· · · Institutional Repository

There is a large demand for an online sensor to monitor the quality of lubricating oils. By constant monitoring the engine oil quality, a better understanding of oil condition can be found. The information on oil condition will lead to a much more accurate determination of proper oil change time, while it also provides increased insight into the actual state of the engine, and ultimately it will lead to more sustainable utilization of oils. A MEMS implementation of such a sensor is highly desirable due to its integration capability. Additionally reduced mass and size allow placing the MEMS device in places where a traditional system would not be able to fit. Microacoustic devices have been successfully used for various sensing applications where the sensitivity of the oscillating surface with respect to mass loading is utilized. These devices can be used for sensing liquid property for many applications. In this thesis work we aim at answering the following questions: · Which Microacoustic based sensor is able to resolve density and viscosity of a liquid? · Which Microacoustic based sensor has miniaturization capability? We choose shear mode resonators as the most suitable devices. Shear mode resonators can be used to measure the density and viscosity of contacting fluids separately. These devices are ideal for many sensing applications: they are small and inexpensive, very robust (can operate in high temperature, high vibration, and highly corrosive environments), have no macroscopic moving parts, have a good miniaturization capability, and can be configured to function as in situ fluid monitors. In contrast to other sensors, our solution allows to resolve the density and viscosity with help of one single corrugated device (in contrast to smooth surface devices). Shifts in resonance frequency and damping in an extremely rough resonator are related to the square root of the density-viscosity product and also to an offset term proportional to density. Experiments designed in order to support the analytical model have been carried out. The results obtained indicate that this is indeed an effective method for separate determination of liquid density and viscosity with shear wave mode resonators.

A small, low-cost, single-chip multi-band WCDMA/GSM transceiver solution is needed to address the high growth segment of 3G enabled smart phone ICs, while still keeping compatibility with widely deployed 2G GSM networks. In this thesis, an inductor-less LNA+mixer combination for multi-band WCDMA is presented. By designing the mixer for high IIP2, the inter-stage saw filter between LNA and mixer, as is normally used in WCDMA receiver designs has been eliminated, resulting in substantial cost reduction. Although mainly intended for WCDMA, the LNA+mixer combination also meets the requirements for GSM. For the LNA design, a noise cancellation topology has been used to achieve wide-band input matching, while still providing the required low noise figure. For this configuration, tradeoffs between noise, linearity and power consumption are explored in detail. Several modifications over basic noise cancellation topology are presented, which result in a significant reduction in noise figure. A passive mixer topology with a switching core followed by transimpedance amplifier (TIA) has been used for the mixer design. To reduce the noise contribution of the TIA, a new approach, which makes use of a current buffer between the switching core and the TIA is presented. This current buffer also allows the use of higher gate lengths and widths for the switching core transistors, which is beneficial in meeting the high IIP2 required to eliminate the inter-stage saw filter. The design of the transimpedance amplifier is also explained in detail. The LNA+mixer combination presented in this thesis achieves a worst case total gain of 35.5dB and noise figure of 2.9dB across all WCDMA bands. S11 is below -15dB from 870MHz to 2.17GHz. The IIP2 at transmit frequency offset is greater than 50dBm across all WCDMA bands. Across all these bands, the worst case IIP3 at transmit frequency offset is -8.2dBm and input referred 1dB cross compression point is -20dBm. The combination of LNA and I/Q mixer consumes 25mA from 1.2V supply. The design has been carried out using IBM 65nm CMOS process technology.

