Spin torque oscillators (STOs) are microwave oscillators with an attractive blend of features, including a more-than-octave tunability, GHz operating frequencies, nanoscale size, nanosecond switching speed and full compatibility with CMOS technology. Over the past decade, STOs' physical phenomena have been explored to a greater extent, their performance has been further improved, and STOs have already shown great potential for a wide range of applications, from microwave sources and detectors to neuromorphic computing. This thesis is devoted to promoting the STO technology towards its applications, by means of implementing the STO's electrical model, dedicated CMOS integrated circuits (ICs), and STO-CMOS IC integration. An electrical model, which can capture magnetic tunnel junction (MTJ) STO's characteristics, while enabling system- and circuit-level designs and performance evaluations, is of great importance for the development of MTJ STO-based applications. A comprehensive and compact analytical model, which is based on macrospin approximations and can fulfill the aforementioned requirements, is proposed. This model is fully implemented in Verilog-A, and can be used for efficient simulations of various MTJ STOs. Moreover, an accurate phase noise generation approach, which ensures a reliable model, is proposed and successfully used in the Verilog-A model implementation. The model is experimentally validated by three different MTJ STOs under different bias conditions. CMOS circuits, which can enhance the limited output power of MTJ STOs to levels that are required in different applications, are proposed, implemented and tested. A novel balun-low noise amplifier (LNA), which can offer sufficient gain, bandwidth and linearity for MTJ STO-based magnetic field sensing applications, is proposed. Additionally, a wideband amplifier, which can be connected to an MTJ STO to form a highly-tunable microwave oscillator in a phase-locked loop (PLL), is also proposed. The measurement results demonstrate that the proposed circuits can be used to develop MTJ STO-based magnetic field sensing and microwave source applications. The investigation of possible STO-CMOS IC integration approaches demonstrates that the wire-bonding-based integration is the most suitable approach. Therefore, a giant magnetoresistance (GMR) STO is integrated with its dedicated CMOS IC, which provides the necessary functions, using the wire-bonding-based approach. The RF characterization of the integrated GMR STO-CMOS IC system under different magnetic fields and DC currents shows that such an integration can eliminate wave reflections. These findings open the possibility of using GMR STOs in magnetic field sensing and microwave source applications. / <p>QC 20151112</p>

Les travaux de cette thèse s'inscrivent dans un projet de grande envergure, le projet 3D-IDEAS, financé par l'ANR. Le but de ce projet est d'établir la chaîne complète de l'intégration de circuits en technologie 3D. Les densités de puissance dans ces circuits sont telles que les problèmes liés à la température - électromigration, désappariement des courants et tensions de polarisation, etc. - sont susceptibles de remettre en cause la conception du circuit. Le coût élevé de la fabrication de ces circuits oblige le concepteur à valider le comportement électrothermique des circuits préalablement à l'envoi en fabrication. Pour répondre à ce besoin, un simulateur électrothermique précis et fiable doit être à disposition. En outre, en raison de la complexité extrême de ces circuits, il est judicieux que ce simulateur soit compatible avec l'approche de modélisation haut niveau. L'objectif de cette thèse est de développer un tel simulateur. La solution proposée intègre ce simulateur dans un environnement de développement CAO pour circuit intégré standard, Cadence®. La contrainte sur la précision des résultats nous a amené à développer une nouvelle méthodologie spécifique à la modélisation électrothermique haut-niveau. Ce manuscrit comporte deux grandes parties. Dans la première, la démarche adoptée pour concevoir le simulateur est détaillée. Ensuite, dans la seconde partie, le fonctionnement du simulateur ainsi que la méthode de modélisation haut-niveau mise en place sont présentées, puis validées.

