30 March 2010
Author: Giorgos LazaridisPC Fan Controlling Methods
Computers are well introduced into our lives. As they grow bigger and faster, they need more and more power to operate. Unfortunately, the power efficiency of electronics is still very low. Most of the power that computers draw, is turned into heat losses. Heat is the major enemy of electronics, and all computers needs a way for cooling. During time, many different ways have been applied. You can read some interesting PC cooling methods in this link. Some are cheap, some other expensive, some are efficient, some others have great heat capabilities... Among them, the passive cooling is the most power efficient -as it uses of no extra power!- followed by the active air cooling. Passive cooling needs expensive parts, and most of the times, pure passive cooling is just incapable to dissipate the amount of heat power of a part such as a CPU.
Active air cooling - using fans of course - is by far the most power efficient way of cooling and yet capable to dissipate the heat load of a PC with ease. In this article, i will present the reasons and the methods that these fans can be controlled.
Why should we control the speed of the fans?
A fan controller is an added cost to the cooling of the box, and also it is an added power consumption - a tiny one but still a power consumption. There should be some good reason for someone to add speed control on the fans. Well, as a matter of fact there is:
Noise reduction from the operation of the fan
No matter what, a moving air mass inside a closed place such as the PC box will cause noise. The higher the speed and the volume of the moving air, the more the sound. Most of the times, the volume of the air that is pushed in the box in order to cool it, is more than needed. Either because there are too many fans, or the air is well humidified and cool. Moreover, the fan itself will cause a lot of noise. Fans are usually not dynamically balanced too well. Therefore, they will cause vibrations which they cause the distinctive and annoying noise. Also, the geometry of the fins will cause acoustic waves as they "slap" the air to push it forward. By reducing the speed of the fans when possible, the noise is dramatically decreased. Some people - just like i do - use more fans than needed, but running in low rpm.
Reduced power consumption of the fan
Circuits consume electric power indeed. Chips and transistors needs an amount of power to operate, while other components such as resistors will dissipate heat which means more energy loss. Frankly, modern controllers can be made with low power consumption chips. A PIC for example will consume only 180 uA. A lousy PWM controller made with a 555 timer operating at 5 volts, will draw not more than roughly 30mA. That is 150 mWatts. A 300mA typical fan consumes 3.6 Watts, that is 3600 mWatts. The 150 mWatt that the circuit consumes is nothing, if the fan will consume with half the nominal power. So... yes, the power consumption is significantly decreased.
Increased lifetime of the fan -> Increased reliability
The fan has bearings, and in these bearings friction is generated. This friction will slowly and steady wear the bearings. This will be the end of the fan. By rotating the fan slower, the friction and the heat generated by the friction if reduced, therefore the lifetime and the reliability of the fan is increased.
Reduced dust accumulation
The air carries these small annoying particles called "dust". Filters are used to reduce the dust. They don't seem to do a very good job though, because even with filters, the dust will sneak into the box. The dust causes 2 problems: first of all, it clogs the filter and this will reduce the air that goes in the box. Also, the dust will stick on the surface of chips and heat-sinks, reducing their ability to dissipate heat sufficiently. Therefore, the temperature inside the box is increased. If the incoming air volume is reduced to the required volume for cooling by reducing the speed of the fan, fewer amount of dust will be pushed through the filters and in the box.
How many types of PC fans are there on the market?
To answer this question, someone must first set some criteria. I will distinguish the fan types according to the number of wires that the fan has. Now that i have set my criteria, i can answer this question easily. There are 3 types of PC fans:
Type 1: The 2-wire fans
This is the simplest and nowadays most rarely used fan type. It has only one ability: to rotate and generate air flow. The wires are used to power the brushless motor of the fan.
Type 2: The 3-wire fans
A much improved fan type that provides one more ability: the rpm feedback! These fans have an extra wire that is actually the output of the Hall sensor. This third wire will return the pulses from the sensor. For each turn, the wire will return two pulses with 50% duty cycle. The frequency depends only on the speed that the fan rotates. Therefore, the fan controller can read the speed and can understand if the fan is running too slow or if the fan is stalled. The signal from the 3rd wire is usually referred as "tach signal"
Type 3: The 4-wire fans
This is the latest and usually most expensive fan type. The first 3 wires of the fan have the same functionality as the simple 3-wire fan. The 4th wire is the control wire. These fans are often called "PWM fans". They have a built-in FET, and the control wire is connected to the gate of the FET. These fans are designed to be controlled with PWM pulses. Although all fans can be controlled with PWM controllers, the 4 wire fans, due to the built-in FET and their extra wiring, can have simultaneously rpm feedback from the tach, something that the 3-wire fans cannot do it. This is their advantage.
