Thesis (Ph. D.)--University of Wisconsin--Madison, 1981. / Typescript. Vita. eContent provider-neutral record in process. Description based on print version record. Includes bibliographical references (leaves 335-340).

No description available.

No description available.

The cathode and the anode fall of the DC arcs are measured by fast oscilloscope for Ag-CdO contacts over a range of gap-lengths from 0.05 mm to 1 mm, and currents of 4-10 Amps at atmospheric pressure, with a known electrode closing speed, using the Moving Electrode Method. It was observed that the anode fall can occur in a few places within the arc voltage waveform, and is dependent of the electrodes' surface condition. Both cathode and anode falls increase with gap-length and decrease with current. It was found that when arc length is shorter than electrode separation, discontinuity within the arc voltage waveform during closure is caused, in many cases, by vapourisation of the first point of contact or by a high electric field set up between the two electrodes. These discontinuities are named as Voltage Step Phenomena. These voltage steps are related to the cathode and anode fall voltages, and their regular occurrences are a function of surface roughness. The fluctuations in the arc voltage waveform are thought to originate mainly from the cathode. A technique has been developed to measure the temperature of the electrodes accurately by using a T-type thermocouple, 0.075 mm diameter, in conjunction with a DC amplifier with a gain of 247. The thermocouple is placed as close as possible under the electrode surface (200 µm). This enables the temperature of the contact to be measured, after breaking contact, for an arc-duration even as short as 1ms. The time-constant of the probe (contact containing the thermocouple) is measured to be approximately 18ms. With this technique the temperature of the electrodes are measured for currents and gap-lengths ranging 4-10 Amps and 0.05-1 mm respectively. The effect of contacts being new and change of polarity have been investigated. From these results it is concluded that the co-existence of layers of foreign material on one, or both, surfaces causes the temperature of the electrodes to be high for the first 50-100 operations, before reaching to steady-state conditions. Change of polarity suggests that the moving electrode, either anode or cathode, due to the effect of air movement over its surface, is cooled relative to when fixed. The power transfer to the electrodes is calculated for various currents and gap-length using thermal analogue formulae derived from the transient response of an RC circuit to a d.c. pulse. The results show that below 0.2 mm the sum of the anode and cathode power is approximately equal to the arc power, and hence losses are negligible. At around 0.125 mm, for currents of about 6A and 12A, they both receive an equal amount of power from the arc. This has been related to the thermal energy of the electrons being negligible, at such separation, at the anode end of the plasma column. The power balance equations are solved to calculate the positive ion current to the cathode, and the thermal energy of the electrons in the plasma column, under various test conditions. In the investigation of erosion, the S.E.M. studies show that most of the power dissipated on the surface of the electrodes is used in melting and evaporation. The x-ray analysis shows that the melted metal is composed mostly of Silver. To operate the test rig and collect the generated data automatically, a computerised test system, with a mini data acquisition system, has been designed and constructed here.

Thesis (M.S.)--University of Wisconsin--Madison, 1962. / Typescript. eContent provider-neutral record in process. Description based on print version record. Includes bibliographical references (leaves 108-109).

M.Ing. (Electrical and Electronic Engineering) / This work reports on a study in the fields of "High power pulsed power supplies and electromagnetic pulse compression". A background discussion on the application of pulsed power on lasers is given. Pulse compression and high power pulsed power supply configuration is described. The work concentrates on simultaneous energy transfer electromagnetic pulse compression which is a technique used to achieve a saving of core material in a pulse compressor. An analytical expression for the saving was derived and verified experimentally. This result is applied in different pulse compressors designed for the pulsed power supplies of copper vapour lasers. The test results are provided and discussed. A windfall from the practical work is a design solution for the voltage sharing compensation network needed when switching semiconductors in series.

D.Ing. (Electrical & Electronic Engineering) / 'Ihis thesis considers the application of non-linear, voltage dependent, saturating capacitors as turn-off snubbers for power electronic switches. The concept of using a non-linear turn-off snuJ::ber for the relieving of semiconductor switching devices is shown theoretically and experimentally. As proven by the results, the most outstanding advantage of non-linear snubbers is the much smaller quantity of energy which is being stored, during and after snubbing, in such a non-linear snubber device, compared to normal linear snubber elements. Depending on the saturation level of such a non-linear capacitor, the stored energy can (for existing ceramic capacitors) be an order of magnitude lower than the energy in a comparable linear capacitor. After the profitability of using non-linear saturating capacitors has been demonstrated, the non-linear capacitive snubber is analyzed by means of a computer simulation which uses the exact capacitancevoltage curve, as stored on data file. The circuit is, however, also analyzed analytically by approximating the non-linear capacitor characteristics in two different ways: (L) by an exponential approximation, and (ii) by a two-step piecewise linearization. Most of the results are within about 20 %of those obtained by the exact analysis, and the approximate analyses can therefore be regarded as very useful. Using the analysis of the non-linear snubber, an optimization in terms of minimum energy losses is carried out on a general turn-off snubber circuit which consists of a non-linear saturable capacitor in a regenerative snubber configuration. The equations and results are awlicable to most snubber cases. From the results it is evident that the same performance as obtained with complex linear regenerative snubber circuits can be obtained by a simple non-linear dissipative snubber...