The field of FPGA design is ever-growing due to costs being lower than that of ASICs, as well as the time and cost of development. Creating programs to run on them is equally important as developing the devices themselves. Utilizing the increase in performance over software, as well as the ease of reprogramming the device, has led to complex concepts and algorithms that would otherwise be very time-consuming when implemented on software. One such focus has been towards a search and optimization algorithm called the genetic algorithm. The proposed approach is to take an existing application of the genetic algorithm on an FPGA, developed by Fernando et al. [1], and create several instances of it to make a parallel genetic algorithm engine. The genetic algorithm cores are interfaced with a controller module that will control the flow of data between them to implement the parallel execution. Both coarse-grained and fine-grained parallelism are tested and results collected to find the best performance when compared to the single core design. Initial experimental results show some improvement over the number of generations required to reach the optimal fitness level, as well as more significant improvement for the number of generations needed for the average fitness to reach the optimal level.

Long-term hot-carrier induced degradation of MOS devices has become more severe as the device size continues to scale down to submicron range. In our work, a simple yet effective method has been developed to provide the degradation laws with a better predictability. The method can be easily augmented into any of the existing degradation laws without requiring additional algorithm. With more accurate extrapolation method, we present a direct and accurate approach to modeling empirically the 0.18-ìm MOS reliability, which can predict the MOS lifetime as a function of drain voltage and channel length. With the further study on physical mechanism of MOS device degradation, experimental results indicated that the widely used power-law model for lifetime estimation is inaccurate for deep submicron devices. A better lifetime prediction method is proposed for the deep-submicron devices. We also develop a Spice-like reliability model for advanced radio frequency RF MOS devices and implement our reliability model into SpectreRF circuit simulator via Verilog-A HDL (Hardware Description Language). This RF reliability model can be conveniently used to simulate RF circuit performance degradation

This research addresses problems in designing analog and mixed-signal (AMS) systems by bridging the gap between system-level and circuit-level simulation by making simulations fast like system-level and accurate like circuit-level. The tools proposed include metamodel integrated Verilog-AMS based design exploration flows. The research involves design centering, metamodel generation flows for creating efficient behavioral models, and Verilog-AMS integration techniques for model realization. The core of the proposed solution is transistor-level and layout-level metamodeling and their incorporation in Verilog-AMS. Metamodeling is used to construct efficient and layout-accurate surrogate models for AMS system building blocks. Verilog-AMS, an AMS hardware description language, is employed to build surrogate model implementations that can be simulated with industrial standard simulators. The case-study circuits and systems include an operational amplifier (OP-AMP), a voltage-controlled oscillator (VCO), a charge-pump phase-locked loop (PLL), and a continuous-time delta-sigma modulator (DSM). The minimum and maximum error rates of the proposed OP-AMP model are 0.11 % and 2.86 %, respectively. The error rates for the PLL lock time and power estimation are 0.7 % and 3.0 %, respectively. The OP-AMP optimization using the proposed approach is ~17000× faster than the transistor-level model based approach. The optimization achieves a ~4× power reduction for the OP-AMP design. The PLL parasitic-aware optimization achieves a 10× speedup and a 147 µW power reduction. Thus the experimental results validate the effectiveness of the proposed solution.

Microgrids containing photovoltaic (PV) cells and wind power gain more and more interest. These microgrids may work in stand-alone mode ("islanding") or be conncted to the main grid. In both modes of operation, power quality must be monitored and controlled. This report focuses on microgrids and aims to implement a monitoring system based on FPGA. In the monitoring system, two applications can be achieved, firstly a PAS-MPPT algorithm in a DC-DC boost converter to improve the maximun power point tracking of a PV unit, and secondly a detection and switching system of the grid mode - stand-alone or connected to the main grid. Simulation results prove the Verilog programs in FPGA are suitable to be used in microgrids.

