The idea of using a spark-gap (SG) as the active element of a converter did not come as a sudden illumination. For a long time, I had been looking for an “acceptable” way of using high-voltage, low-current power-supply sources. For me, “acceptable” simply meant technically feasible without exotic technology and sufficiently efficient to be usable. As it was a selfimposed challenge, I had no definite specifications or deadline.
When I had first reviewed the possible ways of implementing such a converter, I had considered the SG as well as a number of alternatives, such as mechanical converters and potential dividers. But as an electronician, my preferences went for purely static, semiconductor- based solutions. During the following years, I made some attempts with oscillators and charge pump schemes, but none was fruitful. The raw voltage was too high, and even the smartest control circuit ate up most of the precious few microamps provided by the source.
In the end, I decided to give the SG a try. Until then, I had ignored it because of preconceptions about its crudeness, primitiveness, and electrical “dirtiness.” I improvised a quick test using some salvaged parts: an old trigger transformer, a high voltage ceramic cap, a makeshift SG, and a rectifier.
My expectations were low. I simply wanted to “tick the box” to close an avenue. I was therefore all the more surprised when I realized that this rudimentary “thing” on my bench gave out significant amounts of power. I decided to somewhat refine my setup to measure some key parameters, such as pseudo-resonant frequency and peak currents. I was then able to make calculations and estimations and design a dedicated transformer.
Next, I built a “proper” prototype with a Schottky rectifier and an organic semiconductor (OSCON) capacitor and gave it a try. As soon as it was powered, I knew I had a winner. Just looking at the sparks told the story. “Regular” sparks tend be bright blue and make a loud crackling noise. Here, they were faint and dark, of a dim indigo color, barely visible. The noise they made was soft too. All this meant that very little energy was lost there and the spark current was comparatively low. This, in turn, had an impact on another important aspect of the design as the level of interference generated was much lower than could be expected from a spark generator.
After this initial success, I went on testing, trying many variations of the configuration, input voltages, and power levels—and blew a number of things in the process! For instance, I tried to improve the efficiency by artificially increasing the leakage inductance of the transformer. I also attempted to take advantage of the negative backswing of the resonant waveform by adding a diode and capacitor to generate an additional negative output. But although these alterations did bring some efficiency increase, it was modest and not really worth the added complication.
The concept of the SG as a switching element is not revolutionary. In fact, it is as old as the SG itself. Yet I think using it in a step-down converter, together with modern components and topologies, is rather novel. It makes practical applications that no other technique allows. Think about it—how do you convert a 10-kV/20- µA supply into 5 V at 20 mA? In addition, there is no threshold effect, no standby power.
An important objection has been raised: the radiofrequency interference (RFI) level. In reality, thanks to the almost purely inductive discharge path and the ease of confinement of the “hot” parts of the circuit, this issue can be adressed rather easily. Another potentially more serious issue, in my opinion, is the durability. SGs have a limited life. A cumulative charge transfer of 400C is typical for an average model. This is not that much, if permanent operation is sought. Note that the actual life will be significantly longer, due to the “gentle” discharge conditions, and is applicable to a standalone, hermetic component. A custom-made, open-construction SG could last longer. This is because a closed SG only has a limited volume, compared to an open one, which is limited solely by electrode erosion.
This circuit was initially designed with high-voltage piezo-elements in mind, but it can find many other uses. Deriving a small auxiliary supply from high-voltage power lines is tricky, and it’s usually accomplished by a current transformer clamped on a conductor. This can only work if enough current is present, and it’s impossible on the shield wire, for example. But an electrostatic harvester whose outer shell is the collecting electrode could still work to provide power for leakage-current monitoring equipment.
As for the theoretical possibility of using it for harvesting atmospheric potentials, it is exactly that: a theoretical possibility. Don’t repeat Georg Richmann’s attempt.