A Phototransistor works in the same way as a Photodiode, in that light (or IR) gives energy to the atoms freeing electrons to become negative charge carriers in parts of it, causing a current. The main difference is because the effect is used as the base of a transistor, the output is many times greater. The potential difference (A-B) was 0v when the sensor detects no IR, and higher (closer to 6V), when the sensor detected a lot of Radiation. The current is proportional to the Intensity within limits as after a certain point, no more free electrons can be released.
I first Investigated what contrast in signals I would get using a digital voltmeter to measure the Potential Difference across A-B. With the sensor 5mm from the material, I found that the silver paper reflected the IR to give around 4v Potential Difference. The Black paper did not change the Potential Difference, implying that it reflected none of the IR radiation. I now assessed the readings I could get from the cork using a Cathode Ray Oscilloscope to measure the PD. I found that at first, the PD readings for the stationary cork at the silver points varied, when they clearly should not. I soon realised that this was due to some scattering of the IR radiation by the imperfectly curved cork. To minimize this effect, I found that using the minimum distance possible between the cork and the sensor would lessen scattering.
To measure the speed of rotation from the CRO, I could measure the time units (x-axis) it took for one rotation. Since I had split the cork into quarters, one rotation was represented by two peaks (silver sections of high reflection) and two troughs (matt black sections of no reflection).
I noticed the peaks were extremely irregular, I found that this could to some extent be attributed to the cork rotating around a slightly off-centre axis. This led to the distance (and angle) between the cork and the sensor, varying. This, together with the scattering I had previously observed, explained the non-uniform peaks. This in itself was not a problem, as the important detail as regards to measurements was the time units for two troughs and two peaks, which was still clearly defined. I also noticed that the peaks were of irregular shape, alternating between two distinct shapes. This could be attributed to differences between the two silver quarters.
I initially used a manufactured anemometer to calibrate my own. Using a large directional fan, I measured the anemometer speed readings, and my own rotations-per-second readings over a wide set of preliminary speeds. Assuming the anemometer gave correct readings for the wind speed, I was able to use the results to create a calibration graph.
I noticed from these results that the central section showed a clear linear trend, while at extremes of speed (<3 or >6), the graph flattened out. This implied that these results were outside the sensors range limits. The reasons for the range limitations seemed to be (from observation), at higher speeds due to excessive vibration of the rotating axel and its mounting, and deformation of the fan blades rather than low resolution in the Phototransistor itself, as I had tested it at higher speeds using an electric motor to rotate the cork, with positive results. At lower speeds, it is likely to be due to the friction between the axel and it’s mounting, causing the wind force not to overcome the inertia of the blades. I made some adjustments to the mounting to minimize these effects, but for the purposes of the Actual experiment, I limited the readings I would take to between 3m/s and 6m/s actual speed.
Again, the limits of the sensor became apparent, but the clear linear pattern of the central results shows that the sensor I had built was reacting correctly to changes in the wind speed. I noticed during the experiment that the points of contact between the axle and the mounting had become warm. This was clear evidence that some of the kinetic energy I was measuring had been lost to friction as heat. The linear part of the graph roughly conforms to the linear equation y=-2.4x-4, which, when rearranged to x=(y+4)/2.4, allows me to predict the wind-speed at any revolution speed within the limits of the sensor.
To extend the use of my sensor by decreasing its limits, I would need to improve its construction, both in materials and methods. Using more precision machinery to manufacture the parts from more uniform materials, I could reduce any friction within the apparatus (incidentally lowering its inertia point), and improve its sturdiness to prevent vibrations. I would also need more lightweight and sturdy materials if I wished to make the sensor truly portable. To include direction sensing, it would be simplest to include a second, weathercock based rotary potentiometer sensor, the results of which could be calibrated to give a direction. It would be difficult (if not impossible) to combine the two functions in one sensor of the type with which I have been experimenting.