the linearity of that resistor is going to be temperature dependent and non-linear. it may be necessary to get another device to characterize it.

EDIT: phtt... look at the blog post and the awesome linearity adjustment.

I want to highlight this awesome part of your blog post:

The next step is to see if quantitative fluorescent measurements could be done by using another LED as a detector (http://www.instructables.com/id/LEDs-as-light-sensors/). This could allow for measurement of the oil in water for example. If you choose an LED with a green or red emission they have build in filters so a combination of a blue (405nm) LED and a green LED as a detector could be used to determine quantitatively the amount of oil in water. It looks like devices already exist that use LEDs to do this measurement. This may interest the publiclab community as the characterisation of the type of oil could be done with a low cost spectrometer and the concentration measured using the colorimeter. (http://publiclab.org/wiki/oil-testing-kit)

Thanks Mathew,

The linearity in the sensor is a problem however it should be exponential so it can be removed.

I am imagining the measurements of the standards done at the same time as the samples as everything will have changed since the last measurement including the battery voltage (therefore the LED intensity) and the temperature which will effect the sensors accuracy. Another option is to use an arduino to provide a steady voltage and a temperature sensor to adjust the resistor value.

The measurement of oil using LEDs is already done in industry so this should be relatively easy to implement I will give it a go when I get back into the lab.

Check out this link. http://doras.dcu.ie/17889/1/Novel_Optical_Sensing_System-MCzugala_FINAL1.pdf I added the photo they have in this paper for the detection. LED_detector.tiff I have some more papers I will dig up that use LEDs as detectors as well.

Here is the paper I was thinking of and it is an open paper so everyone should be able to read it. http://www.mdpi.com/1424-8220/8/4/2453 There is a section on using LEDs as detectors as well.

Another paper (Dasgupta, P. K.; Eom, I.-Y.; Morris, K. J.; Li, J. Anal. Chim. Acta 2003, 500, 337–364.) which isn't open unfortunately shows a more advanced circuit that uses an opamp to convert a photocurrent from the LED to a voltage and I think it amplifies the signal also. LED_circuit.tiff They reported "As indicated in Section 4, the ability of three photodiodes connected in parallel as fluorescence sensors were initially tested with three different amplifiers (TL082, OPA627 and INA121, all from [155]) with 100nM fluorescein as analyte, resulting in S/N values of 1, 2, and 3, respectively. Compared to this, the PMT provided an S/N value of 35, even in this preliminary testing. As such, we report here fluorescence data mea- sured only with the PMT sensor. For a tighter budget or for pedagogic purposes, photodiode detection of fluorescence still provides sub-microMolar detection of com- mon fluorescent dyes with the instrumentation ampli- fier chip (INA121 is ca. US$ 5 each). If the ultimate in detection limits are not an issue, multicomponent fluorescence detection may actually benefit from a photodiode array where the array elements are individually read and are provided with different optical filters"

So it looks like they might be sensitive enough and if not then a silicon photodiode with a filter and opamp would give better resolution. They also note in the first paper I referenced the lack of use of LEDs as detectors due to sensitivity issues so it all depends on the minimum amount of oil that can be detected.