This protocol describes a hot-wire low-speed anemometer based on a light bulb filament, used for measuring wind.
The temperature of an electrically-heated element will vary with the speed of any wind impinging on it. If the resistance of the element varies with temperature, then using it in a Wheatstone bridge makes it the sensor of an anemometer that is especially good for low-speed work. The resistance of almost all resistive elements-deliberately designed resistors for electronic circuits are exceptional-varies with temperature. For some semiconductors, such as that of the least expensive thermistors, resistance decreases with temperature, while for most other materials resistance increases with temperature. In practice whether the coefficient is negative (thermistors) or positive (other materials) makes little difference, although positive resistance materials are more forgiving of circuit idiosyncrasies.
The resistance of an ordinary incandescent light bulb increases about twelve-fold from unpowered to when it is supplied with electricity at its rated voltage. If the glass envelope is removed and the filament supplied with a much lower voltage-so as not to burn it out when operating in air-it provides a sensory element sensitive to very low speed air movement. While it lacks the spatial sensitivity of commercial hot-wire anemometers or of improvised thermistor anemometers, it can form the basis of an entirely practical instrument. In particular, the light bulb filament anemometer avoids the very high cost of hot-wire devices and the difficulty of procuring small numbers of the smallest bead thermistors.
Units, terms, definitions
Breaking a light bulb without damaging the filament can be done, I find, by placing the bulb on its side on a very hard surface and striking a minimal blow with another hard surface. For a second surface I use a paving brick held between my two hands. Remaining glass can be broken outward, bit by bit, with an ordinary pair of pliers. Figure 1, below at left, shows the bulb with exposed filament in an appropriate socket. The geometry of the filament will vary depending on the bulb’s rating and the manufacturer; any common form should work adequately inasmuch as the bulbs are designed to shine light equally in all directions.
Figure 2, below (also attached file at base of page), gives a circuit diagram for the anemometer. Since the device draws several amperes of current, the resistors dissipate significant heat and should be rated at at least 1 watt each; the commoner 1/2-watt varieties will burn out unless ganged in parallel. Housing and mounting arrangements I leave to the discretion of the user, noting only that no special constraints apply.
Power for the circuit can be supplied by any ordinary rechargeable wet-cell automotive or marine battery. Ones intended for use with motorcycles should be more than adequate in size. A voltage regulator, such as a Zener diode, of adequate capacity ought to be inserted between battery and circuit; alternatively a variable resistance can be readjusted using the digital voltmeter each time the device is used.
For a 60-watt bulb intended to operate with a 120-volt supply, the circuit in Figure 2 has worked satisfactorily. For bulbs intended for 220-volt usage, resistances are higher; that can be offset with one having a higher power rating. But even a 100-watt or 150-watt 220-volt bulb still has a higher resistance, than the 60-watt, 120-volt one, so the resistor opposite the bulb will have to be increased. At the same time, the supply voltage ought to be decreased from 12 to 10 volts to maintain about the same potential across the filament.
Calibration requires a source of air whose speed is adjustable and known. If a pre-calibrated hot-wire instrument cannot be borrowed, arranging such an air source is only a mild nuisance. I used the home-made Venturi tube shown in the photograph at the right in Figure 1 and diagrammatically in Figure 2 to obtain the calibration curve of Figure 3. It has an inner diameter of 12.7 mm and a constriction diameter of half that, so the area reduction is four-fold and the speed at the constriction correspondingly increases four-fold. Such tubes are not hard to construct; the insert that forms the constriction needs to be tapered, but in my experience the exact taper is not critical. The standard formula for Venturi tubes is the following…
where p is the pressure difference, the density of air, V the speed in the unconstricted part of the tube, S1 the cross section of the unconstricted part of the tube, and S2 that of the constricted portion. Readings in millimeters of water have to be converted to pascals to get speed in meters per second-the multiplier is 9.8.
The data points in Figure 3 represent averages of ten readings of the digital voltmeter. A good way to avoid bias is to open one’s eyes briefly enough to sample the meter and then enter the datum on a hand-held calculator, repeating, here, nine more times.
The manometer contains water with a little blue colorant. Paper shielding at the sides (but not top and bottom) of the sensor, not shown in the diagram, reduces the influence of air movements in the room. The speed in the wide pipe to which the sensor is exposed will be that of the Venturi tube (from the formula above) divided by the ratio of the cross sections of wide pipe and Venturi tube. The compressed air supply in a laboratory should provide sufficient volume flow to operate the system.
Ideally one should be able to enforce still air and set the zero point with the potentiometer in the circuit. In practice, free convection at zero imposed wind makes this impractical. I advise setting the zero-velocity point to a low but positive voltage on the meter. Figure 3 gives the calibration curve for the version reported here. One can see the malign influence of free convection in the odd lower end of the curve. Less heating of the filament (lower voltage) could reduce the effect, but it would render the instrument more sensitive to changes in ambient temperature.
The advantages of a heated-filament anemometer are its good low-speed sensitivity and its intrinsic omnidirectional responsiveness. It has the disadvantage of some sensitivity to ambient temperature changes; resetting the zero point to the value used during calibration will minimize the effect. Still, calibration at something close to the intended operating temperature is advised.