A 2–18-GHz 360° Phase Detector Based on Switched Dual Multipliers and Fixed Delay Lines

This article presents a 360° phase detector in the 2–18-GHz frequency band [160% relative bandwidth (RB)] that uses a fixed delay line as a phase shifter and a double multiplication, in-phase and out-of-phase, between the input signals. Initially, several classic 90° power splitter designs are analyzed under the new perspective that they can be used when integrated into this type of topology, where it is not necessary to comply with the quadrature condition. Simulations show increases of up to 600% of the RB, going from a typical value of 20% (9–11 GHz) in branchline coupler structures to 140% (3–17 GHz). It is slightly lower in those that use a Schiffman phase shifter (325%), where it goes from 40% (8–12 GHz) to 170% (1.5–18.5 GHz). This is followed by the design of the phase detector, with the main innovation being the use of a fixed delay line that can easily be implemented in other frequencies and technologies, which presents a theoretical phase shift of <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$90~\pm ~72^{\circ }$ </tex-math></inline-formula> between 2 and 18 GHz. The prototype is intended to show the possibilities of the design and the conceptual idea. In order to be part of a measurement array, the circuit has a cell-type topology. It features four switches that prevent the use of power dividers and line crossover, which facilitates its implementation in other frequencies and technologies but increases the measurement time. Phase errors derived from the measurement system and amplitude noise at the multiplier output are analyzed. The measurement system uses a procedure based on quasi-synchronized generators, which guarantees a measurement error of less than ±3°. The manufactured prototype shows a deviation of the quadrature of <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\beta = \pm 72^{\circ }$ </tex-math></inline-formula> in the 3–18 GHz range, but at the 2-GHz frequency, it displays a value of <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\beta = 73^{\circ }$ </tex-math></inline-formula>, which considerably increases the phase errors compared to noise in the output voltages of the multiplier. The errors stay below ±5° in the 4–18-GHz band when considering an amplitude variation of ±7.5 mV (±1.5 ADC voltage resolution) in the output voltages of the multiplier, yet they increase to ±7° y [<inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$+ 6.5^{\circ } - 10.5^{\circ }$ </tex-math></inline-formula>] at 3 and 2 GHz, respectively. The error decreases below ±1.5° between 4 and 17 GHz for a variation of ±2.5 mV (±0.5 ADC voltage resolution), increasing to ±2.3° at 18 GHz, ±3.3° at 3 GHz, and ±4.6° at 2 GHz.