Another idea I'm thinking of that would be a good use of the MCU would be an antenna rotator. Given what I've read so far, controlling 2 motors from the Nerdkit shouldn't be a problem. My system would have one motor for azimuth and one for elevation. The resulting system would be used to automatically point an antenna at an Amateur Radio satellite and follow it through it's path through the sky. Breaking this project down, the first challenge is how to use the Nerdkit and a motor to "point" to a specific azimuth given a known starting position.

Moving an antenna around azimuth and elevation seems fairly simple: just construct a mount that moves in both directions, attach the firmware to drive the mount in each direction, and wire in the motors to the controller. However, how would one keep track of the direction and elevation that the mount is pointing? With most equatorial mounts, you set the system up and align it to a known value (usually magnetic or true north and zero elevation using a level), then let the antenna rotator system know these values.

From there, the system would be responsible for moving the mounts to a desired position. But how would it determine that position? Initially I thought that the time it took to move the mount could be used. A bit of calibration would be involved at first, but then once a baseline set of values was established, it could be used again to position the direction, right? Well, it doesn't appear this solution would be very accurate. Will the motors always move the same amount in a given time? Probably not due to differences in temperature affecting speed, resistance from the antenna due to wind, age of the motor affecting speed, condition of the batteries driving the motor, etc.

The next thing I was thinking was to use a fixed reference of some sort, which would provide feedback to the controller. The first thing that came to mind was some sort of opto-electric system. My initial ponderings focused on a disk of some sort with scribed points for each degree of azimuth (either markings of sufficient contrast to be detectable, or holes through which a light source could be detected by a collector). Once initialized, the controller could start rotating the mount from a known azimuth and simply "count" how many degrees it has moved successfully. This seems to negate the issues that the time vs. motor speed method would have had.

Does anyone have any other ideas for determining position for this design?

Also, what would be the best way to integrate a detector as described above? I was thinking either a small LED and a phototransistor for the "hole through the disk" method, or perhaps painting white lines on a black disk and using some device to "see" when it passes a white line.

Any input/experience in this kind of design would be greatly appreciated! :)

Current C-band satellite dish systems use linear actuators. These linear actuators have a simple reed switch to signal to a receiver every rotation of the feed screw. The receiver memorizes its start location and uses the fact that polarity of the motor dictates direction of antenna movement and changes in switch state confirm and measure that movement. I hope you find this useful.

Right. Let me add a little to it. The receivers count the pulses from one position to the next and it remembers where each position is. Initially, it requires the operator to manually set the positions and then storing it in the system. Some times, they can get lost with the lower resolution counters but it only happens when they are changed a lot. Most receivers don't know if it is counting up or down but it assumes from the direction control as to which way the motor in the actuator is going.

I am no expert but there are a number of things that you may want to consider. The feedbacks from a linear straight line actuator like the one used on a C band dish are useful to return a pervious position. Using this type of mechanical linkage to track at a constant rate through the rotational range is another thing. As the actuator is moved, in or out, the mechanical linkage slowly but significantly changes in angle. That is to say that the angular rotation of the dish will not be linear at all, even if the changes in the actuator length are linear. You would require a calculation or a table to translate the change in actuator length into change angular direction that takes into account the current mechanical linkage angles. Secondly the total amount of possible rotation is quite limited with some types of linkages hence the portion of the sky that you could reach would be somewhat limited. I am also looking for a simpler solution to rotational position feedback. I have used an opto-electric RF reflective surface array and wheels printed with a graycode pattern. I have no trouble achieving 1 degree of rotational resolution but the system I made is not very robust and certainly not weather resistant. For those reasons I have considered using a graycode wheel made using holes instead of a printed pattern but this would require a considerable amount of machining and construction time. I am hoping to find a simpler inexpensive solution at a reasonable price.