Electronic Design

# For Optimal Cooling, Rely On Closed-Loop, Fan-Speed Control

With Minimum Fuss, A Closed-Loop System Can Yield Maximum Performance For The Most Demanding Applications.

The benefits to designing a cooling system with closed-loop, fan-speed control abound. If done properly, this control method can result in fan-noise reduction under normal thermal conditions, significant savings in energy costs, and increased system reliability. Also, because a closed-loop, speed-control system senses the equipment exhaust air temperature, a temperature-sensing alarm output can be a very economical safety byproduct. With the ready availability of off-the-shelf ac and dc controller boards, as well as dc fans with built-in, closed-loop control, this technologfy is becoming even more attractive (see Fig. 1).

The principle of closed-loop speed control is really quite simple; the thermostat in your home operates on a closed-loop principle. The thermostat senses air temperature and keeps it nearly constant by turning the heating system on and off. In an electronic-system cooling application, fans run at whatever speed is necessary to hold equipment air temperature near constant. This converts a fan from a "flow-regulating device" to a "temperature-regulating device." Isn't that what you chose a cooling fan for?

With your fans now regulating temperature, you can add or remove a circuit card, or you can take your system to Denver or Death Valley. You don't have to worry about customers changing air filters on time, or the air conditioning breaking down. Even in these cases, your fans will hold equipment temperature near constant by running at the required speed.

Closed-Loop System Design
When it comes to designing a closed-loop system, a number of key parameters or variables must be taken into consideration. The first of these is the equipment temperature rise, ΔT (the temperature rise of the equipment with fans running at full speed). This can be defined as:

ΔT = TE - TA

where:

TE = Exhaust temperature of equipment at full-fan speed.

TA = Ambient room temperature.

A designer can vary ΔT by choosing air movers with different flow rates, measured in cubic feet per minute.

Another variable to consider when designing closed-loop control is the control temperature. This is defined as the equipment-exhaust temperature, above which fans run at full speed. The recommended control temperatures are shown (see Table 1).

Example
A cooling system has no speed control. Thus, fans constantly run at full speed. The system temperature rise is 8°C. In this example, as the equipment inlet air temperature varies from, say, 21° to 32°C, the exhaust temperature varies from 29° to 40°C (Fig. 2).

Using the same system with a closed-loop control at a 40°C control temperature, with the fans idle at half-speed, below a room ambient temperature of 21°C. The fans run at full-speed above a room ambient temeprature of 32°C. In this example, as the equipment inlet air temperature varies from 21° to 32°C, the exhaust temperature varies from 37° to 40°C (Fig. 2, again).

Other Options
At this time, one might ask, "Why not sense the inlet air temperature (open-loop control)?"

A cooling system could be designed by varying fan speed in response to inlet air temperature. This is an open-loop, or compensating, design. An open-loop design may work nearly as well as a closed-loop design—but not all the time. Going back to the thermostat in your home, an open-loop design is equivalent to placing the thermostat outdoors. This could work—until you light the wood stove or open a window. The internal thermostat (closed loop) can sense these changes, while the external thermostat (open loop) cannot.

The electronic equivalents to the wood stove and window example are: adding or removing cards from a card cage, blocking an air inlet, or having a fan failure (in a system with multiple fans). A closed-loop design (sensing exhaust air temperature) senses these changes, and adjusts fan speed to hold equipment temperature near constant. The open-loop design (sensing inlet air temperature) does not.

A closed-loop design also increases or decreases fan speed according to altitude; an open-loop design does not. A fan in San Francisco can run at a lower speed than the same fan in Denver, for the same thermal load. As stated, the fans in a closed-loop system run at the speed which holds exhaust temperatures near constant. In Denver, where the air is thin, higher velocity is required to achieve the same amount of cooling (see Table 2).

Noise Reduction
A significant advantage of sensing temperature to control fan speed is the potential for considerable reduction in fan noise (under normal thermal conditions). Most equipment designers select fans for the worst-case thermal conditions anticipated, such as a 90°F day in a Denver office, with a broken air conditioner. This creates excessive, unnecessary fan noise under normal thermal conditions, such as a 70°F office building in New York.

