Forced Air Cooling: Sourcing an Appropriate Fan

Dissipating the heat generated by certain components is a key part of electronics design. Consequently engineers need to be sure that they put an effective thermal management system in place in order to ensure that operational reliability is maintained. There are a variety of different technologies that can be employed for heat dissipation purposes. These will be based on the principles of either conduction, convection or radiation. By having a good understanding of the thermal path for removing unwanted heat from their respective systems, engineers will be able to decide on the best optimised technology to achieve desired results.

Though conduction represents the easiest way of removing heat from a system, if the electronic components involved are placed inside an enclosure (such as a rack-mount), this will impinge on the efficiency with which heat removal is carried out. Therefore, the vast majority of systems where an enclosure is being utilised, will rely on forced air cooling (for which some form of fan will be needed) to tackle the build-up of heat. There are a broad array of fans on the market, so proper consideration should be given to which option is most suitable for the application in question.

When looking to implement a thermal management solution, engineers are advised to create a detailed thermal profile of their electronics system, so that they have complete visibility of where heat is being generated and what are the quantities involved. This should cover every possible operating condition. By distributing temperature sensors around the PCB and within the enclosure itself, all the necessary data can be acquired. Once this is done, the engineer can look at the positioning and the extent of the thermal management needed to attend to the system. In addition to compiling information on the heat generated by the constituent electronic components, this profiling can also provide engineers with valuable insight into the airflow around the system and where there might be impedance.

By knowing the system impedance (in terms of the drop in air pressure witnessed between air inlets and outlets), is pivotal when it comes to accurately estimating the overall airflow that the specified fan will need to produce. It is possible to derive the system impedance through measurement of the pressure drop using sensors. Alternatively, by placing the whole system in an air chamber, better quality results can be achieved. When larger systems are involved (like modern high density data centres), more sophisticated computational fluid dynamics are employed. Through this technique thermal profiles can be determined to a higher degree of precision.

Implementing a thermal management mechanism that is effective, while respecting both budgetary and space constraints can be a difficult balancing act. Key to this is knowing how much the internal temperature of the system can rise without the risk of operational failure being increased. Assessing the system and identifying the ‘most critical’ component in relation to its operating temperature will enable the maximum ambient temperature to be obtained. The combined power dissipation for all relevant components (such as MOSFETs, microprocessor units, etc.) will result in a total power dissipated by the overall design. Power dissipated (in W) directly equates to energy (Joules/s), which manifest itself in the form of heat. Equation 1 describes the relationship between temperature rise that stems for the operation of the electronic components in the system and the airflow.

Equation 1:

q = w x C_p x DT

q - amount of heat absorbed by the air (W),

w - mass flow of air (kg/s)

C_p - specific heat of air (Joules/kg • K)

ΔT - temperature rise of the air (°C)

By knowing the maximum temperature that is permissible within the enclosure and the amount of heat generated (based on the cumulative power/heat dissipated by the components), an exact calculation of the amount of airflow required can be made - as shown in Equation 2.

Equation 2:

 Q = [q/(rx C_p xDT)] x 60

Q - airflow (m³/min),

q - amount of heat to be dissipated (W)

r- density of air (kg/m³)

C_p - specific heat of air (Joules/kg • K)

ΔT - temperature rise of the air (°C)

By substituting constants for C_p and rat 26°C, the following general equation for calculating airflow can be arrived at.

Equation 3:

Q = 0.05 x q/DT

The airflow figure that corresponds to the system can now be compared against fan specification, so that a suitable product can be sourced. Fan manufacturers will characterise their fan units by providing a performance graph that plots airflow against static pressure - as shown in in Figure 1. The fan described in Figure 1 is from CUI’s CFM-120 offering.These 120mm x 120mm frame axial fans have a dual ball bearing construction and support speeds of 4600 RPM.

Click image to enlarge

Figure 1: Example of a manufacturer’s fan performance curve

It must of course be acknowledged that Equation 3 gives an idealised representation, with effects such as back pressure from the enclosure not taken into account. In reality there will always be some system impedance (as discussed earlier) to allow for. To determine the real-world requirements of the fan, it is important that the system impedance is calculated. This can then be plotted on the fan’s performance curve, taking where the lines cross as the operating point for the fan - as seen in Figure 2.

Measuring the airflow through an enclosure, as previously mentioned, can be done via use of an airflow chamber. If that is not feasible, then an alternative approach is to settle on an operating point that is comfortably above the figure derived from Equation 3 so that some headroom is allowed for in terms of the airflow supported.

