Closed-Loop Rotational Speed Control of Cooling Fan

Download report in PDF

Download PowerPoint slides


            The main objective of this project is to design and build a control system for a computer fan. The user inputs a desired angular velocity in RPM into the system, and the computer alters the voltage sent to the fan in order to maintain a constant angular velocity. We then show that this closed-loop control system is functional by inputting different angular velocities, and varying the airflow resistance to the fan. Through all of this, the control system should approximately keep the RPM of the fan constant. We also want to find the power consumption of the computer fan and see if it increases as we add resistance to the airflow. From this power consumption data, we can determine the efficiency of the fan.



Front view of mechanical setup with encoder


            To accomplish our goal, we built a mechanical setup that could introduce varying levels of resistance to the airflow. We also built a nice homemade encoder using a light and a photoresistor to give two pulses every revolution of the fan. Through some other circuitry, we turned that signal from the photoresistor into a rectified square wave. This then went into a tachometer, basically converting a frequency into a voltage. After some more signal conditioning, this voltage was inputted into the computer via the ADC. After multiplying by a calibrated conversion factor, we find the angular velocity of the fan. In similar ways, the current across the fan is inputted to find the power used, and the voltage outputted by the DAC were both inputted into the ADC. The data in LabVIEW was dumped out into Excel and further analyzed.



Signal processing circuit board



LabVIEW user interface



Proportional controller in LabVIEW code (download here)





            The graph above shows both the actual and desired angular velocity of the fan in RPM as a function of time. For this case, the opening in the back of the fan was nine square inches. Here we can see the five different trials represented by the thin line. The thick jagged line shows what the actual RPM of the fan is. As the desired RPM is instantly switched, the actual takes a bit of time to catch up. There is a marked overshoot, but the system compensates less and less each time so that at about 10-15 seconds, the RPM of the fan becomes approximately constant.




            The power drawn by the fan obviously increases as the angular velocity of the fan is increased. This relationship can be fit with a binomial line to a good degree of accuracy. Something else that this data tells us is that the power curve is higher for larger amounts of air resistance. The line that corresponds to the back of the fan being completely blocked off is the greatest, and the line that corresponds to the fan being totally open is the lowest. This is what we would expect because for the same RPM, more power must be used to overcome the air resistance. Another thing to note is that this difference in power is more noticeable as the velocity is increased

            In the end, our setup and circuits worked beautifully as planned. The encoder setup with the light and photoresistor gave us a very clean square wave, making the frequency easy to find. At times, the minute details of all the circuits and the shear number of wires made things difficult and confusing, but all in all, we were able to work through all of our problems. As expected the efficiency of the fan decreased as we added resistance to the fan and increased the angular velocity. The system did a superb job of maintaining the desired angular velocity as well. Though working on this project we put together all of the knowledge about various circuits and LabVIEW to create a meaningful closed-loop control system.


Produced by David Choi, Jay Gainer, Shed Laws & Karl Stensvad – December 2001 – Prof. Espinosa, ME224