This is because the pre-firing is caused by a slow breakdown of the gas related to a phenomenon called “corona discharge”. With more capacitors attached, the power supply’s fixed output charges the bank more slowly, so the switches stay longer at high voltage. Because of this problem, the present configuration of the switches will not work with all 12 capacitors attached. Ten capacitors is our current maximum.
However we know the cure for this. We’ll move the electrodes in the switches further apart (see the attached diagram for further explanation). We will be doing this in our general redesign over the next few months. In addition, if the tungsten trigger rods get worn down so they are too thin or too rough, the field gets too concentrated, and this also leads to pre-firing. Temporarily, we have replaced the thinnest rods and are sanding the others carefully on a regular basis. In our re-design, we will use much thicker rods—probably one-quarter-inch diameter instead of one-eighth inch.
We also had trouble eliminating breakage. While our large stabilizer block prevented any cracking of the insulator above the plate, the rapid movement of the tungsten rods was still breaking the Lexan insulators near the tip. After several tries, we have used a solid cylinder insulator to provide maximum strength. So far, they have lasted 40 shots with only two cracking, so this is adequate for now, although too soon to tell their real lifetime. Again, we know the solution to the mechanical breakage: making the rods thicker so they bend less, and making the insulators thicker so they are stronger. These changes require replacing the top plates of all the switches and making the spark plug holes larger. This hole size has limited our past efforts with the spark plugs.
Finally, to prevent the electrical break-down of the insulators, we also have to make them thicker. All of this can be calculated, based on the experiments we have done, and we hope to complete design work very soon, probably early in January. However, there are considerable ordering delays on some items, such as the tungsten rods, and some additional testing will be needed, so realistically we will not complete the new switches until March. Until then, we will be running with the existing 10–capacitor bank.