Abstract
M.Ing.
Pulse Compression is a technique that may be employed for the generation of
extremely high amplitude current and voltage pulses. These pulses can be as short
as 50 to 100 ns, and may have amplitudes in the kiloampere and kilovolt ranges.
Pulse Compression entails the compression of relatively "flat" pulses in the time
domain, to pulses of very high amplitudes and extremely short duration. The pulse
amplitudes and durations necessary to be achieved in this research, lie in the range
where the switching speeds and other parameters of semiconductors are inadequate
and where even the working life of conventional gas discharge apparatus are
drastically reduced by the extreme switching demands. The burden of excessively
high current densities and unmanageable current rise-rates can be transferred from
the semiconductor switches to electromagnetic switches, by making use of pulse
compression. Pulse compression can be carried out simultaneously or separately for
the compression of the current or voltage content of pulses derived from slowly
switched sources, to obtain pulses of extremely short duration and very high
amplitudes.
The main theme of this dissertation is Current compression. Current
compression is accomplished through series-resonance in capacitors and saturable
inductors connected in a transmission-line configuration. Energy is transferred in
this process from one stage to the next, with reduction in pulse-time in each
successive stage and a commensurate increase in amplitude. The generated pulses
can attain gigawatt amplitudes and nanosecond durations, whilst loading on the
semiconducting switches remains low. In addition to design of the pulse-compressor
proper, the work also includes design and development of a voltage-controlled pulse
power supply, suitable for generating the initial pulses which are to be compressed.
Multistage pulse compression is based on the non-linear characteristics of saturable
inductors. Dynamic analogue-time simulation is indispensable in a study
thereof, as new theory has to be validated and because non-linear analysis is
complex and capable only of being executed by employing approximation methods.
Because of the difficulties involved, a considerable amount of attention has been
devoted to the development of suitable analogue-dynamic simulation programs for
execution on a digital computer. A numerical technique has been developed to
express non-linear parameters in differential form. This technique makes it possible
to model and simulate virtually any non-linear, physically realizable lumped
parameter system with ease. The program is based on State Space techniques
and has been developed for its versatility, to accomplish the simulation of a wide
variety of circuit configurations.