Large format heterodyne arrays for terahertz astronomy

For future ground, airborne and space based single aperture telescopes, multipixel heterodyne imaging arrays are necessary to take full advantage of platform lifetime, and facilitate science requiring wide field spectral line imaging. A first generation of heterodyne arrays with ~10 pixels has already been constructed, i.e. CHAMP, SMART, HERA, DesertStar, PoleStar and HARP. Our group is now constructing SuperCam, a 64 pixel heterodyne array for operation in the 350 GHz atmospheric window. This instrument will realize another order of magnitude increase in array pixel count. Several new techniques were used for SuperCam to maximize integration and modularity. Unlike other SIS array receivers, SuperCam is built around 8 pixel linear mixer modules, rather than independent mixer blocks. These modules house 8 single ended waveguide mixers with SOI substrate SIS devices. Each device is tab bonded to a MMIC based LNA. These modules dissipate only 8 mW of heat, while still maintaining 5 K IF noise temperature and 32 dB gain. Blind mate IF and DC connectors allow each module to be inserted in or removed from the focal plane as a unit. The modules are machined using a state-of-the-art CNC micromilling machine acquired specifically for this project. IF signals are processed by 8 channel IF downconverter boards, which provide gain, baseband downconversion and IF total power monitoring. A real-time FFT spectrometer implemented with high speed ADCs and Xilinx 4 FPGAs produce spectra of the central 250 MHz of each channel at 0.25 km/s spectral resolution. For arrays with an additional order of magnitude increase in pixel count, several additional technical problems must be overcome. Kilopixel arrays will require advances in device fabrication, cryogenics, micromachining, IF processing and spectrometers. In addition, seemingly straightforward receiver systems will require new approaches to realize a kilopixel heterodyne array with manageable complexity and cost. Wire count and 4K heat load must all be reduced significantly compared to SuperCam. IF and DC cabling and interconnects may be replaced with multiconductor microstrip or stripline ribbon. Parallel biasing of LNAs, magnets and even SIS devices is feasible if device uniformity is good enough. IF processing will require further integration, possibly with integrated MMIC chips containing all parts of a IF downconversion chain. Continued advances in FFT spectrometers could allow processing many hundreds of gigahertz of IF bandwidth for a realizable cost. We present results from final SuperCam receiver integration and testing, and concepts for expanding heterodyne arrays to kilopixel scales in the future.