Two types of sun sensors are typically encountered. The first type is the coarse sun sensor, which usually takes the form of a photodiode or photocell. The output of the coarse sun sensor is a current approximately proportional to the intensity of the light falling upon the photodiode. Since in low earth orbit the light intensity from the sun does not vary much over time while the satellite is not in shadow, the light intensity is a function of the 1-axis angle to the sun, and the response can be calibrated to obtain that angle. With three such angle outputs it is possible to construct a unit vector towards the sun. Therefore, a coarse sun sensor is typically placed on each of the 6 faces, such that 3 eyes are always exposed to sunlight. The accuracy is poorest when the incidence angle is near normal to a particular sensor. Moreover, other light sources like the Earth’s albedo also factor into the output of the coarse sun sensor, such that the error can be very great.

The intensity $I_j$ is given by

where $n_j$ is the unit vector in the direction of the eye’s outward normal and $s$ is the unit vector in the direction from the spacecraft to the Sun.

The second type of sun sensor is the fine sun sensor. The HT-SS200 we use can achieve 1-sigma accuracy of within 0.3 degrees, which allows sufficient accuracy for the SelfieSat mission. It gives two-axis incidence angles, which together with the sensor’s normal vector can on its own be used to get a unit vector from the satellite to the sun. However, the field-of-view is limited, at around 110 degrees. Moreover, only one or two fine sun sensors will be available, and they will not always have the sun in their FOV. Therefore, it is necessary to devise a way to get the sun into the view cone.

The orientation of the sun sensor relative to the sun can be estimated in several ways. The first way depends on the inertial and magnetometer data only. Consider the inertial+magnetic attitude estimate, which after an hour or so in shadow might have error of some 10 degrees (for example only). The rough position in space of the satellite is also known from onboard propagation. This will be a simple model like SGP4, but even the along-track error of some 10s of km after several days will not really be relevant to this calculation. The location of the sun is also known. The satellite can be commanded to rotate such that the normal vector of the fine sun sensor is aligned with the estimated sun-satellite unit vector. The sun should then be within the field of view.

Alternatively, the sun-satellite unit vector is calculated as above using inputs from coarse sun sensors. This has the advantage of not depending on potentially fickle onboard environmental and state modelling. A rotation is commanded in the same way.

Currently specifications for a coarse sun sensor do not exist. The datasheet for the fine sun sensor is restricted by NDA. If you need to know more, reach out for more information!