We’ve made good progress in the last week – we have a basic wireframe for the Photosynq Android app! The Android app is one of the most important parts of the system – it connects the Teensy arduino-based sensor to the backend database and web, and provides immediate feedback to the user as to how things are going. Here’s a screenshot of the app’s main page. As users create research projects, they may want only data in a certain area (for example, if I’m studying unique plants in Borneo, I don’t want someone adding data from Lansing), or in a specific season, or a specific time of day. The main page helps sort some of these things automatically, and identifies research opportunities that apply to you: ie you are in the right place and right time to take part in. Any user can create their own research through the web page, and it be searchable via the app. I modelled this roughly after the Google Play store.
If you’d like to give us some feedback, feel free to download the whole wireframe and walk through it. It’s located on our project management page at Open Design Engine here (you’ll also have to install Pencil from here).
We also managed to address one of the primary problems with the actual measurement device itself – correct calibration. For measurement instruments, it’s often easy to get any old signal, but very hard to get a very accurate signal. In our case, we want to measure very very small levels of fluorescence (in our case fluorescence is infra-red (IR) light in the 700 – 800nm range) which is emitted by the plant. However, there’s lots of IR light coming from light, sun, and reflections in the room – so we have to filter that out to drill down the signal coming from the plant.
Robert came up with a clever way of getting rid of all of the static IR light from the sun or lights in the room, but we still had some left over which was coming from the very LEDs we were using to stimulate the plant with. In our large scale bench units, this is most often done with a series of optical filters which removes any non-desirable wavelengths from the measuring LED. But we don’t really want those in a tiny handheld unit – too expensive, too big, too easy to misplace. After looking through about 30 different LEDs, we finally identified a few which produced much much less IR – this significantly reduced the impact of a bad calibration and was a big improvement.
But that wasn’t enough 🙂 We really want to use this tool even on very very weak or dilute samples, like Ben’s algal bioreactors. In this case, the sample being tested could be a very dilute solution of algae, which produces a really weak signal. With signals that small, even a very small amount of interfering IR can overwhelm the signal. In our tests, even with our fancy new low IR LED, the interfering IR was 2 – 3x larger than the signal itself. We even tried using lasers, which have very very specific emissions, with the hope you could eliminate this IR all together, but no dice.
Soooo…. we decided to pursue a software solution. We decided to flash the plants with a very very weak 800nm light, which doesn’t cause them to fluoresce at all. BUT, our de
tector can measure the amount of the 800nm light reflected from the sample. Then, assuming that the IR generated by our measuring LEDs produces a similar level of reflection off the plant, we can estimate the interfering IR and remove it. Each sample will reflect IR differently, so by measuring every sample’s reflection using our 800nm light we can calibrate each sample as accurately as possible. Initial tests showed that the results were in fact more accurate than previously calibrations or no calibration at all. Yay! You can find that research report here (IR calibration method.odt is the openoffice text file) which goes into more detail.
Now we’re ready to dig into the Teensy and move away from the big and bulky LED and detector controller systems we currently use. More on that in the next post!