<p>This thesis investigates four D flip-flops with data transmission look-ahead circuits. Based on logical effort and power-delay products to resize all the transistor widths along the critical path in µm CMOS technology. The main goal is to verify and proof this kind of circuits can be used when the input data have low switching probabilities. From comparing the average energy consumption between the normal D flip-flops and D flip-flops with look-ahead circuits, D flip-flops with look-ahead circuits consume less power when the data switching activities are low.</p>

<p>The impressive evolution of modern high-performance microprocessors have resulted in chips with over one billion transistors as well as multi-GHz clock frequencies. As the silicon integrated circuit industry moves further into the nanometer regime, three of the main challenges to overcome in order for continuing CMOS technology scaling are; growing standby power dissipation, increasing variations in process parameters, and increasing power dissipation due to growing clock load and circuit complexity. This thesis addresses all three of these future scaling challenges with the overall focus on reducing the total clock-power for low-power, multi-GHz VLSI circuits.</p><p>Power-dissipation related to the clock generation and distribution is identified as the dominating contributor of the total active power dissipation. This makes novel power reduction techniques crucial in future VLSI design. This thesis describes a new energy-recovering clocking technique aimed at reducing the total chip clock-power. The proposed technique consumes 2.3x lower clock-power compared to conventional clocking at a clock frequency of 1.56 GHz.</p><p>Apart from increasing power dissipation due to leakage also the robustness constraints for circuits are impacted by the increasing leakage. To improve the leakage robustness for sub-90 nm low clock load dynamic flip-flops a novel keeper technique is proposed. The proposed keeper utilizes a scalable and simple leakage compensation technique. During any low frequency operation, the flip-flop is configured as a static flip-flop with increased functional robustness.</p><p>In order to compensate the impact of the increasingly large process variations on latches and flip-flops, a reconfigurable keeper technique is presented in this thesis. In contrast to the traditional design for worst-case process corners, a variable keeper circuit is utilized. The proposed reconfigurable keeper preserves the robustness of storage nodes across the process corners without degrading the overall chip performance.</p> / Report code: LiU-TEK-LIC-2006:21.

This thesis investigates four D flip-flops with data transmission look-ahead circuits. Based on logical effort and power-delay products to resize all the transistor widths along the critical path in µm CMOS technology. The main goal is to verify and proof this kind of circuits can be used when the input data have low switching probabilities. From comparing the average energy consumption between the normal D flip-flops and D flip-flops with look-ahead circuits, D flip-flops with look-ahead circuits consume less power when the data switching activities are low.

This thesis includes two topics. The first one is a low-power double-edge triggered flip-flop.The other is a orthogonal frequency division multiplex (OFDM) demodulator compliant with the Digital Video Broadcasting Handheld (DVB-H). Low-power double-edge triggered flip-flop (DETFF) is based on multi-Vth transistors technique. Since low threshold voltage transistors are able to generate large leakage current, they are suitable to drive big loads. By contrast, high threshold voltage transistors are more appropriate to latch data due to their low leakage. Therefore, a single latch double-edge triggered flip-flop utilizing multi-Vth transistors can be a low power and high speed design without paying the price of large area. The proposed OFDM demodulator is compliant with the DVB-H standard. The received DVB-H signal is processed by an RF front-end and the following analog-to-digital converter. Then, the digital signal is fed into the demodulator to adjust and calibrate the frequency, timing offset and channel estimation. The proposed DVB-H demodulator is mainly composed of five blocks : symbol timing synchronization block, carrier frequency offset compensation block, fast Fourier transform block, scatter pilot detection block and channel compensation block.

Microelectronic devices and systems have been extensively utilized in a variety of radiation environments, ranging from the low-earth orbit to the ground level. A high-energy particle from such an environment may cause voltage/current transients, thereby inducing Single Event Effect (SEE) errors in an Integrated Circuit (IC). Ever since the first SEE error was reported in 1975, this community has made tremendous progress in investigating the mechanisms of SEE and exploring radiation tolerant techniques. However, as the IC technology advances, the existing hardening techniques have been rendered less effective because of the reduced spacing and charge sharing between devices. The Semiconductor Industry Association (SIA) roadmap has identified radiation-induced soft errors as the major threat to the reliable operation of electronic systems in the future. In digital systems, hardening techniques of their core components, such as latches, logic, and clock network, need to be addressed. Two single event tolerant latch designs taking advantage of feedback transistors are presented and evaluated in both single event resilience and overhead. These feedback transistors are turned OFF in the hold mode, thereby yielding a very large resistance. This, in turn, results in a larger feedback delay and higher single event tolerance. On the other hand, these extra transistors are turned ON when the cell is in the write mode. As a result, no significant write delay is introduced. Both designs demonstrate higher upset threshold and lower cross-section when compared to the reference cells. Dynamic logic circuits have intrinsic single event issues in each stage of the operations. The worst case occurs when the output is evaluated logic high, where the pull-up networks are turned OFF. In this case, the circuit fails to recover the output by pulling the output up to the supply rail. A capacitor added to the feedback path increases the node capacitance of the output and the feedback delay, thereby increasing the single event critical charge. Another differential structure that has two differential inputs and outputs eliminates single event upset issues at the expense of an increased number of transistors. Clock networks in advanced technology nodes may cause significant errors in an IC as the devices are more sensitive to single event strikes. Clock mesh is a widely used clocking scheme in a digital system. It was fabricated in a 28nm technology and evaluated through the use of heavy ions and laser irradiation experiments. Superior resistance to radiation strikes was demonstrated during these tests. In addition to mitigating single event issues by using hardened designs, built-in current sensors can be used to detect single event induced currents in the n-well and, if implemented, subsequently execute fault correction actions. These sensors were simulated and fabricated in a 28nm CMOS process. Simulation, as well as, experimental results, substantiates the validity of this sensor design. This manifests itself as an alternative to existing hardening techniques. In conclusion, this work investigates single event effects in digital systems, especially those in deep-submicron or advanced technology nodes. New hardened latch, dynamic logic, clock, and current sensor designs have been presented and evaluated. Through the use of these designs, the single event tolerance of a digital system can be achieved at the expense of varying overhead in terms of area, power, and delay.