Click image to enlarge

Figure 2: System impedance plotted on fan performance curve to determine operating point

During the design phase effort should be made to keep the system impedance to a minimum, as this will have clear knock on effects when it comes to sourcing a fan - since a lower spec unit can then be chosen. It is advised that, whenever possible, the areas close to air inlets and outlets remain fairly clear of components. Component placement on the PCB should encourage airflow to and around critical components, using guides to accentuate this when needed. Furthermore, due consideration should be taken if the system will be deployed at altitude (such as in aerospace applications). The equations previously cited assume for air density at sea level. As air density reduces with altitude, a significant increase in the airflow will be required accordingly.

Fan selection

In addition to being offered in both AC and DC configurations, fans are generally categorised by the way in which air enters and leaves the unit. For axial fans air will exit in the same plane that it entered (drawing air in from one side then expelling it from the other), while centrifugal fans (or blowers) are constructing so that the airflow is redirected and leaves on a completely different plane. The latter can effectively compress the air, allowing it to deliver a constant airflow under different pressures. The volume of airflow needed and the static pressure of the system will dictate which is the most suitable type of fan to meet given application requirements. Axial fans are suited to systems where high airflow is mandated and there is low static pressure. Conversely, centrifugal fans offer lower airflow, but can deliver it against higher static pressure levels.

Thought should also be given to issue of noise (both of an audible and electrical nature). DC fans may exhibit impressive specs, but the audible noise that is associated with them could be a problem for some system designs. Generally axial fans will have lower audible noise than centrifugal fans. As already noted, reducing system impedance, will lower the airflow needed by the system and this will, in turn, will curb the audible noise. DC fans will also generate electro-magnetic interference (EMI) and this needs to be negated. Ferrite beads, shielding or filtering can be used to do this. 

Activecontrol technology

Due to there combination of low power operation, compact size and high airflow, axial fans are widely used in rack-mount enclosures. In many cases additional features are incorporated to further boost performance parameters by providing greater control over the speed of operation. This means that their impact on the overall power consumption of the system can be minimised.  As already described, calculating the minimum airflow rate required to adequately cool a component-laden PCB inside an enclosure will enable engineers to select a fan capable of delivering the cooling needed under all conditions. This makes the assumption that the fan will run constantly, even when maximum cooling is not required. While this is not likely to result in failure, it does adopt worst-case conditions at all times and is clearly therefore quite inefficient from a system point of view, with the operating lifespan of the fan being significantly shortened as a result. Because of this it has become common practice to monitor the temperature within an enclosure and only turn the accompanying fan on when required.  While this approach will improve the fan’s lifespan and reduce audible noise, it can lead to problems in relation to thermal lag. It may also introduce a fault condition if, for some reason, the fan is unable to start because there is an obstruction.

To address this, modern DC axial fans, like the CFM series from CUI, have auto-restart protection functionality built in. This detects when the fan motor is prevented from rotating and automatically cuts the drive current. Models in this series, which include the popular CFM-60fan, also offer optional controls such as tachometer and rotation detection sensors. The tachometer gauges the rotational speed of the fan motor and provides a pulsed output that can be used within control circuitry (as shown in Figure 3). If the motor stops, the output stops pulsing and stays at either a logic high or logic low state. The rotation detection feature doubles as a lock sensor, so that if the fan motor stops the output is driven to a logic high state and remains at a logic low during normal operation (see Figure 4).

Click image to enlarge

Figure 3: Schematic showing how the signal supports speed detection

Click image to enlarge

Figure 4: Schematic that illustrates output signal indicating stall/lock fault

In summary, with semiconductor devices exhibiting high degrees of complexity and PCBs becoming more densely populated, thermal management is proving to be an ever-greater challenge. Forced air cooling is a highly efficient method via which to dissipate heat from PCBs situated inside enclosures. By correctly specifying and subsequently sourcing a fan that is well matched to the requirements of the system, a long and trouble-free operational lifespan can be benefitted from. Contrary to this, if the level of forced air cooling is insufficient for the system’s needs, the fan will almost certainly be responsible for the resulting failure - even though that failure will typically arise from some other critical component failing. Care and attention must be given to the fan selection process, so that the right fan for the system design is quickly and cost-effectively sourced. Mouser and CUI work closely together to offer the engineering community highly advanced forced air cooling productsthat are optimised for their particular needs.  

Mouser Electronics