Multibeam Phased-Arrays Using Dual-Vector Distributed Beamforming: Architecture Overview and 28 GHz Transceiver Prototypes

Abstract

This article presents a dual-vector distributed beamformer architecture that employs a series-feed network and is capable of supporting up to four simultaneous beams. The multibeam array uses scalar functions within each front end to create Cartesian-weighted signals needed for phase shifting. A dual-vector series-feed network combines/distributes these signals for the receiver/transmitter whereas a global quadrature interpolator is used to create two conjugate beams. By using an interpolator on either end of the series feed, a total of four beams can be obtained. Among these four beams, a first beam can be controlled independently, a second beam is formed as an image of the first, a third beam is offset from the first based on the amount of phase shift within the series-feed structure, and a fourth beam is an image of the offset beam. In this work, the theory of operation of the series-feed DVDB is presented and then two different four-element 28 GHz DVDB transceiver array prototypes in 120 nm SiGe BiCMOS technology are described. One uses a hybrid coupler for global interpolation at radio frequency (RF) and the other uses quadrature mixers for global interpolation at baseband. Measurement results for the array employing passive interpolation at RF show excellent phase-shifting performance, including < 1 dB root-mean-squared (RMS) gain error, < 2 degree RMS phase error, 24% 3 dB bandwidth, with 16–18.6 dBm saturated output power in transmit mode and 4.9–7.3 dB noise figure in receive mode. Measurement results for the array employing mixer-based interpolation likewise show excellent phase-shifting performance with similar RMS gain and phase errors and slightly degraded overall RF performance. Comparing the two, we conclude that the DVDB with passive interpolation at RF is better suited for partitioned systems where beamformers and transceivers are realized on separate chips to support larger, scalable arrays. In contrast, the DVDB with mixer-based interpolation is better suited for integrated systems where beamformers and frequency translation functions must be integrated together.