Power consumption is one of the most important design factors in modern electronic design. With a large market increase in portable battery-operated devices and push for environmental focus, it is of interest for the industry to decrease the power consumption of modern chips as much as possible. However, as circuits scale down in size the leakage current increases. This increases the static power consumption, and in future technologies the static power is expected to make up most of the overall power consumption. Power gating can decrease static power by isolating a circuit block from the power supply. In large chips, this requires state-retention flip flops and non-volatile memories in order to keep the circuit functioning continuously between power gating sequences. A design concept utilizing this is a Normally Off computer, which is in an off-state with no static power for the majority of the time. This is achieved by using non-volatile logic and memories. This concept has been realized by using a new semiconductor technology developed at Semiconductor Energy Laboratories Corporation Ltd., which is known as crystalline In-Ga-Zn oxide semiconductor material. This technology realizes transistors with an ultra-low off-state current, and enables several novel designs of state-retention circuits suitable for Normally-Off computers. This thesis presents two different architectures of state retention flip flops utilizing In-Ga-Zn oxide semiconductor transistors, which are produced and compared to determine their tradeoffs and effectiveness. These flip flops are then implemented in a 32-bit Normally-Off microprocessor to determine the performance of each implementation. This is evaluated by calculating the energy break-even time, which is the power gating time required to overcome the power overhead introduced by the state-retention flip flops. The resulting circuits and the work in this thesis has been presented at two conferences and submitted for publication in one scientific journal.

Dissertation (Ph.D.)--Auburn University, 2007. / Abstract. Vita. Includes bibliographic references (p.107-111).

Avec l'explosion du marché des applications portables et le paradigme de l'Internet des objets, la demande pour les circuits à très haute efficacité énergétique ne cesse de croître. Afin de repousser les limites de la loi de Moore, une nouvelle technologie est apparue très récemment dans les procédés industriels afin de remplacer la technologie en substrat massif ; elle est nommée fully-depleted silicon on insulator ou FDSOI. Dans les circuits numériques synchrones modernes, une grande portion de la consommation totale du circuit provient de l'arbre d'horloge, et en particulier son extrémité : les bascules. Dès lors, l'architecture adéquate de bascules est un choix crucial pour atteindre les contraintes de vitesse et d'énergie des applications basse-consommation. Après un large aperçu de l'état de l'art, les bascules à impulsion explicite sont reconnues les plus prometteuses pour les systèmes demandant une haute performance et une basse consommation. Cependant, cette architecture est pour l'instant fortement utilisée dans les circuits à haute performance et pratiquement absente des circuits à basse tension d'alimentation, principalement à cause de sa faible robustesse face aux variations.Dans ce travail, la conception d'architecture de bascule à impulsion explicite est étudiée dans le but d'améliorer la robustesse et l'efficacité énergétique. Un large panel d'architectures de bascule, avec les fonctions reset et scan, a été comparé dans le domaine énergie-délais, à haute et basse tension d'alimentation, grâce à une méthodologie de dimensionnement des transistors. Il a été montré que la technique dite de « back bias », l'un des principaux avantages de la technologie FDSOI, permettait des meilleures performances en énergie et délais que la méthodologie de dimensionnement. Ensuite, comme le générateur d'impulsion est la principale raison de dysfonctionnement, nous avons proposé une nouvelle architecture qui permet un très bon compromis entre robustesse à faible tension et consommation énergétique. Une topologie de bascule à impulsion explicite a été choisie pour être implémentée dans un banc de registres et, comparé aux bascules maître-esclave, elle présente une plus grande vitesse, une plus faible consommation énergétique et une plus petite surface. / The explosion market of the mobile application and the paradigm of the Internet of Things lead to a huge demand for energy-efficient systems. To overcome the limit of Moore's law due to bulk technology, a new transistor technology has appeared recently in industrial process: the fully-depleted silicon on insulator, or FDSOI.In modern ASIC designs, a large portion of the total power consumption is due to the leaves of the clock tree: the flip-flops. Therefore, the appropriate flip-flop architecture is a major choice to reach the speed and energy constraints of mobile and ultra-low power applications. After a thorough overview of the literature, the explicit pulse-triggered flip-flop topology is pointed out as a very interesting flip-flop architecture for high-speed and low-power systems. However, it is today only used in high-performances circuits mainly because of its poor robustness at ultra-low voltage.In this work, explicit pulse-triggered flip-flops architecture design is developed and studied in order to improve their robustness and their energy-efficiency. A large comparison of resettable and scannable latch architecture is performed in the energy-delay domain by modifying the sizing of the transistors, both at nominal and ultra-low voltage. Then, it is shown that the back biasing technique allowed by the FDSOI technology provides better energy and delay performances than the sizing methodology. As the pulse generator is the main cause of functional failure, we proposed a new architecture which provides both a good robustness at ultra-low voltage and an energy efficiency. A selected topology of explicit pulse-triggered flip-flop was implemented in a 16x32b register file which exhibits better speed, energy consumption and area performances than a version with master-slave flip-flops, mainly thanks to the sharing of the pulse generator over several latches.

