voltage from the temperature sensor to positive input, and the voltage to be subtracted, the bias voltage, is applied to negative... It turns out the subtraction of one from the other, and the result is also multiplied by a huge number, almost to infinity, and another comparator is obtained.
But we do not need infinity, since in this case our temperature window narrows to a point on the temperature scale and we have either a standing or a madly rotating fan, and there is nothing more annoying than an on and off compressor of a shovel refrigerator. We also do not need an analogue of a refrigerator in a computer. Therefore, we will lower the gain by adding to our subtractor feedbacks.
The essence of the feedback is to drive the signal from the output back to the input. If the voltage from the output is subtracted from the input, then this is negative feedback, and if it is added, then it is positive. Positive feedback increases the gain, but can lead to signal generation (submachine gunners call this a loss of system stability). A good example of positive feedback with loss of stability is when you turn on the microphone and poke it into the speaker, usually immediately there is a nasty howl or whistle - this is the generation. We need to reduce the gain of our opamp to a reasonable level, so we will apply negative coupling and bring the signal from the output to the negative input.
The ratio of the feedback and input resistors will give us the gain that affects the width of the regulation window. I figured thirty would be enough, you can count it to fit your needs.
Saw
It remains to make a saw, or rather to assemble a sawtooth voltage generator. It will consist of two opamp. The first, due to the positive feedback, turns out to be in the generator mode, giving out rectangular pulses, and the second serves as an integrator, turning these rectangles into a sawtooth shape.
The capacitor in the feedback of the second op-amp determines the pulse frequency. The lower the capacitance of the capacitor, the higher the frequency and vice versa. Generally in PWM generation the more the better. But there is one jamb, if the frequency falls into the audible range (20 to 20,000 Hz), then the engine will squeak disgustingly at a frequency PWMwhich is clearly at odds with our concept of a silent computer.
And I could not get more than fifteen kilohertz from this circuit - it sounded disgusting. I had to go the other way and drive the frequency to the lower range, to the region of twenty hertz. The engine began to vibrate slightly, but it is not audible and is felt only with your fingers.
Scheme.
Dachshund, we figured out the blocks, it's time to look at the schematic. I think most have already guessed what's what. And I'll explain it anyway, for clarity. The dotted line on the diagram indicates functional blocks.
Block # 1
This is a saw generator. Resistors R1 and R2 form a voltage divider to supply half of the supply to the generator, in principle they can be of any value, the main thing is that they are the same and not very high resistance, within a hundred kilo-ohms. Resistor R3 paired with capacitor C1 determine the frequency, the lower their values, the higher the frequency, but again I repeat that I could not bring the circuit out of the audio range, so it’s better to leave it as it is. R4 and R5 are positive feedback resistors. They also affect the height of the saw from zero. In this case, the parameters are optimal, but if you do not find the same, then you can take about plus or minus a kilo. The main thing is to maintain a proportion between their resistances of about 1: 2. If R4 is greatly reduced, then R5 will have to be reduced.
Block # 2
This is a comparison unit, here PWM pulses are formed from the saw and DC voltage.
Block # 3
This is just the circuit that suits the calculation of temperature. Thermal sensor voltage VD1 applied to the positive input, and the negative input is supplied with a bias voltage from the divider to R7... Rotating the trimmer knob R7 you can move the regulation window higher or lower on the temperature scale.
Resistor R8 it can be within 5-10 kOhm more is undesirable, less too - the temperature sensor may burn out. Resistors R10 and R11 must be equal to each other. Resistors R9 and R12 must also be equal to each other. Resistors R9 and R10 can be, in principle, any, but it must be taken into account that the gain, which determines the width of the regulation window, depends on their ratio. Ku \u003d R9 / R10 based on this ratio, you can choose denominations, the main thing is that it is not less than a kilo-ohm. The optimal, in my opinion, is a coefficient equal to 30, which is provided by 1kΩ and 30kΩ resistors.
Mounting
Printed circuit board
The device is made with printed wiring to be as compact and accurate as possible. The drawing of the printed circuit board in the form of a Layout file is posted right there on the site, the program Sprint Layout 5.1 to view and simulate printed circuit boards can be downloaded from here
The very same printed circuit board is made once or twice by means of laser-ironing technology.
When all the parts are assembled, and the board is etched, then you can start assembling. Resistors and capacitors can be soldered without fear, because they are almost not afraid of overheating. Special care should be taken with MOSFET transistor.
The fact is that he is afraid of static electricity. Therefore, before you take it out of the foil in which you should wrap it in the store, I recommend taking off your synthetic clothes and touching the bare battery or faucet in the kitchen with your hand. Mikruhu can be overheated, so when you solder it, do not hold the soldering iron on the legs for more than a couple of seconds. Well, and finally, I will give advice on resistors, or rather on their marking. Do you see the numbers on its back? So this is the resistance in ohms, and the last digit indicates the number of zeros after. for example 103
this 10
and 000
i.e 10 000
Ohm or 10kOhm.
Upgrading is a delicate matter.
If, for example, you want to add a second sensor to control another fan, then it is absolutely not necessary to fence the second generator, it is enough to add a second comparator and a calculation circuit, and feed the saw from the same source. To do this, of course, you will have to redraw the circuit board drawing, but I do not think that this will be a big deal for you.