M.Ing. / Power supplies for pulsed laser applications are required to produce high energy pulses of microsecond and sub microsecond duration. Gas phase switches have traditionally been used in these applications. The exorbitant replacement as well as running costs, coupled with the limited lifetime of these switches, has led to the investigation of the implementation of an equivalent semiconductor switch. The semiconductor switch consists of thyristors coupled in series. This switch is connected to a two stage ferrite pulse compressor in order to achieve the required output pulse shape. The use of series coupled thyristors as high voltage switches in fast switching applications has been limited to laser and radar applications. Little has been published about work in this field and it was necessary to investigate the characteristics of the switching elements, under pulsed conditions, in order to optimise the switching and conduction losses of the elements as well as the switch as a whole. This investigation led to the development of an 18 kV switch which produces 750 nanosecond pulses at a repetition rate of 10 kHz. The applicability of similar switches in the radar and laser fields is discussed.

The 21st century is emerging into an era of wireless ubiquity. To support this trend, the RF (Radio Frequency) front end must be capable of processing a range of wireless signals (cellular phone, data connectivity, broadcast TV, GPS positioning, etc.) spanning a total bandwidth of nearly 6 GHz. This warrants the need for multi-band/multi-mode radio architectures. For such architectures to satisfy the constraints on size, battery life, functionality and cost, the radio front-end must be made reconfigurable. RF-MEMS (RF Micro-Electro-Mechanical Systems) are seen as an enabling technology for such reconfigurable radios. RF-MEMS mainly include micromechanical switches (used in phase shifters, switched capacitor banks, impedance tuners etc.) and micromechanical resonators (used in tunable filters, oscillators, reference clocks etc.). MEMS technology also has the potential to be directly integrated into CMOS (Complementary metal-oxide semiconductor) ICs (Integrated Circuits) leading to further potential reductions of cost and size. However, RF-MEMS face challenges that must be addressed before they can gain widespread commercial acceptance. Relatively low switching speed, power handling, and high-voltage drive are some of the key issues in MEMS switches. Phase noise influenced by non-linearities, need for temperature compensation (especially Si based resonators), large start-up times, and aging are the key issues in Si MEMS Resonators. In this work potential solutions are proposed to address some of these key issues, specifically the reduction of high voltage drives in switches and the reduction of phase noise in MEMS resonators for timing applications. MEMS devices that are electrostatically actuated exhibit significant non-linearities. The origins of the non-linearities are both electrical (electrostatic actuation) and mechanical (dimensions and material properties). The influence of spring non-linearities (cubic and quadratic) on the performance of switches and resonators are studied. Gold electroplated fixed-fixed beams were fabricated to test the phenomenon of dynamic (or resonant) pull-in in shunt switches. The dynamic pull-in phenomenon was also tested on commercially fabricated lateral switches. It is shown that the resonant pull-in technique reduces the overall voltage required to actuate the switch. There is an additional reduction of total actuation voltage possible via applying an AC actuation signal at the correct non-linear resonant frequency. The demonstrated best case savings from operating at the non-linear resonanceis 50 % (for the lateral switch) and 60 % (for the vertical switch) as compared to 25 % and 40 % respectively using a fixed frequency approach. However, the timing response for resonant pull-in has been experimentally shown to be slower than the static actuation. To reduce the switching time, a shifted-frequency method is proposed where the excitation frequency is shifted up or down by a discrete amount 'Ω after a brief hold time. It was theoretically shown that the shifted-frequency method enables a minimum realizable switching time comparable to the static switching time for a given set of actuation frequencies. The influence of VDC on the effective non-linearities of a fixed-fixed beam is also studied. Based on the dimensions of the resonator and the type of resonance there is a certain VDC,Lin where the response is near linear (S ' 0). In the near-linear domain, the dynamic pull-in is the only upper bound to the amplitude of vibrations, and hence the amplitude of output current, thereby maximizing the power handling capacity of the resonator. Apart from maximizing the output current, it is essential to reduce the amplitude and phase variations of the displacement response which are due to noise mixing into frequency of interest, and are eventually manifested as output phase noise due to capacitive current nonlinearity. Two major aliasing schemes were analyzed and it was shown that the capacitive force non-linearity is the major source of mixing that causes the up-conversion of 1/f frequency into signal sidebands. The resonator's periodic response (displacement) is defined by a set of two first- order nonlinear ordinary differential equations that describe the modulation of amplitude and phase of the response. Frequency response curves of amplitude and frequency are determined from these modulation equations. The zero slope point on the amplitude resonance curve is the peak of the resonance curve where the phase ('dc) of the response is ±π/2. For a strongly non-linear system, the resonance curves are skewed based on the amount of total non-linearity S. For systems that are strongly non-linear, the best region to operate the resonator is the fixed point that correspond to infinite slope ('dc = ±2π/3) in the frequency response of the system. The best case phase noise response was analytically developed for such a fixed point. Theoretically at this fixed point, phase noise will have contributions only from 1/f noise and not from 1/f2 and 1/f3. The resonators phase can be set by controlling the rest of the phase in the loop such that the total phase around the loop is zero or 2π. In addition, this work has also developed an analytical model for a lateral MEMS switch fabricated in a commercial foundry that has the potential to be processed as MEMS on CMOS. This model accounts for trapezoidal cross sections of the electrodes and springs and also models electrostatic fringing as a function of the moving gap. The analytical model matches closely with the Finite Element (FEA) model. / Ph. D.

No description available.