Thin film silicon solar cells are produced by using plasma deposition techniques. With this technique a thin film layer of silicon, most commonly amorphous silicon (a-Si: H), can be deposited on a substrate. The material properties depend largely on the plasma properties and the growth mechanism. The investigation of the correlation between material properties and the ion bombardment, which are important for electronic device operation, is the sole aim of this thesis. Expanding Thermal Plasma Chemical Vapour Deposition (ETP-CVD) is used for the deposition of a-Si: H layers. With this deposition technique, growth rates in excess of 1nm/s can be achieved. Through this, the production cost of solar cells can be reduced because of the increase in the production output. The growth mechanism of a-Si:H layers deposited with this technique has been extensively investigated and it has been shown that SiH3 radicals contribute to about 90% of the growth. The growth mechanism of a-Si: H layers can be manipulated by applying a bias to the substrate. In this way, the contribution of ions to the growth and the ion energy can be varied. To induce ion bombardment during film growth, a non-sinusoidal pulse-shaped signal is applied to the substrate and the material properties, which include optical properties, structural properties, and electrical properties, are measured. Denser a-Si:H films are obtained with increasing ion energy, indicating material densification. Higher growth rate is observed because of the production of more radicals around the substrate. An increase in photoconductivity is obtained up to 100V, which correlates with the decrease in the Tauc band gap and the densification of the material. Increase in the ion bombardment leads to decrease in the microstructure, which indicates reduction of voids in the material. However, defect density and Urbach energy increases with the ion bombardment due to the creation of weak bonds which end up as defects in the a-Si:H films. The optimized layers of the measured samples are incorporated into solar cells. Solar cells that are produced with pulse-shaped bias voltage up to 50V yielded a conversion efficiency of 6.2%. The efficiency is lower at higher voltages, which is mainly caused by smaller Voc and FF because of high defect densities present in the intrinsic layers at higher voltages.

Bit-error rate (BER) of comparators is becoming one of the limiting factors in the design of high speed ADCs. BER measurement setup is introduced and implemented in this thesis. Using this BER measurement setup gives us the opportunity to compare the BER of different comparators. It also enables us to study the effect of different parameters such as bias current, and power supply variations on the BER of these comparators. Capacitive based comparator is also proposed in this work which is a new topology for comparators and simulations show that it can perform better than the other conventional comparators with respect to BER. The capacitive based comparator and 2 conventional comparators are implemented in the BER measurement setup so that they can also be compared on silicon.

As we are moving toward nanometre technology, the variability in the circuit parameters and operating environment (Process, Voltage and Temperature (PVT)) are increasing, causing uncertainty in the circuit performance. Statistical Static Timing Analysis (SSTA) is a category of methodologies to analyse the variations in delay due to PVT variations. This thesis work is a part of the MODERN project, which is involved in developing a new SSTA methodology. In this thesis, the variation of the delay in 45nm standard cells is analysed. In industry practice, the Monte Carlo method is often used to estimate the statistical moments. This method needs a large number of simulation iterations and these simulations are parameter distribution dependent. A fast statistical moment estimation method is proposed in this work. The proposed methodology is at least 100 x faster than the Monte Carlo method and simulations are independent of the parameter distribution. In the SSTA methodology of the MODERN project, the signal waveforms with their variations are preserved at each pin of the standard cell. The concept of a "set of waveforms" as a representation of a variable electrical signal is also developed in this thesis work. Possible methods to represent the set of waveforms and their integration with the timing analysis methodology are analysed. The pseudo circuit based representation turns out to be the most compact model. A methodology for the analysis of the accuracy and efficiency of the pseudo circuit model is proposed.

This thesis reports on the design and modeling of decoupled and tunable bandwidth MEMS vibratory gyroscopes that were designed for the medical applications of tremor compensation and micro-surgical tool navigation. Two different designs are presented. The first design is a dual mass de-coupled gyroscope that consists of a drive and sense mass implemented in a drive frame architecture. The device achieves a theoretical maximum resolution of 0.01 deg/s and a maximum sensitivity of 5.247 ƒF/deg/s. However, this design is prone to quadrature error that drastically reduces the sensitivity of the device and completely destroys its performance. The second design is a three-mass doubly-decoupled gyroscope that consists of a drive, Coriolis and sense mass and is designed to overcome the sensitivity of the dual mass de-coupled gyroscope to quadrature error. The three-frame structure of this gyroscope, with an outer sensing frame, leads to an improved electrical sensitivity over conventional architectures. It leads to a full decoupling between the sense and drive modes that makes the device robust against quadrature error and eliminates the cross axis sensitivity that usually limits the angular rate sensing performance. The device achieves a theoretical maximum resolution of 0.006 deg/s and a maximum sensitivity of 1.255 ƒF/deg/s. Optimizing this design based on the DVA principle leads to a 60 % reduction in the size of the Coriolis mass, resulting in a 37 % increase in the resolution and an 19 % increase in the sensitivity. The drive and sense mode resonance frequencies of the designs are 2500 Hz and 2830 Hz respectively, with an input angular rate bandwidth tunable between 40Hz and 330Hz. Both COMSOL finite element simulations and macro models implemented in Agilent ADS were used to validate the structures.

