Methods that improve high frequency Single Flux Quantum electronics design
Paul le Roux
Abstract
ENGLISH ABSTRACT: In this thesis, we cover various methods that improve high-frequency Single Flux Quantum electronics design. Accurately modelling high-frequency effects is vital in high-frequency Single Flux Quantum electronics design. The passive transmission line is, currently, the most inaccurately modelled circuit component. We, therefore, developed a new 2D quasiTransverse-Magnetic method to calculate transmission line properties of superconductor structures. The new method can take into account quasi-particle losses as well as dielectric losses. The frequency response of a passive transmission line with arbitrary length can then readily be obtained from the transmission line properties. The high-frequency response is informative, but a time-domain simulation model is necessary to determine the effect on high-frequency circuit dynamics. Accordingly, we developed the Superconductor Vector Fitting algorithm so that we can take arbitrary superconductor frequency responses and convert them into time-domain simulation models. Complex dynamical behaviour, made more complicated by the new numerical simulation models, currently prohibits designing circuits by only using design equations. Automatic optimization routines are used to adjust circuit parameters to improve fabrication yield. Existing routines suffer from issues ranging from suboptimal circuits to severe performance issues. We developed a new and novel yield optimization algorithm, Distance-To-Failure-Maximization, that can be used to optimize superconductor circuits that are verified u sing time-domain simulations. High-frequency circuits can be sensitive to changes in bias current from parasitics mutual inductances. We developed the fundamental inductance model to model all inductive effects accurately. The fundamental inductance model can also be used to construct design equations which accounts for parasitic effects as well as reduce the number of unknowns in the yield optimization procedure. We use all the above methods in a detailed case study on the design of a JTLT. We take into account attenuation, impedance mismatch, flux mismatch, and resonance during the design of the JTLT circuit. With this, we conclude that the methods we developed have brought significant improvements to high-frequency Single Flux Quantum design.