More Data!

 I don't have many responsibilities on Friday.  That means today I played around with the LabJack from my thermometer post.  Instead of messing with temperature readings, I played around with a photodiode and light sensing.  

 

This also let me try out my serial port on the DAQ.  The breakout card (thing with the rows of white labels) was purchased from Winford Engineering, a company in Auburn, MI of all places for.  For the non-Michigan readers, it is very close to both Saginaw and Midland.  It cost $35.  Most other sources were selling the same thing for between $50-100.  It was gratifying to know I can get these sorts of things from a small company right near the school.  In any case...

I managed to get the thing to detect me getting up, walking past the sensor, turning on the second overhead room light, and walking past the sensor again to sit down.  The sensor was aimed straight up, and I stayed about 2 feet away.

I am very pleased with this.  I played some games with averaging multiple readings and whatnot.  I can easily collect data every 20 ms (100 times a second), reduce that by a factor of 4 by averaging points and arrive at a noise level on the order of 0.5% of the signal.  I tried to collect data faster, but ran into what I think is the fluorescent light flicker causing more noise.  It would be interesting to see if I can observe that.

One issue I've had the past few years is making a reasonable light sensor for collecting data.  Yes, I can buy one for a few hundred dollars.  And, yes, we have some from PASCO that could be used.  But the PASCO ones are not great, are expensive, and require proprietary software.  In addition, the software must have some "secret sauce," as Chris calls it, cooked into it.  Much of the data we get from it is too clean and consistent.  When we modify an experiment using the same PASCO equipment but a slightly different way of collecting the data, we get more noise.  I don't understand what it's doing, which means I can't really troubleshoot it.  Nor can I feel fully confident in the data. 

By using the LabJack (or some other generic DAQ card), Python, and less expensive parts I can create more experiments and more flexible experiments.  Plus, I know what I'm doing to get the data.  I can make deliberate changes that "break" things so I understand the experiment.  Adam Savage once said on Mythbusters that the difference between science and screwing around is in science we write stuff down.  That is one component.  The other part of science is the deliberate "breaking" of the experiment or mathematical model so you know the limits.  Sun Tzu quoted the proverb "Know your enemy and know yourself and you need not fear the result of a hundred battles."  In science we would say "Know your biases and know your equipment and you will understand the experiment."

That this worked as well as it did with very little equipment is exceptionally encouraging.  It made me wonder if I could take an array of, say, 16 of these and use them as a crude pattern detection system.  Maybe as a model of how a CCD array (like what's in your digital camera) functions.  If that worked, it could be modified to show how rastering allows for more resolution at the expense of time.  It could be a nice student project.  What is really appealing about it is that it can be repeated by multiple students over time.