<p>This report is the outcome of a thesis work done at Linköpings University, campus Norrköping. The thesis work was part of the development of a RF transceiver chip for implantable medical applications. The development was done in cooperation with Zarlink Semiconductor AB, located in Järfälla, Stockholm.</p><p>The transceiver is divided into three main blocks, which are the wakeup block, the MAC block and the RF block. The wakeup block is always operating and is awaiting a wakeup request in the 2,45GHz ISM-band. The RF-block is operating in the 400MHz ISM-band and is powered up after wakeup The MAC is the controller of the whole chip. All three blocks in the transceiver structure should be integrated on the same chip, using TSMC 0,18µm process design kit for CMOS (Mixed Signal /RF).</p><p>The purpose of the thesis work was to develop the wakeup circuit for the transceiver. The main purpose was to develop the digital control logic in the circuitry, using RTL-coding (mainly VHDL) but the thesis work also included a system analysis of the whole wakeup block, including the front-end, for getting a better overview and understanding of the project.</p><p>A complete data packet or protocol for the wakeup message on 2,45GHz, is defined in the report and is one of the results of the project. The packet was developed continuously during progress in the project. Once the data packet was defined the incoming RF stage could be investigated. The final proposal to a complete system design for the wakeup block in the RF transceiver is also one of the outcomes of the project. The front-end consists mainly of a LNA, a simple detector and a special decoder. Since the total power consumption on the wakeup block was set to 200nA, this had to be taken under consideration continuously. There was an intention not to have an internal clock signal or oscillator available in the digital part (for keeping the power consumption down). The solution to this was a self-clocking method used on the incoming RF signal. A special decoder distinguishes the incoming RF signal concerning the burst lengths in time. The decoder consists of a RC net that is uploaded and then has an output of 1, if the burst length is long enough and vice versa.</p><p>When it was decided to use a LNA in the front-end, it was found that it could not be active continuously, because of the requirements on low power consumption. The solution to this was to use a strobe signal for the complete front-end, which activates it. This strobe signal was extracted in the digital logic. The strobe signal has a specific duty cycle, depending on the time factors in the detector and in the decoder in the front-end. The total strobing time is in the implemented solution 250µs every 0,5s.</p><p>The digital implementation of the control logic in the wakeupblock was made in VHDL (source code) and Verilog (testbenches). The source code was synthesized against the component library for the process 0,18µm from TSMC, which is a mixed/signal and RF process. The netlist from the synthesizing was stored as a Verilog file and simulated together with the testbenches using the simulator Verilog-XL. The results from the simulations were examined and reviewed in the program Simvison from Cadence. The result was then verified during a pre-layout review together with colleagues at Zarlink Semiconductor AB. During the implementation phase a Design report was written continuously and then used for the pre-layout review. Extracts (source code and testbench) from this document can be found as appendixes to the report.</p>

As predicted by Gordon Moore, the number of transistors on a chip has roughly doubled every two years. Microprocessors featuring over a billion transistors are no longer science fiction. For example, Intel’s Itanium 9000 series and Intel’s Xeon 7400 series of processors feature 1.7 and 1.9 billion transistors respectively. To keep up with the emerging needs of contemporary very large scale integration (VLSI) design, industrial hardware description languages (HDLs) like Verilog and VHDL must be significantly enhanced. This thesis pinpoints some of the main shortcomings of the latest Verilog standard (IEEE 1364-2005) and shows how to overcome them by extending the language in a backward compatible way. To be able to cope with more complex circuits, well-understood higher-level abstraction mechanisms are needed. Verilog is already equipped with promising generative constructs making it possible to concisely describe a family of circuits as a parameterized module; however these constructs suffer from two problems: First, their expressivity is limited and second, they are not adequately supported by current tools. For instance, there are no static guarantees about the properties of the description generated as a result of instantiating a generic description with particular parameter values. Addressing both problems while remaining backward compatible led us to select a statically typed two-level languages (STTL) formal framework. By formalizing a core subset of Verilog as an STTL, we were able to define a static type system capable of: 1) checking the realizability of a description, 2) detecting bus width mismatches and array bounds violations, and 3) providing parametric guarantees on the resources required to realize a generic description. The power of the chosen framework is once more demonstrated as it also allows us to enrich the language with a new set of constructs that are designed to be expanded away when instantiated. To experiment with these ideas we implemented VPP, a Verilog Preprocessor with a built-in type checker. VPP is an unobtrusive tool accepting extended Verilog descriptions but generating descriptions compatible with any tool compliant with the Verilog standard. Our experience throughout this research showed that STTLs present a particularly suitable framework to formalize and implement generative features of a language. / Rice University, National Science Foundation (NSF) SoD award 0439017, Intel Corporation, Semiconductor Research Corporation (SRC) Task ID 1403.001

