A 200-GHz Modulable Transceiver With 35-dB Tx On–Off Isolation and 16-Gb/s Code Rate for MIMO Radar in 130-nm SiGe Process

This article describes a modulable 200-GHz transmitter (Tx) and receiver (Rx) chipset for multiple-input–multiple-output (MIMO) radar applications. To achieve signal modulation and orthogonality, a mode-switchable modulator and an on–off isolation enhanced power amplifier (PA) are incorporated in the Tx. Based on this architecture, a simulated switching isolation exceeding 40 dB is realized even at PA saturation output. In addition, a compact, high-performance distributed balun design is proposed and analyzed, optimizing balanced-to-unbalanced performance and impedance matching. Moreover, the chipset features a wideband folded-dipole on-chip antenna with a vertical chip-to-waveguide transition for broadband radiation. The Rx includes a noise-canceling low-noise amplifier (LNA) and an IQ down mixer with a compact local oscillator (LO) generation network. The Tx chip exhibits a 3-dB saturated power bandwidth of 52 GHz and a maximum output power of 10.5 dBm, and the Rx chip exhibits a 42-GHz, 3-dB small-signal bandwidth, 24-dB conversion gain (CG), and 12-dB SSB NF. Enhanced by the isolation architecture, the chipset achieves switching isolation better than 35 dB within 170–230 GHz. Integration of the folded-dipole on-chip antenna generates a maximum equivalent isotropically radiated power (EIRP) of 12 dBm across a 4-dB bandwidth from 180 to 232 GHz. This performance is further enhanced to 28 dBm with a chip-to-waveguide transition connecting a WR-05 horn antenna. The modulation abilities are demonstrated by a non-return-to-zero (NRZ) signal with a 16-Gb/s code rate. A stepped-frequency continuous-wave (SFCW) waveform is utilized for high-resolution radar detection and a range resolution less than 6 mm is well measured with a 40-GHz sweep bandwidth. Angular detection capabilities are validated with a 2×2 MIMO radar prototype, distinguishing two corner reflectors clearly via 2-D fast Fourier transform (2D FFT).