Cristina Maurette-Blasini

This article presents the design and modeling of two distributed resistive downconverter monolithic microwave integrated circuits (MMICs), realized in 100- and 70-nm AlGaN/GaN high-electron-mobility transistor (HEMT) technologies operating up to D-band frequencies (110–170 GHz). A dedicated transistor model that describes both the 100-nm and the scaled 70-nm HEMTs has been developed and extracted and has been used for the design of high-frequency mixers. The model is based on the physics-based advanced SPICE model for GaN HEMTs (ASM-HEMT). The standard ASM-HEMT model has been adopted by including a more precise model of the influence of velocity saturation and channel-length modulation (CLM) for both drain current and intrinsic capacitances and a cross-coupling (CC) capacitances model, to improve low-voltage mixer simulations up to at least D-band frequencies. The proposed MMICs can be operated either in a drain-pumped or gate-pumped mode. At 9 dBm of local oscillator (LO) power and a fixed intermediate frequency (IF) of 200 MHz, the 70- and 100-nm gate-length drain-pumped MMICs have an average conversion gain (CG) of −8.7 dB and −9.3 dB over a wide radio frequency (RF) band of 50–150 GHz, respectively. In gate-pumped operation, the CG is approximately 2 dB lower. The 100-nm HEMT MMIC has an RF input power related −1-dB compression point ( $P_{\mathrm {-1\text {dB}}}$ ) of 3 dBm in the drain-pumped mode and the $P_{\mathrm {-1\text {dB}}}$ increases to 9 dBm in gate-pumped operation. The 70-nm HEMT MMIC, in comparison, achieves a $P_{\mathrm {-1\text {dB}}}$ of 2 and 10 dBm for drain and gate LO-driven operations, respectively.

This paper presents an ultra-wideband vector modulator using an attenuator-based amplitude modulation covering the frequency range from 100 to 300 GHz. The proposed MMIC achieves up to a 6-bit phase resolution. This work depicts the individual components design, the entire MMIC, architecture as well as calibration used that enhance the chip performance. The design is implemented using a 35 nm InGaAs metamorphic HEMT process and achieves an RMS phase error down to 1 $^\circ$ , and an RMS amplitude error of 0.5 dB after calibration. The chips, including RF and control pads, consume 1 × 1.25 mm $^{2}$ .