In dit ontwerprapport wordt een systeem ontworpen dat via ledverlichting in een warenhuis elektronische prijskaartjes aanstuurt. De informatie die op het display van elk kaartje moet verschijnen, wordt onzichtbaar voor het oog gecodeerd in het uitgestuurde licht. Het ontwerp bestaat uit een enkele zender, namelijk de ledverlichting van een warenhuis, die uniform en simultaan dezelfde informatie uitzendt, en meervoudige ontvangers, namelijk alle prijskaartjes die informatie op het display afbeelden die uniek is per prijskaartje. In deze thesis wordt de zenderkant van het systeem ontworpen, die bestaat uit een centrale terminal vanwaar de data worden verstuurd naar een Power Line Carrier Modem (PLCM) die de data superponeert op het lichtnet. Bij de aansluitklemmen van de verlichting worden de data met een tweede PLCM van het lichtnet gehaald en gebruikt om een elektronische schakeling aan te sturen die de ledverlichting van stroom voorziet.

With the performance of single-core processors approaching its limits, an increased amount of research effort is focused on chip multiprocessors (CMP). However, existing lock-based synchronization methods that are critical to performing parallel computation possess limited scalability and are inherently complex to use while programming. This thesis uses the concept of transactional memory implemented within a synthesizable fabric named TMFab, containing all the requisite hardware components needed to prototype a scalable chip-multiprocessor. Its processor independent nature enables the instantiation and use of any suitable soft-processor core inside the fabric without significant modifications to the fabric hardware. Additionally, the fabric offers scalability on account of its 3D interconnect architecture that supports die-stacking to add additional processor cores to the CMP without increasing its area footprint. The hardware transactional memory system of the fabric reduces performance overheads of transactional operations, allowing transactions to complete execution faster. TMFab is shown to provide speed up as high as 3.44x for correctly partitioned independent transactions and can be used to analyze the points of contention for conflicting transactions. The fabric was synthesized for both Field Programmable Gate Array (FPGA) as well as 90nm semi-custom targets.

Wireless communication has encountered a tremendous growth over the past few decades. The increased plurality in communication standards, characterized by the use of different operating frequencies and data rates, has translated into very tough specifications for the broadcasting base station power amplifier, in terms of efficiency, bandwidth and linearity. For this reason, currently many high efficiency power amplifier concepts are investigated for their suitability to handle the upcoming generations of wireless communication standards. At this moment the Doherty power amplifier (DPA) is a popular concept, which gives good effciency in the power back-off, making it a suitable choice when dealing with signals that have a high peak-to-average power ratio. To be efficient in power back-off operation, the Doherty power amplifier is composed out of two linear amplifiers with an impedance inverter as an output power combiner. Due to its high complexity, the traditional Doherty amplifier is limited for its RF bandwidth. In view of this, the objective of this thesis is to design a linear wideband class-B power amplifier cell which allows incorporation in the DPA. For this purpose an LDMOS based push-pull topology together with baluns (implemented by bondwires) at the input and output of the transistors has been adapted. The described topology helps to achieve an orthogonal relation between the fundamental path and its second harmonic, resulting in wideband high efficiency operation. The resulting amplifier provides an output power of 60 W with a power added efficiency greater than 48% over a relative bandwidth of 40% centered around 1:8GHz.

