Ultra-pure digital sideband separation at sub-millimeter wavelengths

Context. Deep spectral-line surveys in the mm and sub-mm range can detect thousands of lines per band uncovering the rich chemistry of molecular clouds, star forming regions and circumstellar envelopes, among others objects. The ability to study the faintest features of spectroscopic observation is, nevertheless, limited by a number of factors. The most important are the source complexity (line density), limited spectral resolution and insufficient sideband (image) rejection (SRR). Dual sideband (2SB) millimeter receivers separate upper and lower sideband rejecting the unwanted image by about 15 dB, but they are difficult to build and, until now, only feasible up to about 500 GHz (equivalent to ALMA Band 8). For example ALMA Bands 9 (602-720 GHz) and 10 (787-950 GHz) are currently double sideband (DSB) receivers. Aims: This article reports the implementation of an ALMA Band 9 2SB prototype receiver that makes use of a new technique called calibrated digital sideband separation. The new method promises to ease the manufacturing of 2SB receivers, dramatically increase sideband rejection and allow 2SB instruments at the high frequencies currently covered only by DSB or bolometric detectors. Methods: We made use of a Field Programmable Gate Array (FPGA) and fast analog-to-digital converters (ADCs) to measure and calibrate the receiver's front end phase and amplitude imbalances to achieve sideband separation beyond the possibilities of purely analog receivers. The technique could in principle allow the operation of 2SB receivers even when only imbalanced front ends can be built, particularly at very high frequencies. Results: This digital 2SB receiver shows an average sideband rejection of 45.9 dB while small portions of the band drop below 40 dB. The performance is 27 dB (a factor of 500) better than the average performance of the proof-of-concept Band 9 purely-analog 2SB prototype receiver developed by SRON. Conclusions: We demonstrate that this technique has the potential of implementing 2SB receivers at frequencies where no such instruments exists, as well as improving the image rejection of current millimeter 2SB receivers to a level where sideband contamination is so low that would become negligible for any known astronomical source.