Les peptides pénétrants (CPP) se caractérisent par deux propriétés : ils pénètrent dans l'espace intracellulaire et favorisent l'internalisation de cargaisons moléculaires auxquelles ils sont associés. Si les CPP sont très utilisés comme vecteurs en recherche fondamentale, la méconnaissance des mécanismes de pénétration et de leurs distributions intracellulaires limite leur utilisation thérapeutique. Il est admis que les CPP et leurs cargaisons sont internalisés par transport actif (endocytose) et par transport passif (translocation directe). J'ai étudié la translocation directe à l'échelle moléculaire en utilisant des membranes modèles. Les CPP usuels sont internalisés et permettent l'accumulation de cargaisons dans des vésicules unilamellaires. J'ai alors démontré que la translocation directe se déroule via la formation de complexes neutres et hydrophobes CPP-phospholipides.La pénétration intracellulaire des CPP est le plus souvent étudiée par microscopie confocale. J'ai démontré que des fortes concentrations locales de CPP induit une auto-inhibition de leur fluorophore. Cet artefact a conduit à des erreurs d'interprétation dans la littérature quant à la localisation des CPP. Un protocole permettant de révéler la fluorescence éteinte a été proposé et conduit à réévaluer la localisation subcellulaire des CPP ainsi que l'importance relative des mécanismes d'internalisation.Ces résultats ont permis de développer rationnellement de nouveaux vecteurs pénétrants : les oligoarginines acylées par des chaînes grasses dont des insaturations sont de stéréochimie cis. Leur internalisation passive particulièrement importante conduit à la libération de la cargaison dans le cytosol. / Cell penetrating peptides (CPPs) are short cationic sequences capable of shuttling bioactive cargoes into eukaryotic cells. If CPPs are common delivery tools in basic research, their therapeutic use is currently limited because their internalization mechanisms and intracellular distributions remain to be elucidated. In living cells there is evidence for both endocytosis of the CPPs and for “direct translocation”, an energy-independent uptake pathway. I analyzed the direct translocation phenomenon at the molecular level with model membranes. CPPs are internalized into large unilamellar vesicles and trigger the internalization of various cargoes. I then demonstrated that direct translocation occurs through membranes via the formation of a neutral and hydrophobic CPP-anionic phospholipids complex. CPPs internalization is mostly analyzed by confocal microscopy. I demonstrated that fluorescence self-quenching occurs if fluorescently labeled CPPs are locally too concentrated. This severe artifact led to misinterpretation of the subcellular distribution of CPPs. I developed a reliable procedure to avoid this artifact and ranked subcellular regions of living cells depending on their CPP concentration. As a result, I was able to rationalize the subcellular distribution of CPPs and to deduce their penetration mechanisms. The studies that I performed provided valuable information that I used to design a new family of delivery vectors: minimalist oligoarginines peptides acylated by unsaturated fatty acids (cis unsaturations). The direct translocation of these lipopeptides is particularly important yielding to an efficient delivery of a cargo inside the cytosol of living cells.

<p>Bit-serial multiplication with a fixed coefficient is commonly used in integrated circuits, such as digital filters and FFTs. These multiplications can be implemented using basic components such as adders, subtractors and D flip-flops. Multiplication with the same coefficient can be implemented in many ways, using different structures. Other studies in this area have focused on how to minimize the number of adders/subtractors, and often assumed that the cost for D flip-flops is neglectable. That simplification has been proved to be far too great, and further not at all necessary. In digital devices low power consumption is always desirable. How to attain this in bit-serial multipliers is a complex problem. </p><p>The aim of this thesis was to find a strategy on how to implement bit-serial multipliers with as low cost as possible. An important step was achieved by deriving formulas that can be used to calculate the carry switch probability in the adders/subtractors. It has also been established that it is possible to design a power model that can be applied to all possible structures of bit- serial multipliers.</p>