The objective of this thesis is defining the feasibility of using an Inductive Wireless Power and Data transfer system (IWPD) as sounding rocket umbilical. An Umbilical is used to supply power and communicate with the rocket in the last moments before launch. The power is used to keep the batteries fully charged and the communication is used for arming and disarming procedures. An IWPD system uses magnetically coupled coils, or inductors, to transfer power and data. This principle is used in transformers where a magnetic core creates a high coupling. In the umbilical the coils cannot be placed around a single core. This comes from the fact that one coil is outside the rocket and one coil is inside the rocket. Coupling of two coils separated by air is obtained by the magnetic field they produce. The coupling is defined as the magnetic field shared by the two coils in reference to the field that only passes one of the coils. To determine the coupling a simplified model was made which allowed the coupling to be plotted to the distance between the coils. The power transfer between a pair of coupled coils is dependent on the coupling between the coils. The magnetic field will not dissipate energy unless it goes through conductive materials. This means that magnetic energy is stored in the field and can be reclaimed. Making the inductors resonate with the addition of capacitors keeps the energy in the circuit. This resonance makes the power transfer non linear to the coupling of the coils. Circuit calculations are presented to determine the transfer of the IWPD in respect to frequency, supply and load conditions. The transfer is then used to determine the best setup to be used for an IWPD umbilical. The resonance of the system is used as frequency response of an oscillator. This oscillator is used to overcome the frequency shift of the system. The proposed system is compared with a more conventional umbilical. This comparison is made in three fields. Performance, Interference and Safety. In Performance an IWPD is as suited for the task as conventional system. Both will be able of supplying sufficient power and date. In Interference the IWPD is much better. It separates the umbilical from the skin and structure of the rocket. For Safety the results are balanced. The conventional systems perform better on electromagnetic interference (EMI). While the IWPD performs better on Electrostatic discharge (ESD). The IWPD might perform less on EMI this however does not have to be a problem. Increasing the electromagnetic compatibility (EMC) of the electronics and the distance to dangerous materials in the rocket will create a safe working situation.

Nowadays most of the energy needed comes from fossil fuels such as natural gas (NG). Dwindling stocks of all fossil fuels, including locally occurring natural gas reservoirs, have spurred the effort to both decrease dependence on local fuels and switch to renewable, green energy. The drawback to gas sources like Subsitute Natural Gas (SNG), Biogas or Syngas is that they are too variable in composition to be readily mixed into the existing gas infrastructure. An electrical gas sensor with continuous data output needs to be developed for production safety, burner control and measuring the value of the gas. Photoacoustic infrared absorption (PAS) is a good measurement principle to measure the composition of gas. Three components are essential for a photoacoustic system: a modulated light source emitting light in the spectrum which the target gas absorbs, a pressure sensor, usually a microphone and a photoacoustic cell to hold the sample. In this cell the photoacoustic signal is generated. The PA-signal can be at a resonance frequency of the cell, in which case the cell functions as an acoustical amplifier. The parameter to describe this amplification is the Q-factor. The composition of gas from different sources creates a list of gases that are possibly included in the mixture. The components are classified in four categories: combustible fuel sources, hazardous gases, neutral gases and other gases. For each of these components, Non-dispersive infrared absorption (NDIR), Thermal conductivity detection (TCD) and PAS are considered as sensor principles. The function and requirements of the sensor have to be determined before starting the design and follow from the application of the sensor. This thesis proposes a six step design methodology for designing a PA-sensor. It is a structured approach which tries to divide the process in discrete steps. Because the acoustic behavior of the cell is very important, the resonance profile has to be determined. Analytically calculating the behavior of sound inside the cell is only possible for simple structures. Therefore two methods of acoustic profiling are considered: experimentally and by modelling the acoustic behaviour with the Finite Element Method (FEM) software tool COMSOL Multiphysics 4.1. Measurements were done on different sizes and shapes of resonance chambers in the laboratory. First measurements are performed on simple structures, which show clear longitudinal resonances. Later more complex structures are discussed as well. Measuring gives an good acoustic profile, but to create a large scale model sample chamber takes a lot of effort. Therefore, measuring is useful when verifying a design, but not during the design phase. FEM simulations do not have these drawbacks. Using COMSOL in the design process gives the possibility to do an eigen frequency analysis and a frequency domain analysis. Simulation results are compared with measurements to build confidence in the simulations. Three design cases are discussed. First choosing the position of source and microphone is discussed, then the complexity, meshing and simulation time of the model, and finally excitation by two sources. COMSOL is a good tool to help design the acoustical properties of the sample chamber.

