Hybrid CPV-T with combined electrical and thermal output is well suited to solar generation from fixed limited areas, such as on the roof of an industrial or commercial facility with need for heat. This application will become especially attractive once overall electrical conversion efficiency of 40% is reached, as is projected for REhnu CPV systems using multijunction cells of 50% efficiency, anticipated in a few years. We outline here a configuration of dish- based CPV trackers optimized for close packing on a flat roof in a triangular grid, with a mirror area-to-ground area ratio of 50%. When the geometry of shadowing averaged over a year is taken into account, 80% of all the sunlight that would strike the rooftop is directed into the receivers. Such an array on a given area of flat roof will generate more electrical energy than would be possible with conventional PV panels, even if covering the entire rooftop, because of silicon's relative inefficiency. For example, in Tucson, the annual average global flux of 5.7 kWh/m2/day on a horizontal surface covered with 22% silicon modules will yield 1.25 kWh/m2/day. We show that a CPV system collecting 80% of all the direct sunlight of 7.0 kWh/m2 and converting it with 40% efficiency will yield 2.24 kWh/m2/day of rooftop area, nearly twice as much4. Thermal power will double again the total energy yield. A dual axis CPV-T tracker designed specifically very close spacing has been built to carry a single dish mirror of the standard type used in REhnu's M-8 generator, described by Stalcup et al in these proceedings1,2. Sunlight is collected and focused by a single square paraboloidal mirror, 1.65 × 1.65 m with focal length of 1.5 m. For closest possible packing without mechanical interference, and for broad distribution of load on a rooftop, the mirror and receiver are mounted to a C-ring structure, configured such that the elevation and azimuth axes intersect at a virtual pivot, at the center of the sphere that just clears the receiver and the corners of the mirror. Initial tests of closed loop tracking show an accuracy of 0.03°rms under calm conditions, and 0.04°rms in 6 m/sec wind.