RadioLocman.com Electronics ru
Advanced Search +
  

04-28-2008

Build Yourself a Fan Temperature Control

Tillmann Steinbrecher

CPU coolers have been getting more and more efficient over the last years. However, this higher performance usually has a price: More noise. Noise is annoing and reduces well-being and productivity - so why not do something against it?

The solution: A temperature control
In many PCs, the fans are always spinning at max speed - no matter if the CPU is running at full load or is idle, and no matter if the outside temperature is 30°C or 16°C. Of course, this doesn't make sense. Some modern CPU coolers feature temperature controls; many motherboards also have circuitry for controlling fan speed.

Problems with conventional temperature controls
Temperature controls for fans aren't exactly a new idea. Temperature-controlled fans are widely available. However, the most commonly used fan temperature controls have some major limitations:

  • The temperature at which the CPU (or case) should be kept cannot be adjusted by the user. This is a major problem: The maximum operating temperature from varies greatly between CPU types, so it's unlikely that the control will provide the best possible regulation for your particular CPU. Also, coolers with a temperature control that can't be adjusted by the user are totally unsuitable for overclocking, since an overclocked CPU must usually be kept below a very precise temperature, at which it crashes.

    Most conventional temperature controls only regulate fan speed, but don't turn the fan off. Especially for case fans, this doesn't make sense. It is much better if the fan is totally shut off when it's not required. Certain types of CPUs will produce so little heat when idle, that the CPU fan can be turned off when the CPU is not busy.
  • Each fan needs its own sensor - with must current fan temperature controls, each fan needs its own sensor; it is not possible to control more than one fan with a single sensor.

Instead of continueing to whine about current solutions, we present a better one. For about $4, you can build yourself a temperature control with the following features:

  • Temperature can be adjusted by the user. The desired temperature can be adjusted over a wide range, so the control is suitable both for regulating case temperature and CPU temperature.
  • Fans are switched off if the temperature is low enough
  • Several fans can be controlled with just one sensor and temperature control

Building the unit

With all these features, you might think that this temperature control will be hard to build and requires lots of expensive parts. Wrong guess. It consists of just three electronic parts!

  • A MOSFET Power transistor (N-Channel), price between $1 and $2
  • A 10K spindle trimming potentiometer, price around $1
  • A 10K NTC temperature sensor, price around $1

All of these are standard, easy-to-get parts.

The MOSFET can be just about any N-channel Power MOSFET, as long as it can handle the 12 V voltage and the amperes the fan requires. Even the cheapest Power MOSFETs in the $1 area can usally handle over 50 V and over 10 amperes, so they will be by far sufficient for this circuit. We used an IRFZ24N MOSFET. If you're in the US, you can use an IFR510 Power MOSFET. The pinout of Power MOSFETs is standardized, as you can see on the image. It is very unlikely that the MOSFET you buy has a different pinout, unless you get a really exotic one.

The spindle trimming potentiometer should be a 10K type. If you cannot get a 10 K unit, 20 K or 25 K will also do. However, this pot is rather large and bulky, a spindle trimming potentiometer is really the more elegant (and also cheaper) solution. Finally, for the NTC thermistor, get the cheapest 10K NTC you can get. It doesn't matter if it has a high tolerance. You can also use the flat thermistors for CPU monitoring that ship with some motherboards.

Build Yourself a Fan Temperature Control

Apart from the electronic parts, you should also get:

  • Heat-contractable tubing

    Since the unit is so small, you don't need to put it in a special case. However, it is absolutely required that the entire unit is insulated. Otherwise, it might cause a short circuit when touching your case (see warning below). The easiest and cheapest solution is to simply put all the circuitry in a (large enough) heat-contractable tube. If you have trouble getting heat-contractable tubings that are large enough, you can also wrap up the circuit with good old electrical tape. Also, the NTC's contacts must be insulated (see photo). For this, you need smaller heat-contractable tubing. Since this tubing also provides som thermal insulation, the wisest thing to do is to apply it in a way that the head of the NTC sticks out, but the contacts are insulated - see photo.
  • A little PCB

    Since we're dealing only with three components, it's not strictly necessary to put them on a PCB. However, it is more elegant to use a PCB, and it makes things easier.
  • Possibly a heatsink for the MOSFET

    Under normal conditions, it is not necessary to put a heatsink on the MOSFET. However, if you'd like to use the temperature control for an unusually high load (more than three fans, or fans with particularly high power usage), then you should add a little heatsink to the MOSFET. Again, the heatsink must be insulated. Depending on where you buy the parts, the total price of all parts of the temperature control will be between $3 and $6. Feel free to use the Heatsink Guide Discussion Board for help, suggestions, tips for places to buy, and "coshopping".

