Fabrication, characterization, and error mitigation of non-flat sun sensor

We report the design, fabrication and error analysis of a sun sensor array composed of six photodiodes. The sensor estimates the direction of the sun using a linear least squares method. The performance of the sensor is deteriorated by three major sources: fabrication errors, scattered environmental light, and inexact modeling of photodiodes. Using a calibration procedure and modeling the uniform component of the environmental light, we mitigate the first two errors and significantly reduce root mean squared error from 2.63° to 0.83°. For a Field of View (FOV) of 110°, the maximum estimation error also drops from 3.8° to 1.6°. Through exact mathematical modeling of photodiodes, we demonstrate the feasibility of further reducing the root mean squared error, and discuss the effect of the maximum allowed angle of incidence on the sensor array performance. We propose a method to compute the optimum value of this angle and verify it experimentally. The effect of the number of illuminated photodiodes on the sensor accuracy is also investigated experimentally.