Microwave Oscillator Based on Two Optimally Tuned YIG Filters

We report a novel X-band loop oscillator, whose resonator is formed from two optimally tuned commercial yttrium iron garnet (YIG) filters. We call this a <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\Lambda $</tex-math> </inline-formula>-resonator, owing to the characteristic <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\Lambda $</tex-math> </inline-formula>-shape of its amplitude transfer function (ATF). At offset frequencies greater than 70 Hz, oscillator phase noise is limited by the contribution from the amplifier, as described by Leeson’s model. The effective Q-factor of the <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\Lambda $</tex-math> </inline-formula>-resonator (2.6×10<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">3</sup>) is determined by the slope of its phase transfer function on resonance, which is directly proportional to the group delay of the filters. With off-the-shelf components, we achieve single-sideband (SSB) phase noise of −151 dBc/Hz at 1 MHz offset for an oscillation frequency of 9.2 GHz. This work points to the potential for improvements with custom built filters optimized for high group delay. The high selectivity of the <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\Lambda $</tex-math> </inline-formula>-resonator (made of two seventh-order YIG-filters) allowed a strong suppression of both phase and amplitude fluctuations of the transmitted signal at Fourier frequencies outside the resonator’s bandwidth. We also investigated relevant oscillator noise sources, including power-to-frequency conversion as a function of Fourier frequency, and use a novel, highly sensitive technique to characterize the noise.