The purpose of this sensor is to measure the speed of the air flow exiting household air ducts. The motivation behind this airflow sensor is to develop a low-cost anemometer for household usage. With readily available parts, we can build an anemometer with an analog output voltage that is approximately linear to incident air flow.
The theory behind this airspeed sensor involves counting generated voltage pulses and converting the pulses into a useful voltage output. We generate voltage pulses by means of a reed switch, a magnet, and a wind sensing apparatus, such as a cup and hub assembly. Since a reed switch is a switch that is toggled by a magnetic field, we can use a magnet to close the switch repeatedly to generate voltage pulses. By mounting a magnet onto the sensing apparatus, we can generate voltage pulses from the rotation of the wind sensor. For example, a typical cup and hub assembly has spokes that lead out to the cups. We can mount a magnet onto a single spoke of the assembly, such that as the sensor turns, the magnet will also rotate at a regular frequency. A reed switch can be positioned close enough to the spokes so that the magnet can close the switch as it rotates. The reed switch must be attached to a voltage source to generate the voltage pulse when the switch closes. The voltage pulses can be converted to an analog output voltage by means of a frequency-to-voltage converter. When viewed as a black box, a frequency-to-voltage converter outputs a voltage that is linearly proportional to an input frequency. These devices are produced from a wide range of analog device manufacturers, and are freely available as samples.
A. Design Issues
The design of this sensor was inspired by the “Homemade Easter Egg Anemometer” from Otherpower.com. The webpage details how to use Easter egg halves to construct a cheap and relatively accurate anemometer.
We loosely base our design on the “Easter Egg Anemometer” coupled with some significant changes. We use a DC ball bearing fan without its internal copper coils rather than a hub and cup assembly to detect the presence of airflow. We do this for two reasons: the ease of placement over air vents and the wide availability of DC fans. Furthermore, we have a considerably smaller voltage supply. The original design uses a 9V battery to supply the system. However, our sensor is restricted to a stepped-up 5V supply from the wireless sensor node and an accompanying voltage step-up circuit. Finally, we sample the voltage output of the frequency-to-voltage device using the ADC of the wireless sensor, instead of attaching the output to an ammeter.
To ensure the same basic functionality as the Easter Egg Anemometer, we mount a reed switch extending to the center of the DC fan and glue two magnets to the center hub. The magnet and reed switch are used exclusively for the generation of voltage pulses.
B. Frequency-To-Voltage Converter
To convert the stream of voltage pulses to a DC output voltage, we utilize a frequency-to-voltage converter, the LM2917, from National Semiconductor. According to National, this device will output 1V per 67Hz of input frequency given a 9V supply voltage. Since we are restricted to a 5V supply, the produced output voltage of the LM2917 will deviate from the factory calibration. Therefore, we must perform our own calibration for the sensor.
C. Charge Pump
To produce a 5V supply voltage, we use a 5V DC charge pump, the ST619LB from STMicroelectronics. This device produces a regulated 5V output from the 2.5V reference voltage of the mote.
D. Voltage Inverter
An important aspect to note about the LM2917 (8 pin version) is that it can only measure an AC signal with a negative swing. For instance, a sine wave between 0V to 1V would not produce an output voltage from the LM2917. Thus, it is not possible to send positive voltage pulses to the LM2917 to receive an output. To produce a negative swing for the AC input signal, we use the ADM8660 from Analog Devices as a voltage inverter. As a result, the ends of the reed switch are attached to a positive 5V supply and a negative 5V supply. Also, we place an LED at one end of the reed switch to prevent the switch from being welded shut due to current spikes.
The figure above illustrates the architecture of the sensor. The design flow starts with the two AA batteries powering the mote. The mote then provides a regulated 2.5V supply from its analog VCC pin (Pin 1). The DC charge pump steps up 2.5V to 5V to supply three other components. First, it supplies power to the frequency to voltage IC. Second, it feeds into the input of the voltage inverter. Finally, it attaches to one end of the reed switch. The output of the voltage inverter attaches to the other end of the reed switch. One end of the reed switch connects to the input of the frequency-to-voltage IC. The output of the frequency-to-voltage IC connects back to the ADC input of the mote.
F. Bill of Materials
Component Description Purchase Location(s) Price
LM2917 Frequency to Voltage Converter National Semiconductor, Digikey $0.80
ADM8660 Voltage Inverter Analog Devices, Digikey $2.00
ST619LB DC-DC Charge Pump STMicroelectronics, Digikey $2.25
Reed Switch Reed Switch Cherry Corp, Digikey $5.00
Case Fan 80mm Ball Bearing Fan Electronics Stores $4.00
A. Fan Sensor
The figure below demonstrates the construction of the sensor fan. It is a typical 80mm, ball-bearing, computer case fan with its internal metal coils removed. Thus, the fan can spin freely without cogging, the resistance from the internal metal coils and magnets. It is important to note that sleeve fans do not work as well for this application. Sleeve fans with its internal sleeve removed do not allow the fan to spin as freely as modified ball bearing fans.
The hub of the fan has 2 magnets mounted approximately equidistant from each other in the center. We used an L-bracket to mount the reed switch such that it extends toward the center of the fan. The reed switch can be mounted with any type of durable adhesive. Most importantly, the reed switch must be close enough to the magnets for the sensor to operate correctly.
B. Circuit Schematic
Since we operate the frequency-to-voltage converter at a lower supply voltage, we must calibrate the sensor ourselves. The calibration setup is shown in the figure below.
The collection of clamps represents a layering of three components: a reference DC fan, a hot wire anemometer, and the sensor fan. On the very bottom, we used a 24V rated DC fan to generate reference airflow. Above that, we situated a hot wire anemometer to obtain a reference airspeed reading. On top, we placed the sensor fan. The figure below illustrates how this is done. The hot wire anemometer that we used was the HHF42 from Omega. It is important to note that the hot wire anemometer may not be necessary for this calibration. The instrument provides a higher degree of accuracy, but is very expensive. Using a less expensive handheld anemometer as a reference may produce similar calibration results. However, we performed the calibration with only the hot wire anemometer.
For calibration, we increased the reference fan voltage from 10V to 24V and examined the output voltage of the sensor at each step. We then plotted the relationship between sensor output voltage and reference airspeed, as shown in the figure below. From the plot, we see that the relationship between the output voltage and airspeed is approximately linear for the airspeed range that we are interested in.