Research on unmanned aerial vehicles (UAVs) has steadily increased over the last decades due to the wide range of military, civil and security applications. Within the larger framework of aerial vehicles, the high degree of autonomy of UAVs places constraints on data communication links. Many existing UAV systems include ground stations which allow users to retrieve data from the airborne unit during flight. Air-to-ground connections have been successfully implemented for both low data rates and high-capacity links. A broad variety of UAV applications therefore appears to be feasible. This thesis proposes a possible implementation of a graphical user interface and a ZigBee-based air-to ground communication link for sensor and graphical data transfer. This design is part of a ground vehicle tracking system for UAVs.

Bridge sensors are widely used for accurate measurement of physical quantities such as temperature, pressure, strain or altitude. Such sensors require a low-noise, high-resolution and accurate readout system with high input impedance. In order to meet these requirements, conventional sensor readout systems use multiple stages which typically include a low-noise preamplifier, an anti-aliasing filter and a discrete-time (DT) sigma-delta modulator (ΣΔM). As a result, these systems involve several high-gain loops with total open-loop gain far exceeding the required closed-loop gain. This can lead to sub-optimal power dissipation and greater analog design complexity in design of a sensor readout system. In recent years, Gm-C continuous-time (CT) ΣΔMs have attracted a lot of attention due to their inherent anti-alias filtering, low power dissipation, high input impedance and high resolution. However, their use in precision applications such as bridge sensor readout is limited by the nonlinearity of the input stage. In this work, a new single-bit CTΣΔM topology is proposed that employs an identical nonlinear element in the feedback path along with a low pass filter to enable nonlinearity compensation and achieve high linearity. A feedforward Gm stage further enhances the nonlinearity compensation by increasing the effective loop-gain. This approach enables more than 60 dB improvement in the nonlinearity of the input transconductor stage of the CTΣΔM. A precision sensor readout circuit using the proposed CTΣΔM architecture is designed and implemented in 0.7 µm technology. The modulator achieves a resolution of 20 bits with a 22 nV/√Hz noise floor and an accuracy better than 10 ppm in post-layout simulations. It consumes 240 µA current from a 5 V supply. The resolution and accuracy of the CTΣΔM designed in this work is comparable to that of state-of-the-art readout systems but with lower power dissipation and lesser analog complexity. The proposed modulators achieves 10x better linearity and accuracy compared to the state-of-the-art Gm-C based CTΣΔMs, albeit at low frequencies, with significantly less noise and power dissipation.

To ensure efficient and reliable operation of power amplifiers it is very important to precisely measure and control the power of the signal they transmit. A device most often used to do precise power measurements of RF systems is the so-called RF power detector. As this detector is used as measurement device the precise measurement ability of the RF power detector is very important. For the precise predictability of the output power of power amplifiers the transfer of the RF power detector has to be fixed and accurately known. In this thesis we develop a new calibration method for the transfer stabilization of logarithmic power detectors. Via thorough investigation at system level and circuit level it is shown that the proposed method can be used to continuously calibrate the transfer of a logarithmic RF power detector to a predefined and fixed position over mismatch, part-to-part spread, temperature and input frequency. The method depends on a novel switching algorithm around a log device that is capable to do continuous slope, intercept and dynamic range correction on the transfer of the logarithmic power detector. When accurate enough, the method would make calibration of each individual device unnecessary. Furthermore a new method is presented that can be used to extend the dynamic range of log detectors. System simulations show that the calibration method leads to the wanted transfer stabilization of the logarithmic power detector. A critical part of the new logarithmic transfer calibration method is the need of an accurate multiplication procedure at the input of the logarithmic device. For this accurate multiplication a new accurate gain fixation procedure for a non-linear high bandwidth gain stage was developed. A big part of the thesis is dedicated to the investigation and circuit implementation of this new accurate and fast gain fixation procedure. The gain stabilization method leads to the implementation of a new innovative gain fixation system, including several new architectural innovations. One of these innovations is the implementation of a new accurate ripple blocking system with relative small form factor and fast response time. Simulation results of the circuit implementation of the gain stabilization system prove that the accuracy of the gain stabilization of the non-linear high bandwidth gain stage is well within the required specifications.

