Today, process technology has advanced to 32nm, resulting in an increase in the number of transistors per unit area and a reduction in package size. At the same time, system designers are trying their best to reduce system size by increasing the component density on boards, adding as many features in the design as possible to deliver the best possible products in terms of space and size.
Increased transistor density inside the chip, higher operating speeds, and increased component density on board in modern electronic systems has led to relatively more heat being generated in these systems. All this has made thermal management an integral and critical part of system management in all application domains, including automotive, industrial control, consumer electronics, battery powered systems, and so on. Many systems are equipped with cooling fans to deal with the heat generated, causing designers to realize the necessity of cost-effective, reliable, noise-free, and power efficient temperature-based closed loop fan control systems.
Before going to the implementation of the fan control system, let's have a quick look at the basics of heat generation and heat transfer. Electrical power dissipation (Voltage * Current) is indispensable in any electronics circuit. This electrical power dissipation results in the generation of heat and causes the junction temperature to increase above the ambient temperature. For reliable operation of the device, the junction temperature of the device must be kept within specified limits. Sometimes, the requirement is not only to keep the junction temperature within the specified operating range but, for some applications, to keep it at a specific value since these devices show different characteristics at different temperature ranges. For instance, oscillator frequency, ADC offset, and thermal noise are functions of temperature and variation in these parameters may be undesirable for some applications.
As long as power is dissipating in the device, junction temperature will be more than the ambient temperature. The ability of the device to transfer heat depends upon the thermal resistance. A thermal circuit can be considered similar to an electrical circuit. Figure 1 shows the equivalent circuit of thermal system:
In this circuit, T1 and T2 are temperature and θ is thermal resistance. Temperature can be assumed to be analogous to voltage in electrical circuit. Thermal resistance is analogous to electrical resistance and heat flow is analogous to current flow. Considering this thermal-electrical analogy, it can be seen that a higher temperature difference increases the heat flow. Similarly, the lower the thermal resistance, the more heat flow there is.
In an IC, flow of heat from junction to ambient air depends upon the difference between the junction temperature and the ambient air temperature. Cooling fans operate by blowing away hot air and allowing more heat to flow from the junction to its external surroundings. There are several types of control systems used for fans:
The simplest way to use a fan for thermal management is to use one that needs no feedback control. The fan runs at its maximum rated speed at all times, ensuring foolproof cooling with the least associated cost. However, while this reduces the installation/manufacturing cost, it increases the running cost of the system. The fan's lifetime is also decreased since the fan runs all the time (a fan's life is generally defined by the number of revolutions). Another major disadvantage of such a system is its high power consumption since the fan keeps running at its highest speed even though this behavior is not required at most times. In such systems, a fan is selected for the worst case cooling, and the system rarely runs in that condition. As there is no feedback and no control system, there is no way to check the fan's present condition. If the fan breaks down, smoke or system failure might be the only feedback the user gets.
Another way of controlling the fan is using a linear control system. In such a system, the speed of the fan is controlled by changing the voltage input to the fan--the lower the input voltage, the slower the fan speed. The advantage of linear control is its noise free operation due to lack of coil switching. If we look at the limitations of this control system, we see that the operating-voltage range of the fan limits the speed range. For example, the minimum voltage needed to force the fan to run may be more than half the maximum operating voltage of the fan. Also important to consider is that the efficiency of linear regulators is less at lower output voltages, meaning these systems are not optimized for power consumption at lower fan speeds. Another major drawback is the cost of the linear voltage regulator circuit required for such a system.
The most extensively used technique to control fan speed is using a PWM. Using this method, the fan either runs at its highest speed or it remains off. Major advantages of this control method over linear control include simplicity of the circuit, cost effectiveness, and efficiency. Another advantage is the average speed range possible since the lowest speed is not limited by the minimum operating voltage of the fan as it is in the case of a linear control system. The primary disadvantage of this method is the noise induced due to the switching of coils. However, this can be overcome by operating the fan outside the audible frequency range. Figure 2 shows the basic thermal management system block diagram using the PWM control method.
There are different variants of fans available, and each type can be the best fit for a particular application based upon cooling needs, system cost, reliability requirements, etc. These variants are generally classified based upon the number of wires used: 2-wire fans, 3-wire fans, and 4-wire fans.
A 2-wire fan has two terminals for power and ground. Changing either the supply voltage or the power duty cycle, which is equivalent to changing the average supply, controls its speed. This kind of fan is used mainly in open loop temperature control systems when there is no tachometer signal available to provide fan speed feedback to the control system. This type of fans is used in cheaper thermal management systems. The limitation of a tachometer isn't really a limitation, and it can be worked around. The winding commutation in the motor causes the fan's current consumption to fluctuate. If the frequency of the fluctuations and the number of permanent magnetic poles of the rotor is known, the mechanical rotation speed of the fan can be readily calculated.
A key point to note here is that no one would like to pay for the additional current sensing circuit if there isn't any budget for a 3-wire or a 4-wire fan. As a consequence, a single chip that can do all the jobs of measuring the commutation frequency, temperature measurement, and controlling the fan speed by means of PWM will be an ideal choice for the application. Measuring the commutation frequency can be done by differentiating the current sense resistor's output, amplifying the signal, and then feeding it to a comparator, which will control the capture input of a timer. Once the frequency is known, fan speed can be calculated. Sensing the current fluctuation also helps in recognizing if the fan is working or not. Figure 3 shows the block level implementation of a commutation frequency measurement using PSoC3/5 from Cypress.
A 3-wire fan which has an extra wire for a tachometer output. As far as controlling the fan speed is concerned, the method is the same as for a 2-wire fan. With a 3-wire fan, both the fan motor and hall sensor circuit share the same supply. Thus, the tachometer signal is valid only during the on period of the PWM that is controlling the fan. The tachometer signal reading must be synchronized with the PWM on time and must also be long enough to capture one complete tachometer cycle.