Apart from the electronic parts, you should also get

Assembling it
I won't be bothering you with complicated electronic symbols - instead, I've drawn all the components the way they really look, to make it as easy as possible for you. This way, you can build the circuit without any knowledge about electronics - as long as you know how to use a soldering iron. Note that the NTC is shown larger than it is in real life.

schematics Build Yourself a Fan Temperature Control

When it's done in an elegant way, the entire circuit can be as small as you can see on these photos:

When it's done in an elegant way, the entire circuit can be as small as you can see on these photos

When it's done in an elegant way, the entire circuit can be as small as you can see on these photos

However, if you're not very experienced with soldering, I'd recommend to build the thing on a larger PCB. Before you start testing your device, read the following warning:

WARNING! Whenever you're building an electric or electronic device, good insulation is very important for security. This temperature control isn't an exception. All parts of the temperature control, including the sensor and the MOSFET transistor (even it's metal backplate!) carry a voltage of up to 12 V. Therefore, it is absolutely required that the circuit is properly insulated (e.g. using heat-contractable tubing or electrical tape). If the unit is not properly insulated, a short circuit will occur if the circuit or the sensor touches the case or (even worse) the motherboard.

Fan RPM monitoring will NOT work with this temperature control. Don't try to connect the rpm signal wire to your motherboard. It will not work and might damage your motherboard. Building a temperature control that works with rpm monitoring would require the usage of a P-Channel MOSFET (and a different circuitry), which are not as widely available as N-Channel Power MOSFETs.

Here's the finished unit with proper insulation. The potentiometer's screw sticks out on the lower end.

Here's the finished unit with proper insulation. The potentiometer's screw sticks out on the lower end.

Setting up the temperature control

Before you start using the temperature control, the temperature at which the CPU (or case, etc...) should be kept must be set. This is done by adjusting the spindle trimming potentiometer. A good strategy to do this is: Start with a cool CPU (or case). Turn potentiometer until the fan doesn't spin at all. Then, watch the temperature rise. As soon as the temperature gets close to the maximum it should reach, adjust the potentiometer so that the fan just starts to spin.

Exact adjustment is crucial if you'd like your temperature control to work the way it should. Do invest enough time adjusting it precisely. If the adjustment is not good, the computer will either overheat, or the fans will spin almost at top speed all the time, making the control useless.

You might notice that when turning your computer on, the controlled fans will not spin at all. After a while, as the system gets hotter, they will start to spin; and often they will spin a little faster at first, but then fall back to slower speed after a while. This is normal; the reason is that fans need a higher voltage to start spinning than to keep spinning.

If you change the fan connected to the temperature control, or if you add another fan, the temperature must be readjusted.

Legal stuff

If you build this temperature control, you do it at your OWN RISK. The instructions presented here come with NO WARRANTY, neither expressed nor implied.

