It wouldn’t matter if our hardware was cheap, portable and easy to build if it didn’t actually work. We characterized it with LUDOX and fluorescein to ensure it wasn’t just making up the readings.

We used the interpreter to record the data from device. These readings were then exported and analysed to plot the graphs.

Optical density

Optical density measurements were taken using different concentrations of LUDOX. We made up a dilution series, with initial concentration 50µM and diluting by half each time. We used 6 dilutions. We used the same cuvette (washed with distilled water between readings) to avoid random error introduced by the differences between our cuvettes. We characterized our densitometer using Lee Filters’ - Fire (019) and Medium yellow (010) - and without a filter.

The readings for optical density are consistent with little variation between readings, except for LUDOX concentrations above 12.5µM. Above 12.5µM the readings began to decrease. We are unsure as to why this would happen, but this is clearly not caused by random error as it happened with all the filters.


For the fluorometer function of our device, we used fluorescein to take measurements. We made up the same dilution series, with initial concentration 50µM and diluting by half each time but with fluorescein instead of LUDOX. We characterized our fluorometer using LEE Filters’ Twickenham Green (736) and without a filter.

The results are consistent and demonstrate the linearity of our fluorometer. The pipetting error in the dilutions is visible on the graph, as the distances betwen the points and the lines of best fit are similar both with the green filter and with no filter. The no filter readings are higher as fluorescence is provided in arbitrary units, as they are the light intensity readings without being calibrated against a blank. Without the filter less light is blocked and thus there is increased levels.

We were surprised results without the filter. We expected greater errors caused by variations in ambient light and the input blue light. The input source light must therefore be very consistent and also quite directional so as not to significantly affect the readings. For reference, we used a Nichia Superflux LEDs, product code NSPBR70BSS. However, we don't think there's anything particularly special about this LED - it may be an idea for future projects to test LEDs to see if there is a benefit to using a specific type or brand.


The results follow the trends we expected them to, demonstrating that our combined fluorometer and densitometer has high linearity and thus is an effective measurement tool. It could therefore be developed commercially to be used in diagnostics, especially in developing countries. This is further explored on our silver human practices page.

We characterized the linearity of our light sensor (TSL235R) with optical density readings of LUDOX, and found it performed very well.

Aachen 2014 characterized the linearity of the same light sensor (TSL235R) to detect fluoroescence of purified iLOV,[1] and our results with fluorescein are concordant. Using the same light sensor also means our work further characterizes their fluorometer.

Unfortunately we don’t have a plate reader at our school so we were unable to test the same samples with commercial equipment to compare the performance. However, the linearity graphs and Aachen’s previous work characterizing the light sensor (TSL235R) demonstrate that our hardware has real potential to reduce the costs and increase the accessibility of a screening programme.


  1. (n.d.).Aachen OD/F Device, Retrieved October 24, 2017, from