60GHz Beamforming Receiver Front-End Measurements

I. Introduction This paper describes measurements and some issues involved in characterization of a 60GHz front-end for phased array receivers. The chip is implemented in a CMOS process and consists of two low noise amplifiers (LNAs), two mixers, and a phase locked loop (PLL) featuring a quadrature voltage controlled oscillator (QVCO). The digital phase control of the front-end is implemented in the PLL [1]. The front-end performs a two-step frequency down conversion, first from 60GHz to 20GHz, and then from 20GHz to quadrature baseband. A. Beamforming mm-wave receiver characterization The implemented chip occupies an area of 1.4mm x 0.66mm and is shown in Figure 1(a). It was wire bonded to a PCB with external decoupling capacitors, shiftregister controller, and IF baluns. The RF input was supplied by a microprobe (Cascade Microtech 67GHz), with the measurement setup shown in Figure 1(b) and (c). A network analyzer, with four phase-coherent internal sources, was used to measure the input match (S11), gain, linearity, quadrature accuracy, and the front-end phase control. The main issue for the circuit characterization was to be able to generate phase-coherent signals in order to measure the front-end phase shift set by digital phase control in the PLL. This kind of measurement would be straightforward if the circuit would not include any mixer as it could be measured with a two port network analyzer without any special features. The conversion gain measurement of a mixer is also easy, but becomes complicated if also the phase is of interest as it is a function of the RF and LO signal phases. These phases must therefore be held constant relative to each other to avoid drift. A known solution is to use an external mixer where the phase drift is cancelled, but it would for this measurement require a 48x frequency multiplier since the reference signal to the PLL is at 1.25GHz. Hence, the use of the four port network analyzer with coherent internal sources. The phase control was measured with one source generating the 60.1GHz input signal, the second a 1.25GHz PLL reference signal, and the third a 100MHz tone. The phase difference between the 100MHz tone and the receiver IF output at 100MHz was measured versus the phase control word by doing 80 automated sweeps, using Labview, to suppress the effect of jitter and to improve measurement accuracy. The noise figure was measured with a high frequency noise source (NC5115) from Noisecom and a Rohde & Schwarz spectrum analyzer (FSU50).