You may have to register before you can post comments and get full access to forum.
User Name
Fragments of discussion:Full version of discussion »
  • I am a total noob with electronics - can anyone explain to me why do we need a MOSFET in this circuit. Can we not use only the potentiometer in this circuit with same effect.
  • Hi, this mosfet is acting here as a controlling gate. once it reaches the set point from potentio meter & from NTC simply the gate allows the current flow.. :)
  • Hi i did this circuit.... its awesome thanks for the circuit..... i want to add another function for this. can u pls tell me how to add an over ride signal to drive the fan directly at any time without damaging the circuit??
  • Hi can you please tell me how to add an + positive signal to override the circuit and run the fans directly when i needed.. thanks in advance..
  • Sonda, yes, I can tell you how. I had the thought to do the same thing, then saw your question. A fan override function is as simple as adding a switch. You need only a SPST toggle switch to connect the "ground" of the fan. See how the positive 12v is always connected to the fan? Connecting the ground completes the circuit and turns the fan on. The MOSFET does this when temperature is reached. You will add a 2nd switch that will allow the fan to turn on either way, either by Mosfet/Thermistor automation, or human hand flipping a switch. Simple instructions: Connect one lug of the SPST toggle switch to where Tillmann has has designated "To ground wire from power supply." Connect the other toggle lug to "To fan's black (-) wire." Voila! Let me know if you have any troubles.
  • Hi there! I really appreciate this document. Very useful and exactly what I was looking for. I am a rc helicopter enthusiast and I recently adapter a dell server power supply to power a new charger I purchased. The problem I was having was that the fan had no temp control and just ran full blast. I did build the circuit as you specified and it was perfect, until the PSU stopped working because the built in speed tac noticed the fan wasn't spinning. What would be the best way to build this circuit to allow for a low end adjustment of the fan speed independent of the thermister? IE....an additional pot somewhere in the circuit. See attached. Thanks!
  • Calmeida, As I adjusted the 10k trimpot, the fan slowed and slowed as I continued to adjust until I moved past that sweet spot where the fan stopped altogether - presumably to reactivate at high temperatures. You could try adjusting your trimpot for just before this spot and let your fan slowly spin at all times perhaps. Then it would ramp up at higher temperatures. (I'm using a 25-turn trimpot btw. A single turn trim would be hard to dial in just right.) Dunno if that's enough of a draw to keep the PSU sensor happy. You may have to trick it with a dummy load so it thinks the fan is always spinning.
  • Kato, Thanks for the response. I've tried a number of way to adjust the pot but if it's too cold than the fan wont come on at all and the PSU will not work. If I adjust the fan at the coldest point in the day then by later in the afternoon the fan is almost going full blast, which is unnecessary and loud. I am using a 25 turn POT as well. I think the best bet would be to somehow set a baseline speed where the fan cannot go any slower. Believe me, I have adjusted this thing quite a bit to find a happy medium and either the next time I turn it on the fan doesn't come on or it comes on full blast. Thanks, Chris
  • This controller is exactly what I need to run a fan in my cctv dvr. However, space is limited and I can only accommodate a 5V laptop fan. How do I adapt this circuit to run from 5V supply instead of 12V? (HDD supply has 12v and 5v, but I don't want the fan to run continuously.) Guidance appreciated.
  • Chris, it sounds like your PSU already has some kind of temperature sensor built in which is clashing with the temperature sensor in this circuit. Am I reading that right? I have two different models of Seasonic PSUs on which the fan doesn't spin at all until it reaches a certain temperature. The circuit on this page is presuming a full on fan at all times. But I'm not sure what to suggest other than to bypass that circuitry. Does your psu use a PWM style speed-controlled fan with the 4 pin connector? Maybe you can just skip the "sense" and "control" lines and use only the +12 and ground? (I'm just stabbing in the dark here.)
  • Scooby, I'm pretty sure you can just use the circuit as drawn with any low-voltage under 50v. It seems voltage agnostic. Somebody let me know if i'm missing something obvious.
  • Kato, I had hoped so, but we found that it wouldn't trim to an off condition - fan just ran continuously irrespective of pot position or temperature. I may re-make it as we did try to make it compact.
  • Hmm, I don't know. Hopefully Tillmann will chime in before you rebuild it. If you have a 12v fan handy, I would test the circuit with that just to find out if it's working properly at 12v. Then, worst case scenario, you could calculate a power resistor value to drop down to 5v so as not to fry the fan. Good luck!
  • Hi I would like to use this fan controller for 2 cooling fans in my power supply with 4 heatsink and therefor I need to use 4 NTC 39KOhm or 47KOhm parallel is that possible.
  • nothing is impossible..
Full version of discussion »

News on theme:

Slices ↓
Radiolocman facebook Radiolocman twitter Radiolocman google plus