How We Finally Stopped Burning Marx Generator Electronics – Part 3

Part 1 and part 2 of this series covered what goes wrong — an internal fire from a charging circuit that won’t let the spark gaps recover, and 100’s of kV launched back into your supporting electronics from a self-break. Part 3 is the practical one: what APELC actually does to set up and operate a Marx generator safely, at both low and high repetition rates.

These aren’t theoretical recommendations. They’re the configurations we’ve developed over decades of building and testing these systems — some of them learned the expensive way.

Low Repetition Rate Configuration

Let’s start with the low repetition rate case. The figure below provides a simple configuration map, showing a Marx generator supported by a power supply and a trigger source. For this scenario, we don’t care whether the generator uses resistors or inductors for the (internal) circuit’s charge elements. The new element in this discussion is the inline charge resistor and its placement. We use a long carbon composition resistor with a value of approximately 22 kΩ. The placement of this resistor is important. Generally, we try to keep the length of the cable from the power supply to the resistor (Length1) much longer than the length of the cable connecting the resistor to the generator.

Why do we do this? In a self-break scenario, a very fast transient from the Marx generator will make it onto the charge line. And while the charge voltage may only be 30–40 kV, the transient coming back might be in the 100’s kV to even MV’s. Not something we like dealing with. With enough cable length (Length1), we can use the series resistor to absorb much of the pulse, assuming a simple resistive voltage divider. In essence, when the transient pulse hits the inline resistor, there is a short period of time that the pulse “sees” the cable impedance (or Length1) compared to the inline resistor, and we massively attenuate the pulse’s amplitude (Vp = Vtransient x 50 Ω / (22 kΩ + 50 Ω), or approximately 440X. This configuration works well.

Higher Repetition Rate Configuration

The high repetition application is trickier. In the low repetition model, the losses in the inline charge resistor, during the charge cycle, are manageable and acceptable. Yes, the resistor heats up some, but we typically don’t care, since there is usually plenty of time to cool. Higher repetition rates, however, can lead to substantial heating. In the past, we have flowed air through the resistor housing, which is ok for lower rates. We have also discussed flowing transformer oil through the resistor housings. But now, the overhead is increasing (as well as the potential mess).

The best alternative is to incorporate a passive low pass filter near the power supply and use a long charge cable to help reduce the transients through dispersion and resistive losses. This is a tricky approach, since those transients seem to find their way through most anything, including the ground loops. Designing and building these circuits can also be costly and iterative.

Pulse-Charging

We have had a lot of success using well-timed pulse-charging techniques. In essence, we “command” charge the Marx generator, then turn off the power supply just prior to triggering the Marx generator. The high voltage power supplies seem less prone to the high voltage transients in the disabled state.

Protecting the Trigger

The trigger circuit is very difficult to protect. The challenge is to deliver a very fast transient to trigger the Marx generator while protecting the circuit from an equally fast transient pulse coming back from the Marx generator. The predominant difference between the pulses is the amplitude — we might deliver a 30–40 kV pulse, while seeing a several 100 kV pulse coming back. Diodes seem great; however, they become expensive when failing from every self-break event.

We have found that long cables help; but also deteriorate the trigger pulse. Ultimately, we try to include a parallel (crowbarring) spark gap near the trigger to shunt the unwanted transients that are very high in amplitude.

Why We Wrote this Blog Series

This three-part Lessons Learned series tells the same story from three angles: what burns, what gets destroyed, and how to stop it from happening. The inline resistor, the pulse-charging timing, and the crowbarring spark gap aren’t exotic solutions. They’re the ones you arrive at after losing enough hardware. We’d rather you arrive at them sooner.

If you’re designing or operating a Marx generator system and want to talk through configuration, protection, or system design, APELC is glad to help. Reach out directly, or explore our Marx generator product line to see how we approach these problems from the ground up.

————————-

About the Author
Jon Mayes is the Founder and President of APELC. With over 25 years of experience in pulsed power system design and a Ph.D. in Electrical Engineering, Jon has led the development of industry-leading Marx generators, EMP simulators, and high-voltage test systems. His work has supported the Department of Defense, Department of Energy, and major research institutions across the U.S.

Follow APELC on LinkedIn for more insights on pulsed power engineering, high-voltage system design, and lessons learned from the field.
Follow us on LinkedIn →