The first requirement is easily achieved using readily available published solar tracking algorithms. The processing requirement for this activity is minimal.
The second requirement can be achieved using two light sensors in known positions (relative to the targets) and an algorithm that scans the mirror while looking for light reflected to the sensors. Once the angles needed to direct light at the sensors are known (as well as the local time) the angles for any other target can be calculated.
The third requirement simply requires sufficient memory to track which mirror is set to which target.
The desirable feedback arrangement is possible simply by controlling how many mirrors are allocated to the target in question dynamically according to sensors on the target.
The calibration and feedback adjustment are behaviours that can only be achieved with a rapidly responding directable mirror. Conventional designs are not capable of being moved much faster than is required to track the sun. With such devices calibration involves precise measurement of the installed position and alignment of each mirror. The directable mirror array described here can be rapidly positioned and thus could be traversed through a calibration sequence in a few seconds.
For the actual control electronics a two level design is appropriate with dedicated control systems (single chip controller based) for a number of mirrors providing standardised behaviours (calibration sweep, set target, track sun etc.). The installation as a whole is controlled from a standard PC using a simple interface to the control boxes. This arrangement should be adequate for any size installation.
This is by no means the only method with which it is possible to control an array of directable mirrors many other designs are possible. This design is a cheap and effective one that illustrates the possibilities inherent in these directable mirrors.
Australian patent no. 722115, US Patent no. 6227673, Patent pending Europe
How it works
Control System Design
Estimated Manufacturing Costs