

Lessons learned:
* The BlueSMiRF header on the shield is NOT connected to the UART (D0, D1), but to pins D2, D3. I had to solder my own connector to use the Bluetooth module as a "wireless serial cable". Boo.
* When sending a lot of data (OSC messages at ~100Hz), the BlueSMiRF chokes and is not discoverable from the PC anymore. Putting in lots of delays helped, but made the sensor reporting rather sluggish. Should probably not send anything out until a secure serial connection is established in the first place - maybe use the free microswitch on the shield to start/stop sending?


I just finished building a galvanic skin response (GSR) sensor for two students who want to use it for a gaming study. The sensor measures the varying resistance of human skin, which is suppsedly a proxy for engagement in a game. I guess we'll have to wait for the results of their study to see how well this works.
The sensor has two probes that you strap around two fingers; a black plastic Fry's box where all the electronics live; and a USB connection to send sensor data to a PC.

Since you don't want to send a lot of current through your body, the sensor circuit is built around an op amp (LMC6442 single supply rail-to-rail), based on the schematics published by Michael Sung and Vadim Gerasimov here. I didn't have a 100k pot handy so I approximated with 3x20k pots and a 10k fixed resistor in series. Built on one of the super handy One Pas prototype boards. Output here is an analog voltage in the 0-5V range. This output is hooked to to and ADC port of a small d.tools board, which in turn sends data to a PC, but you can substitute your favorite PC interface here easily (Arduino, Phidgets, ...). On the PC, I graph results in Exemplar.

The finger probes were inspired by the Lego GSR Sensor page. I tried three types of probes: 3/4" copper foil tape on velcro (too thick, green fingers); 1/4" aluminum foil tape on velcro (glue not conductive); thumbtacks on velcro (preferred solution)