Because there is a fifth-power relationship between noise level and fan speed, a small change in fan speed will cause a large change in fan noise. The equation for determining the noise level of a fan, at less than full speed, is given as:

LS = L1 + 50 log S

where:

S = Fan speed as a fraction of full fan speed.

LS = A-weighted noise level at fan speed S.

L1 = A-weighted noise level at full speed.

Example: A 300-ft3/min. fan has a full-speed noise rating of 59 dBA. What would the noise rating be at half speed?

LÞ = 59 + 50 log (Þ)

LÞ = 59 - 15

LÞ = 44 dBA

Energy Savings
Controlling fan speed according to temperature saves energy. There is an approximate square-law relationship between fan speed and power consumption:

PS = P1(S)2

where:

S = Fan speed as a fraction of full-fan speed.

PS = Power consumption at a fan speed of S.

P1 = Power consumption at full speed.

If a fan is running at a reduced speed under normal thermal conditions, the potential for saving power is very high.

Example: A 1200-ft3/min. blower uses 200 W at full speed. How much power would be required to run the blower at 600 ft3/min. (half speed)?

PÞ = 200(Þ)2

PÞ = 50 W

Increased Reliability
Bearing failure, caused by heat and wear, is the most common cause of fan and blower failure. By allowing air movers to run at reduced speeds much of the time, speed control increases fan life. Typical speed-control circuitry can be expected to have a mean time between failure (MTBF) in the 106 hours range—far greater than a typical fan running at full speed. The negative impact of the added control circuitry is significantly outweighed by the increased fan MTBF at reduced speeds.

Adding speed control to a cooling system prevents fluctuations in line-frequency and supply-voltage from adversely affecting fan speed. This ensures that fan speed varies only in relation to temperature.

One of the most intriguing aspects of fan-speed control is the positive effect on system reliability. One might think that adding another feature to any system could only decrease its reliability. This is not the case with speed-control circuitry.

With fluctuating semiconductor junction temperatures being a major source of component failure, maintaining a constant temperature inside the enclosure can greatly improve overall system reliability. A closed-loop, speed-controlled system, with the proper slope (3°C in Figure 2) will tend to hold semiconductor junction temperatures constant, even when:

• Circuit boards are added or removed
• Air inlets are partially blocked
• Equipment changes altitude
• One fan fails in a multiple-fan system

No matter how thermal conditions inside the equipment enclosure change, fans run at whatever speed is necessary to hold junction temperatures constant.

Alarms: Temperature Vs. Tach
The use of fans with a tach output, which senses the speed of the fan, has become very popular during the last several years. This tach output is often used to sense and report fan failure using an LED, audible, or electrical alarm signal. A closed-loop speed control may be used in conjunction with the circuitry designed to accept the fan-tach pulses. This arrangement will trigger an alarm faster than a temperature alarm, and has a diagnostic advantage whereby the system can tell a service technician that a fan has failed. By using a tach alarm, one can also avoid the problem of locating an air- or surface-temperature sensor in the critical spot.

One disadvantage of a tach alarm, versus a temperature alarm, is the problem of cost. If you have a closed-loop, speed-controlled system with six fans, adding tach alarms means using six tach fans (three-wire fans), alarm circuitry for the six fans, and additional wiring for each fan. The costs for adding tach detection continue to increase with each additional fan in the system. Adding a temperature alarm, which senses exhaust air or surface temperatures, requires one simple circuit and one output device, but not a special fan.

Another disadvantage to using tach alarms over temperature alarms is the potential for false positives. For example, if the system described above is alarmed to the hilt, and a user unwittingly blocks the equipment exhaust, the fans rise to full speed because of the closed-loop speed control. However, with nowhere for the hot air to go, the temperature continues to rise inside the cabinet. Unfortunately for the piece of equipment, no alarms are triggered, because all the fans are running.

As the equipment operator, I do not care if all the fans are running. My biggest concern is whether my equipment is overheating. Only the service technician that I call after my machine overheats needs to know if all the fans are running.

To model your own closed-loop cooling system, check out the interactive demo at www.controlres.com, or call (978) 486-4160 and request a 3.5-in. demo diskette.