Need more detailed info about the fans?
I have written an article with details about how PC fans work, You can find it here: How PC Fans Work. You will find there also links for the brushless DC motors, the brushed DC motors and the Hall sensor.
How can someone control the speed of a PC Fan?
PC fans use brushless DC motors to operate. These motors can be easily controlled with a number of ways:
Fixed speed control
This is the easiest way of controlling a fan. A -usually- very simple circuit will power the fan with reduced voltage. A brushless motor, when powered with reduced voltage, will rotate slower. Such a circuit can be as simple as a resistor in series with the positive supply of the fan, or can be more complicated, by using a voltage regulator like an 7806. Using this method, the fan will always have a fixed low speed. The efficiency of the circuit has to do only with the way that the voltage is reduced. Using a Zener for example is much more efficient than using a resistor.Example circuit(s):
Stepped speed control
A variation of the above category is the stepped speed control. Instead of using a simple diode or resistor to drop the voltage for example down to 6 volts, multiple components are connected in series. Each one will cause a smaller voltage drop. A selector switch is used to select different points from the series connection where the power of the fan will be delivered. The more the components that exist before the connection of the fan, the less the voltage that will be delivered to it and thus the less the speed it will rotate. Again the efficiency depends on the type of the component selected for the application. A series of diodes for example are more efficient than a series of resistors.Example circuit(s):
And i do not mean manual on-off control using a switch of course. In this category i have the thermostat circuits. The fan will turn on or off, according to the temperature inside the box. For example, there is always an undervoltaged fan cooling a box. If the temperature in the box exceeds a critical point (for example 40 degrees), a secondary fan will turn on to avoid overheating. Or maybe, the first fan will be powered with full voltage. Usually, the fan connected to the thermostat will run in full speed, thus this system makes noise.Example circuit(s):
Linear speed control
The circuits so far could only set the fan into a standard speed. The linear controllers can set the fan into any speed within a range. The first and most simple way to do this, is with a power transistor. Using a potentiometer, we change the voltage on the base of the transistor. The fan is powered through the transistor. The transistor will dissipate power according to the base voltage. The more the base voltage, the less the power that the pass-component (eg transistor) will dissipate, and thus, the faster the fan will run.
The linear controllers have great advantages over the fixed speed controllers. The speed can be decreased according to the temperature of the box, and thus maintain the noise levels always as low as possible. Yet, the controller i just described carries a great disadvantage: At low speeds, the pass-component will have to dissipate a lot of power and thus it will generate a lot of heat. This means that this system is rather inefficient at low speeds, as far as the power consumption is concerned.Example circuit(s):
Linear DC-DC speed control
This is an improved version of the above method. Instead of using a simple transistor to control the fan, a DC-DC converter is used. Usually, this is done with a switching mode power supply. A PWM generator will send the PWM pulses to the pass component. Large capacitors are connected in parallel to smooth the pulses at the output of the pass-component (eg a MOSFET). The more the duty cycle of the pulses, the higher the output voltage. The output voltage is no longer composed by PWM pulses. The pulses have been smoothed by the large capacitors, therefore it is a smooth direct current. I have written an extensive article about the PWM theory. You can find details about the switching power supplies in this page: PWM theory and switching power supply operation.
The DC-DC converter, like the simple linear controller described previously, will both control the speed of the fan by changing the voltage. The great difference between these two methods is that, the DC-DC converter, instead of dissipating the power on the pass component in order to reduce the power on the fan, it rather do not deliver it at all! The power delivered to the pass component (and finally turned into voltage) is proportional to the duty cycle of the PWM pulses. A 50% duty cycle, will reduce the power delivered to the fan by 50% (ideally). The pass-component will dissipate no power at all, except the amount of power it needs to operate. Therefore, this system is highly efficient.
Ideally, these systems have 100% power efficiency. Yet in the real world, such systems may be from 75 up to 95% power efficient. Their efficiency is increased, as the speed is decreased. This works completely opposite as the simple linear controllers with the power transistors. And this is because the DC-DC converter does NOT dissipate power, instead it does not deliver it at all. The greatest disadvantage that someone could find in these systems is the circuit complexity and the building cost.Example circuit(s):
PWM speed control
This is by far the most efficient method of controlling a fan. Like the above method (the DC-DC converter), the controller will generate PWM pulses to control the speed of the fan. The more the duty cycle, the faster the fan will run. But unlike the previous method, there is no smoothing capacitor after the pass-component. Instead, the pulses are driven as-is directly to the coils of the motor, and the voltage is always fixed to the maximum voltage.