A mechanical system is an excellent choice for studying forces of varying magnitude. The force produces a displacement of the system under study. By measuring this displacement electrically or optically, the force can characterized. In this thesis, the displacement of a nanoelectromechanical system (NEMS) due to thermal energy is of interest. At room temperature, the thermal motion of the system is the smallest amplitude of motion possible and this can be detected if the detector has a high sensitivity. We present a high frequency transmission line resonator operating in the GHz frequency regime as a detector to measure such small changes in motion. Graphene, an atomically thin membrane of carbon will be used in a capacitor geometry with the detector to measure its thermal motion. The motion of graphene is transduced into capacitance which in turn modulates the electrical resonance frequency of the detector. This non-linear effect produces intermodulation sidebands which contain the resonance signature of the graphene membrane which can be read out at thermal noise limited sensitivity. The thesis discusses the design of the detector from analytical modeling to electromagnetic simulations. Fabrication of the complete device is also discussed. Measurement results indicate that the detector has a Q factor of 43 at room temperature. The position and capacitance sensitivity of the detector is estimated to be 0.54 pm/√Hz and 0.58 zF/√Hz respectively at an amplifier noise floor of 1 nV/√Hz .

For neural cell stimulation there is a need of a high supply voltage up to 18V. However because the stimulation is not every time the same, sometimes there is only a need for lower supply voltage. The supply voltage should be variable so that loss of energy due to the wasted supply voltage headroom is minimized. In order to be able to cover this stimulation voltage range the converter needs to convert the 3.3V battery voltage up to a variable output voltage with a maximum voltage of 18V. This needs to be done with a power efficiency of at least 80%. Also the converter must not be sensitive to input/output variations and temperature changes. In this work the boost converter topology is selected and is designed with a new current sensing technique. Also the controller circuit is designed so that the converter works always in the discontinuous current mode. For this controller circuit an efficient capacitor based voltage measurement circuit is designed. Everything is designed at transistor level to be implemented in the 0,35ìm High Voltage I3T25 technology of On Semiconductor. The working principle is illustrated with circuit simulations.

3D is entering the world of Integrated Circuits. While interconnects have always been three-dimensional, the actual silicon inside an IC was essentially still planar. The introduction of the Through-Silicon Via changes that, by allowing connections to be made from one side of a die through its silicon substrate to the other side of the die, so that multiple dies can be interconnected efficiently inside a single package. Like any revolutionary development, TSVs require changes in the tools that deal with them. Not all current Electronic Design Automation software is able to extract chip layouts containing TSVs to a correct circuit. In order to adapt extraction software, a methodology to extract TSVs is required. This methodology, in turn, requires a model. Several models that can be found in literature are compared and contrasted. The one that is selected is improved upon by making some minor corrections. The model is also simplified and the conditions for validity of this simplification are shown. The resulting model is then used to implement an extraction methodology, both with a model-based approach and with a formula-based approach. Simulating the different dies that make up a 3D IC together is demonstrated, giving the designer access to a complete toolchain necessary for designing 3D ICs and verifying their correctness.

Attractive material properties of plasma enhanced chemical vapour deposited (PECVD) silicon carbide (SiC) when combined with CMOS-compatible low thermal budget processing provides an ideal technology platform for developing various microelectromechanical systems (MEMS) devices and merging them with integrated circuits. In this paper we present a generic surface micromachining technology developed using a stressoptimised PECVD SiC as the structural and encapsulation material for MEMS. An overview of selected MEMS applications realised, at DIMES Technology Center (DTC) of TU Delft, using the PECVD SiC surface micromachining technology is provided. Presented MEMS examples include—a pressure sensor, wafer-level thin-film packaging, RF switch and accelerometers. Potential applications for the presented technology include automotive, industrial and medical systems, where devices are often subjected to harsh environments.