Accurate Measurement of Heliostat Reflector Shapes, Using Fully-Sampled Starlight Images

In the method described here, a bright star is used at night to measure the shape of heliostat reflectors in the field, at different elevation and azimuthal settings. The heliostats are oriented to direct starlight to the heliostat array receiver, and are then fixed in orietation while their aberrated and overlapping star images are translated across the receiver by Earth’s rotation. At the receiver, each of a fixed line of cameras forms an image of the heliostats to be measured. The cameras use large aperture lenses and are closely spaced along a zig-zag line, turned to be perpendicular to the translation direction of the star. Exposures are repeated while the entire image star passes by, so as to fully sample the blurred star images in two dimensions. In this way, every point on the reflector surfaces of the viewed heliostats appears in one or another of the recorded images, as bright from reflected starlight. From this data the amplitude and direction of the surface slope at each of these points is computed from Snell’s law, using the known positions of the star at the time of each exposure, of each camera and of each point on each heliostat. The surface shapes are then obtained by integration of the surface slopes. The total time needed to obtain a complete data set depends on the amount of surface error. For example, if the heliostats have peak-to-valley slope errors (from shape and pointing) of 4 mrad, a complete measurement of many exposures may be obtained in 1 minute, independent of heliostat field size, while the Earth rotates 4 mrad. The method will be particularly valuable for measuring heliostats whose reflector shape is actively altered throughout the day, so as to always form sharp images of the solar disc. But also for segmented reflectors of fixed shape, it will yield the detailed shapes and canting of the facets. The method has the advantage of being a null measurement – if there is no shape error, the star image is a point, and all the surfaces of all the heliostats appear in a single exposure of one camera. The method exploits modern, large CMOS imagers with very low read noise, and very fast (f/1.1), large format lenses to measure many heliostats at once.