D-band waveguide probe design, LNA fabrication, and radar front-end analysis

Frequency modulated continuous wave (FMCW) electromagnetic radars offer many advantages for object detection and ranging. Their robustness in poor conditions (rain, snow, darkness) makes them an interesting solution for automotive and military applications compared to ultrasonic sensing or visible light camera sensing. FMCW radars transmit a signal that is reflected by the targets and radiated back. The generation and processing of such signals require elements like Power amplifier (PA), mixers, filters, Low-Noise Amplifier (LNA). FMCW radars currently used in the automotive industry for collision avoidance and Adaptive Cruise Control (ACC) operate over specific frequency ranges. Due to spectrum regulations and standards developed by the ETSI and ITU-R, the 24 GHz unlicensed Ultra Wide Band (UWB), originally used in automotive radar, is no longer available as of January 1 2022; only part of this band remains available for narrowband applications. This lack of wide bandwidth in the 24 GHz band coupled with need for higher performance in radar applications makes the 24 GHz band unattractive for new radar applications [1]. One key benefit of the 77 GHz band is the wide bandwidth available in that frequency. This made the industry shift towards 77 GHz radar applications. Large bandwidth drastically improves the range resolution and range accuracy. Academia and key industry players have identified the D-Band, ranging from 110 GHz to 170 GHz, as a candidate frequency band for beyond 5G and 6G mobile communications as well as for future automotive radar applications [2]. This leads us to developing a LNA for a 140 GHz radar front-end module. This work focuses on the design and packaging of a D-band (110 GHz - 170 GHz) LNA. Especially, we design the waveguide to microstrip transition, analyze how the given low noise amplifier chip can be bonded to the microstrip and design the cavity that hosts the chip. The packaged LNA we have developed achieves a gain of 16 dB over the entire D-band with a noise figure of less than 4 dB. To interface with the LNA’s coplanar waveguide connectors, we designed a waveguide to microstrip transition (probe) and we used ribbon bonding. The designed probe achieves an S11 of less than -10 dB through the D-band with -15 dB for most of the band. The losses, represented by S21 are on average -1 dB. We conducted a study to understand the impact of the shapes of the probes by comparing a rectangular probe with a trapezoidal probe. We modelled wire and ribbon bonds and compared to the experimental results. We then carried a study of the resonances in the cavity hosting the LNA chip and designed elements to reduce them. Under these conditions, at frequencies above 100 GHz, this study shows that the optimal choices are a trapeze-shaped probe with ribbon bonds. Indeed the shape presents a better wideband performance than a rectangular-shaped and the ribbon transmission is superior to a wire bond transmission. We then designed the cavity that hosts the system and included walls to reduce resonating modes. A second part of this masters thesis draws the theory necessary to estimate the performances of a FMCW radar. We show the maximum range that can be achieved at 140 GHz, with a range resolution of 2.5 mm or 14 mm depending on whether the available bandwidth covers the whole 60 GHz D-band or a limited 10 GHz spectrum.