Pulsed Power Engineering

Power is energy per unit time, \(P = W/t\). Pulsed power exploits this ruthlessly: it does not generate vast energy, but takes an ordinary amount and delivers it in an extraordinarily short time. Charge a store over one second and dump it in one microsecond, and the peak power exceeds the charging power by the ratio of those times — a million to one. The energy is humble; the power is colossal. This is power amplification by time compression, and it is the whole philosophy of the field.

The pay-off is a regime of power unreachable any other way. A bench supply delivering a few kilowatts, run for a second into a capacitor, can produce a microsecond pulse of gigawatts — enough to drive a particle beam, flash an X-ray tube, pump a laser, or punch an electric field through living cells. Every pulsed-power system is an answer to one question: how do we store the energy, and how do we let it out fast enough?

There are two places to keep electrical energy: the electric field of a capacitor, or the magnetic field of an inductor.

The two differ in a way that decides the whole machine. To release a capacitor's energy you close a switch — connect it to the load — and good closing switches are plentiful. To release an inductor's energy you must open a switch while it carries the full current, forcing the current to commutate into the load; the inductor then develops a huge \(L\,\mathrm{d}i/\mathrm{d}t\) voltage. Inductive storage has a higher energy density (a magnetic field packs more energy per litre than a practical electric field), but the opening switch — a fuse, a plasma-opening switch, a fast semiconductor — is far harder to build than a closing switch. For that reason the great majority of pulsed-power systems store their energy capacitively, in banks of capacitors, and the rest of this chapter follows that mainstream.

The elementary pulse circuit is a charged capacitor discharged through the unavoidable inductance \(L\) and resistance \(R\) of its leads and load. The character of the pulse depends entirely on how the resistance compares with the circuit's natural impedance \(\sqrt{L/C}\), through the damping condition:

An underdamped circuit rings — the current oscillates and reverses, useful when a sinusoidal burst is wanted but hard on capacitors. A critically damped circuit gives the fastest single pulse with no reversal. An overdamped circuit gives a slow rise and long decay — and this is exactly the double-exponential shape of the \(1.2/50\) impulse from the high-voltage-generation chapter, where one time constant sets the front and another the tail. The peak current of a low-loss discharge is \(I_{\text{peak}}\approx V\sqrt{C/L}\), and the time scale is set by \(\sqrt{LC}\) — so low inductance is the key to a fast, high-current pulse, which is why pulsed-power hardware obsesses over short, fat, low-inductance conductors.

A single capacitor can only be charged to the voltage of its supply, yet pulsed power often needs hundreds of kilovolts or megavolts. The classic solution is the Marx generator, a voltage multiplier of beautiful simplicity. Its \(N\) capacitors are charged in parallel through high-value resistors, each reaching the modest charging voltage \(V\). Then a trigger fires the switches — traditionally spark gaps — that reconnect the capacitors in series. The voltages add, and the generator "erects" to an output of

The Marx is the workhorse of high-voltage pulsed power: ten stages charged to \(50~\mathrm{kV}\) erect to half a megavolt; large machines reach many megavolts. The charging resistors are large enough to behave as open circuits during the fast erection, so they isolate the stages without bleeding away the pulse. The triggered first gap over-volts the next, which fires the next, and the whole stack erects in nanoseconds. The classical Marx uses gas spark gaps, but the modern evolution — replacing each gap with a stack of semiconductor switches — yields a solid-state Marx that fires precisely, repeats rapidly and lasts for billions of shots, a major theme of present-day pulsed-power research.

A capacitor discharge gives an exponential pulse, but many applications — radar, accelerators, lasers — need a flat-topped rectangular pulse of precise duration. The elegant way to make one revives the travelling-wave physics of Chapter 23. Charge a length of transmission line (a pulse-forming line, PFL) to voltage \(V_0\); when a switch connects it to a matched load \(R=Z\), a wave travels to the open far end, reflects, and returns, draining the line in exactly one round trip. The result is a clean rectangular pulse whose width is the round-trip time:

Two refinements complete the toolkit. The Blumlein arrangement uses two lines so that the matched load receives the full charge voltage \(V_0\) rather than half, while keeping the rectangular shape and the \(2\ell/v\) width. And when the required pulse is so long that a real line would be impractically long, a pulse-forming network (PFN) — a ladder of lumped inductors and capacitors — mimics the line's behaviour in a compact box, the number of sections trading off against the flatness of the pulse top. Together, PFLs and PFNs let the engineer dial in a pulse of chosen amplitude, duration and shape.