This is the first method that we control the speed of the fan NOT by altering the supply voltage. This gives a major advantage over the other methods: The fan can start and rotate at extremely low speeds! Unlike all other methods, this method will always generate maximum torque on the shaft of the motor, as the voltage is always maximum. This also prevents the fan from stalling when running at low RPM. Again, the power efficiency of the system is very high, higher than the DC-DC converter, as the lack of the smoothing capacitor will save the system from leaking currents.
A drawback for this system, except the complexity and higher cost, is the acoustic sound that may be generated to the fan. Because the pulses are driven directly to the coils of the motor, the torque is generated also in pulses. If the frequency is inside the acoustic spectrum (20Hz - 20 KHz), then most likely an annoying sound will be heard from the fan. This can be easily avoided if the PWM frequency is above the acoustic spectrum (above 20 KHz). This system is called High Frequency PWM controller.
This method has another major drawback. If high frequency PWM pulses are used to control a 3-wire fan, then the feedback from the tach will be unusable! Read the article "How PC Fans Work" and you will understand why this happens. 3-wire fans cannot be controlled with PWM pulses and have simultaneously rpm feedback from the tach cable. For this, the new PWM fans have been manufactured. These fans have a 4th wire. The pulses are driven to the 4th wire, while the controller of the fan (and the Hall Sensor) are powered with constant current. The 4-wire fans have a built-in pass-component. This pass-component will switch the current according to the PWM pulses ONLY on the coils of the motor and NOT on the whole supply of the motor. Therefore, from the tach wire, pure rpm feedback can be taken. It should also be mentioned that, because 4-wire fans have a built-in pass component, there is no need of an external pass-component to exist on the circuit! The pulses are driven directly from the PWM generator to the control wire.
Nevertheless, there are some tricks to control a 3-wire fan with PWM pulses and have simultaneously rpm feedback from the tach wire. The most reliable and known method is the "Pulse Stretching" method. According to this, the 3-wire fan is powered with PWM pulses. From time to time, the fan is powered with direct current. During this time, the feedback form the tach can be clearly read. The direct current is kept as long as it needs for the microcontroller to read a full period of a pulse from the tach wire. Immediately after the feedback is read, the controller will continue sending normally PWM pulses to the fan. This is a rather complicated method that requires the use of a microcontroller, yet, it will work!
Another method that i tested extensively is to try to smooth the feedback from the hall sensor with a capacitor. The tach feedback will actually be the PWM pulses. But the amplitude and the offset of the pulses is not always the same. It changes according to the pulse period of the feedback (HIGH or LOW). Using a set of filters and amplifiers, the feedback may be smoothed and squared again. I run some tests with such a circuit and frankly, i got some results. But the problem is that it will not always work the same. I found out in general that, this method should be used for fans that rotates not faster than 1200rpm - 1400rpm (like the silent 120mm fans). But again, to be sure if a fan will work with this method someone must test it.Example circuit(s):
Mix and match
Someone can use his imagination to make a mixed controller. For example, you can have a fan controlled with PWM pulses running at a low speed. A thermostat will check the ambient temperature. If it exceeds a critical level, the fan will be forced to run at full speed.Example circuit(s):
What can be done with the tach feedback?
First of all, if the fan is connected to a motherboard, then the tach signal will inform the computer with the speed of the fan. That is the original reason of existing of the tach wire.
But if you make a fan controller by yourself, then probably have not used this wire. Pity it is!!! I will give you an idea. First of all, you can make an rpm counter. Some of the circuits i presented above have already an rpm counter implementation. You may get some ideas. If you do not want to use a microcontroller, then you can do something easier. You can convert the pulses into voltage. This is called Frequency-To-Voltage conversion. Then you can compare this voltage to a pre-set voltage level and monitor the operation of the fan. If the fan is stalled or the rpm is too low, then the voltage will be changed significantly and the comparator will activate the alarm.Example circuit(s):
Let's see the pros and cons:
Fixed speed controllers
Stepped speed controllers
Linear speed controllers
Linear DC-DC converter speed controllers
PWM speed controllers