This report is the outcome of a thesis work done at Linköpings University, campus Norrköping. The thesis work was part of the development of a RF transceiver chip for implantable medical applications. The development was done in cooperation with Zarlink Semiconductor AB, located in Järfälla, Stockholm. The transceiver is divided into three main blocks, which are the wakeup block, the MAC block and the RF block. The wakeup block is always operating and is awaiting a wakeup request in the 2,45GHz ISM-band. The RF-block is operating in the 400MHz ISM-band and is powered up after wakeup The MAC is the controller of the whole chip. All three blocks in the transceiver structure should be integrated on the same chip, using TSMC 0,18µm process design kit for CMOS (Mixed Signal /RF). The purpose of the thesis work was to develop the wakeup circuit for the transceiver. The main purpose was to develop the digital control logic in the circuitry, using RTL-coding (mainly VHDL) but the thesis work also included a system analysis of the whole wakeup block, including the front-end, for getting a better overview and understanding of the project. A complete data packet or protocol for the wakeup message on 2,45GHz, is defined in the report and is one of the results of the project. The packet was developed continuously during progress in the project. Once the data packet was defined the incoming RF stage could be investigated. The final proposal to a complete system design for the wakeup block in the RF transceiver is also one of the outcomes of the project. The front-end consists mainly of a LNA, a simple detector and a special decoder. Since the total power consumption on the wakeup block was set to 200nA, this had to be taken under consideration continuously. There was an intention not to have an internal clock signal or oscillator available in the digital part (for keeping the power consumption down). The solution to this was a self-clocking method used on the incoming RF signal. A special decoder distinguishes the incoming RF signal concerning the burst lengths in time. The decoder consists of a RC net that is uploaded and then has an output of 1, if the burst length is long enough and vice versa. When it was decided to use a LNA in the front-end, it was found that it could not be active continuously, because of the requirements on low power consumption. The solution to this was to use a strobe signal for the complete front-end, which activates it. This strobe signal was extracted in the digital logic. The strobe signal has a specific duty cycle, depending on the time factors in the detector and in the decoder in the front-end. The total strobing time is in the implemented solution 250µs every 0,5s. The digital implementation of the control logic in the wakeupblock was made in VHDL (source code) and Verilog (testbenches). The source code was synthesized against the component library for the process 0,18µm from TSMC, which is a mixed/signal and RF process. The netlist from the synthesizing was stored as a Verilog file and simulated together with the testbenches using the simulator Verilog-XL. The results from the simulations were examined and reviewed in the program Simvison from Cadence. The result was then verified during a pre-layout review together with colleagues at Zarlink Semiconductor AB. During the implementation phase a Design report was written continuously and then used for the pre-layout review. Extracts (source code and testbench) from this document can be found as appendixes to the report.

During the design and verification of the Hyperstone S5 flash memory controller, we developed a highly effective way to use the SystemVerilog direct programming interface (DPI) to integrate an instruction set simulator (ISS) and a software debugger in logic simulation. The processor simulation was performed by the ISS, while all other hardware components were simulated in the logic simulator. The ISS integration allowed us to filter many of the bus accesses out of the logic simulation, accelerating runtime drastically. The software debugger integration freed both hardware and software engineers to work in their chosen development environments. Other benefits of this approach include testing and integrating code earlier in the design cycle and more easily reproducing, in simulation, problems found in FPGA prototypes.