If the capacitor is the muscle of pulsed power, the switch is its heart — and its hardest organ. A closing switch must hold off the full charge voltage as a near-perfect insulator, then, on command, collapse to a near-perfect conductor in nanoseconds while carrying kiloamperes, repeatably and with low timing scatter (jitter). No single device does all of this perfectly, so the choice is a set of trade-offs:

The historical workhorse is the gas spark gap — a triggered breakdown between electrodes that can switch megavolts and megamperes, at the cost of electrode erosion, shot-to-shot jitter and a limited repetition rate. The modern direction is decisively toward solid-state switching: thyristors, IGBTs, MOSFETs, fast GaN HEMTs and avalanche transistors, which fire with sub-nanosecond precision and survive billions of shots. Because each semiconductor holds off only a limited voltage, they are series-stacked — often in a solid-state Marx arrangement — to reach the required level, trading the spark gap's brute simplicity for precision, longevity and high repetition rate. (The dual problem, the opening switch for inductive systems, remains genuinely difficult and is met with exploding-wire fuses, plasma-opening switches and fast semiconductor openers.)

Early pulsed power was single-shot: charge for minutes, fire once, recover. Modern systems increasingly run at high repetition rate, firing thousands of times a second, which transforms the design priorities. The average power is now \(P_{\text{avg}} = W_{\text{pulse}}\times f_{\text{rep}}\), so heat removal, capacitor lifetime and switch longevity dominate — exactly the strengths of solid-state switching and the reason it has displaced spark gaps wherever repetition matters (industrial, environmental and biomedical pulsed power above all). The same fast fronts impose a severe insulation duty of their own: the steep \(\mathrm{d}V/\mathrm{d}t\) stresses the polymeric and oil insulation of the system in the fast-front regime of Chapter 26, demanding careful grading and clean dielectrics.

Diagnostics close the loop, and they are exacting because the signals are large and fast. Voltage is measured with carefully compensated resistive, capacitive or mixed dividers; current with Rogowski coils, Pearson current monitors or low-inductance current-viewing resistors; and all of it captured on wide-bandwidth oscilloscopes, where the instrument's bandwidth must exceed the pulse's spectral content or the measured front will be a fiction. Measuring a pulsed-power shot well is as much an art as making it.

The reach of pulsed power runs from the laboratory to the food factory. In high-voltage engineering itself, the impulse generators that test insulation are pulsed-power machines. In physics and defence, Marx-driven systems power particle accelerators, flash radiography (single X-ray snapshots of fast events), high-power microwave sources, intense lasers, and the giant Z-pinch and inertial-confinement experiments pursuing fusion. But the fastest-growing branch is the environmental and biomedical one, built on the pulsed electric field (PEF): short, intense field pulses that open the membranes of biological cells (electroporation), enabling non-thermal inactivation of microbes in water, wastewater and food, gentler extraction in processing, and emerging cancer and gene-therapy treatments.

These PEF systems are frequently Marx-generator-based, marrying the voltage-multiplying topology of Section 27-4 with the modern solid-state switching of Section 27-6 to deliver clean, repetitive microsecond pulses at the field strengths cells require. They are the natural bridge to the next chapter, which takes up the science and engineering of pulsed-electric-field applications in their own right — and they show how a discipline born to test transformers and drive fusion has become a tool for treating water and healing patients.

For each item, first identify what it tests — the time-compression idea, the storage choice, the RLC discharge, the Marx generator, the pulse-forming line, the switches, or repetition rate — then apply it. Difficulty rises down the list.