To figure out which sensors / switches will reliably trigger sounds from sixteen swinging pendulums on our proposed ‘Lightly Oscillating Linguistic Organ’, I decided to build a test bench. While doing so it occurred to me that like Professor Tanya Monro, I was building a ‘platform for sensing’ (albeit on a slightly smaller scale).
Here is what I did:
- Hung an old broomstick from a ladder
- Added a few bearings to make it swing better
- Added a six inch piece of orange PVC conduit to the top of the broomstick (via some home sprinkler system pieces I had lying around)
- Wired my sensors to servo leads (from my model planes) and stuck em to wooden pop-sticks (actually tongue depressors)
- Whacked a cardboard box on the top with arduino, battery and breadboard with holes to shove my sensors into
- Strapped an android phone running ‘AndroSensor’ to the bottom of the broomstick
- Carted my laptop out to the shed and collected a bunch of data via arduino serial monitor & pasted into a spreadsheet & from a .csv file that an android phone’s sensors was writing to
- Bunged the collected data into a few google spreadsheets for analysis by chart
Here is what I learnt:
- My $10 freetronics light sensor for android (that gives an analog output proportional to light intensity) mounted at the top is very reliable and copes with really sloppy swings.
- A $1 glass reed switch under a strong magnet on the bottom of the broomstick works well (once you make it stronger by sticking to a wooden tongue depressor and cover in clear epoxy and when you figure out how to ‘de-bounce’ the signal).
- My $10 hall effect sensor for android mounted at the top is easy to use & check but hard to position well for consistent results. You need to have a very smooth swinging pendulum and find the exact centre.
- The guys building the physical pendulum wave have expressed concern over perceived slowing due to the magnetic force on each pass. This should be measured & tested before giving up on the simple reed switch..
- The ‘false positive’ readings I was getting from the magnetic reed switch is actually a well known phenomonen called ‘bounce‘ which can be de-bounced in software or via some rather ingenious circuitry! [I’m a little bit chuffed because before I discovered those links, I simply looked at my data on a graph and hacked my code to ignore state changes for a few hundred milliseconds after the first].