Larbi BOUKHEZAR

Abstract: The thesis work will focus on two fundamental aspects concerning the development of millimetre band phase shifters:
- The choice of the structure, passive, active
- The calibration of the phase and amplitude
Two types of millimetre-band phase shifters can be distinguished in the literature: active phase shifters, based on a vector sum principle, and passive phase shifters, based on the use of varactors and/or switches.
- Active phase shifters have been studied for several years in the millimetre band with some recent results in the lower millimetre band around 30 GHz for 5G applications. The ambition of the work proposed for the thesis goes far beyond the current state of the art, with on the one hand the development of active phase shifters in D-band, i.e. far beyond the bands explored in the literature, and in which few works have been published to date, and on the other hand the development of innovative passive circuits constituting active phase shifters based on the experience and skills developed within TIMA and STMicroelectronics in these frequency bands. Indeed, the implementation of the active approach relies on the use of passive functions, in particular power dividers, and depending on the architecture choices, also couplers and baluns. On these points, the recent work carried out in the laboratory, in particular within the thesis of Marc Margalef and the master's course of Larbi Boukhezar, which contributed to improving the concept of coupled line power dividers, will make it possible to improve the performance of active phase shifters with respect to the state of the art.
- Similarly, there is a vast literature on passive phase shifters, but mostly at frequencies below 100 GHz. Here again, the ambition will be to demonstrate the feasibility of passive phase shifters in D-band. This work will be a continuation of David Ouattara's thesis (CIFRE STMicroelectronics funding, TIMA academic direction) which will be defended in autumn 2022. During his thesis, David Ouattara worked on the development of a passive phase shifter operating at 120 GHz, based solely on passive blocks, with a hybrid approach using digital blocks being developed. The aim is to finalise the analysis of these blocks, in the light of the latest characterisations, and to propose improvements while aiming at higher frequencies in D-band. One of the challenges to be resolved will be linked to the low quality factor of the varactors in D-band.
In-situ calibration of the phase shifters, both in production and during operation, is essential to ensure reliable beam steering at very small aperture angles of a few degrees. Calibration must be done in phase, in order to point in the right direction, but also in amplitude, in order to limit the appearance of secondary lobes.
Several recent works have developed some original concepts, but in-situ calibration, i.e. within the chip without having to implement an extremely complex measurement system, remains a challenge in millimetre band.
In this context, the aim of this thesis is to implement a solution based on the Oscillation Based Test (OBT) technique. This work will be a continuation of Marc Margalef's thesis (funded by the TARANTO project, TIMA), which he defended in September 2020. During his thesis, Marc Margalef paved the way for an original and efficient calibration method, based on an 'Oscillation Based Test (OBT)' technique, by laying the first theoretical foundations for the calibration of a phase shifter with respect to a reference phase shifter. Several perspectives set out in Marc Margalef's thesis make it possible to consider the calibration of an entire phased array, by combining mixing techniques that make it possible to determine the relative phase between all the phase shifters of the phased array in a very precise and energy-efficient manner. The first step will be to finalise the theoretical demonstration of this concept, then to implement the circuits on silicon and thus demonstrate a proof of concept.
The final goal will be to be able to compare in a totally objective way the two passive and active approaches for the same operating frequency and the same technology, which has never been done to our knowledge, in particular on state-of-the-art D-band structures. This comparison will have to be as exhaustive as possible, and will focus in particular on (i) power consumption, (ii) linearity, (iii) silicon area, (iv) phase and amplitude errors, (v) robustness to technological variations ('corners'), (vi) compatibility with in-situ test and calibration methods.