Implemented an FFT plot. This plot updates once every refresh (i.e. once every time 1000 samples are taken). The frequency axis looks consistent with testing, but could be slightly off depending on how python handles the FFT function. Regardless, this should give us some good "quick-look" data.
Put in a try/except to convert serial data to integer. This catches the random bad serial data. Be nice to know the cause of it, but at least it doesn't crash the program now. It seems this only happens after the program has been exited once with a reupload of the arduino code. It doesn't happen frequently during a run though.
Increased the speed of sampling by taking (almost) every data point instead of just the freshest. Had some trouble with leftover data due to some blitting issue. Turning blitting off seems to have solved it, but not sure why. Also need to work on the calibration more - it's not quite right yet.
Changed the baud rate to 115200. The old baud rate of 9600 wasn't sufficient for the 3 value array. Also did a crude calibration on the accelerometer. Subtracted away "rest position" values and scaled to g forces. The V/g seems off. Spec sheet quotes a typical value of 800mV where I calculated 1.23 volts, so I may be doing something wrong.