Think about it. If the GPS device is stationary (or even being spun around), it will read the same lat/long and time regardless of the direction it is facing. Most need to go meters away from the prior point to determine a direction. But let's say with a calibrated limit switch to indicate a completed rotation to North (or some known azimuth angle), that a starting angle position could be determined. It then becomes a dead reconing/timing system, as there is a lot of error in the motor and mechanics slewing the panels "westward" due to daily impacts of temperature, wind, and wear in the "mechanics." I've tried this and it is easy to be off by 20-degrees over the course of the day, and day-to-day it varies too. A 2-axis digital compasses such as the HMC6352 ($35 USD) helps provide feedback to the processor of where the panels are pointing in realtime. These also are not perfect. That said, it's fun to try these things. I'm "mostly" in the sun sensor camp, but I see some value in the having some programmed control, such as to set the panels vertical during precipitation and horizontal in high winds, assuming you could sense those too. ;-)