Marx generator troubleshooting can eat a full day if you don’t have a system. There are a lot of components in the circuit, the failure mode isn’t always obvious, and a bad reading in one place often points you somewhere else entirely. Over years of building and testing these systems, APELC has developed a structured debugging process that cuts through most of that — including one technique that can locate a failed capacitor in minutes, without touching anything.
This post covers the process we use for Marx generators built around ceramic doorknob capacitors. If you’re working through a fault on a similar system, or just want to understand how experienced pulsed power engineers approach this kind of problem, here’s the inside view.
A Structured Approach to Marx Generator Troubleshooting
We build a lot of these generators, and when they sometimes fail, it becomes a process trying to figure out the problem. Let’s start with a flow diagram to get to the potential problems. I note that others might have a different, or better process. But, this process seems to work for us.
Now, let’s dive in deeper.
Start With the Basics: Confirm Your Sources Are Working
The quick tests look at the charging power supply and the trigger – easy enough. The figures below provide the quick setups to determine the functionality of each source. In the first figure, we remove the charge cable, and carefully witness the voltage on the cable with a high voltage probe, while only applying a 1 kV voltage. In the second figure, testing the high voltage trigger source, we remove the trigger cable and place it on a metal surface, creating a tiny gap between the center pin and the plate. We then pulse the trigger unit, hopefully observing an arc forming on the gap.
Figure 1. Testing the high voltage supply.
Figure 2. Testing the high voltage trigger supply.
If the Sources Check Out: Open the Generator and Verify Continuity
If both charge and trigger measure good, we then need to remove the Marx generator from the housing to inspect the circuit. This should be done very carefully, since residual charge may be present. As we open the Marx generator, we typically use a resistive grounding stick, with an insulated handle and connected to ground, to short circuit the stage capacitors. Once removed, we lay the Marx circuit on the bench for testing, with the charge and trigger cables removed.
The first test is a continuity check. Below is a simple circuit diagram overlayed to a Marx generator (sorry, we have to cover up the good stuff), noting the measurement points. Whether the capacitors are resistively charged, or inductively charged, the measurement is the same. We are looking for continuity between the input (or charge side) and the output side.
Figure 3. Checking the continuity of the charge elements.
If either set of the charge elements doesn’t show continuity, there is a very good possibility that one of the charge elements has failed and needs to be replaced.
Locating a Failed Capacitor Fast: The Infrared Method
This is where the process gets interesting. Locating a failed ceramic doorknob capacitor used to mean working through the circuit iteratively — removing charge elements one at a time until you found the culprit. It’s slow and tedious. APELC developed a faster method that uses the failure mode itself to point directly at the bad component.
If there is continuity with both sets of charge elements, the problem might be with one of the capacitors. Traditionally, locating a failed capacitor is a difficult task, iterating through many charge configurations, by removing charge elements, not-so different from a programming bubble sort (probably showing my age). However, over the years, we found a neat-o method for quickly finding the failed capacitor. We have found that when the doorknob capacitors fail, without some physical evidence, like a chunk of the epoxy missing, it’s an internal flashover between the ceramic slug and the epoxy, and the flashover usually starts with an applied voltage of about 3 kV. By applying a DC voltage, just above the flashover point, we can produce a steady arc inside the bad capacitor, which will heat the capacitor. We then use an infrared thermometer, or a phone with an infrared imager to locate the failed capacitor. It’s super easy.
To perform this trick, we re-attach the high voltage charge cable, then carefully begin charging the Marx circuit to the approximate 3 kV level. Once the capacitor begins failing, the charge voltage should not increase and the current should be abnormally high. The bad capacitor should “light-up”, indicating that it’s much hotter than the others. The wonderful part of this approach is that it’s a stand-off / no-touch measurement. But again, high voltage is involved and only qualified personnel should be performing this test.
Figure 4. Measuring the temperature of the stage capacitors.
Closing Thoughts
The problems described above are the typical failures we find. I’m sure that there are other issues we could discuss, and maybe we will someday in another blog.
If you’re in the middle of Marx generator troubleshooting — or evaluating whether a system is worth repairing versus replacing — APELC is glad to talk through it. This kind of application-level problem solving is exactly what we do. Reach out directly, or browse our Marx generator product line to get a better sense of how we build and support these systems.
