Building the Testbed
We made a wooden block to keep the relative positions of the sample, sensor, and light source constant as we swap out samples:
Originally we attempted to use a 3mW board-powered laser as the light source, but it was not powerful enough to generate any meaningful differences between samples. We also used AA-powered LED flashlights, but power drained noticeably over the sample period. The best results were obtained from an iPhone onboard LED (pictured), so we are going to try to find a similarly powerful LED that can be attached to the Arduino unit.
Testing vs Measured Samples
The sample range we experimented with was quite small, with samples at 2, 5, and 8 NTU, but the results looked promisingly linear (the NTU values were obtained from a calibrated HF Scientific portable turbidimeter):
In gathering this data, we noticed that the luminosity values for a given sample varied based on the rotation of the glass cuvette. We averaged the response for 8 angles of incidence, which seems comparable to the portable units’ 3 sensors at 90 degrees to each other. It appears that the fluctuations may have been due to scratches or imperfections in the glass (needs looking into).
Our next step is to obtain a wider range of sample cells, which should help us finalize the hardware components.