Access Official Technical PDF

Share Summary:LinkedInEmail

Generating high-voltage pulses from a low-voltage DC supply presents a design trade-off: increasing pulse voltage can require more converter modules, while sequentially charging those modules can constrain pulse repetition. A modular multilevel converter (MMC)-based high-voltage pulse generator addresses this challenge by charging each sub-module capacitor in voltage boost mode before connecting the charged capacitors in series across the load.

The approach is designed for rectangular pulse applications, including pulsed electric field systems. Its central idea is to use the sub-module capacitor and switching devices as part of a boost-converter charging process. This can produce a higher pulse voltage for a given number of sub-modules than charging each capacitor only to the input supply voltage.

Why use a modular pulse generator?

Traditional Marx generators charge capacitors in parallel and discharge them in series. Classical versions use spark gaps; solid-state Marx generators replace those gaps with semiconductor switches. Pulse-forming networks use cascaded inductor-capacitor sections to shape a pulse. These established approaches have trade-offs: Marx-generator structures can be inflexible, while pulse-forming networks have pulse widths linked to their selected component values.

MMC-based pulse generators offer a modular alternative. In sequential-charging designs, sub-module capacitors are charged one at a time and then inserted together in series to produce the output pulse. However, when each capacitor is charged only to the DC input voltage, the pulse peak is limited by the number of sub-modules multiplied by that input voltage. Raising the output then typically means adding modules, which also increases the time needed for sequential charging and the converter footprint.

How voltage boost mode works

The proposed topology uses two arms of half-bridge MMC sub-modules. One arm generates the positive pulse polarity and the other generates the negative polarity. Two additional reverse-blocking switches direct the charging process and help provide a current path for the input inductor. During pulse generation, the selected arm’s capacitors are connected in series across the load; the other arm is bypassed.

Charging each sub-module

During charging, one sub-module at a time is connected with the DC source and input inductor. The sequence has two stages:

  • Inductor energizing: The sub-module’s main switch is turned on, allowing current from the input supply to build in the inductor.
  • Capacitor charging: The main switch is turned off. The inductor’s stored energy drives current into the sub-module capacitor through the associated diode path.

Repeating these stages for the sub-modules charges each capacitor to a voltage above the input supply. The ratio between the inductor-energizing interval and the capacitor-charging interval sets the boost factor, represented by λ. In the ideal relationship described for the design, equal intervals correspond to a boost factor of two; a longer energizing interval increases the factor. Actual limits depend on the supply voltage, semiconductor ratings, and circuit losses.

Once charged, the capacitors in an arm are inserted simultaneously to deliver a rectangular pulse. With N sub-modules per arm, input voltage Vs, and boost factor λ, the pulse peak is expressed as NλVs. Switching between the two arms enables positive and negative pulse polarities without physically changing the power hardware.

Pulse control and capacitor balancing

For rectangular pulses, the control sequence sets the pulse duration and coordinates recharging between pulses. The charging intervals matter to the overall repetition time: increasing the number of sub-modules or the boost factor can require more charging time. The paper identifies a fast controller as an operational requirement because pulse generation and charging are software controlled.

Capacitor voltage differences can also accumulate, particularly when sub-module capacitances are not identical. For more general, multilevel pulse waveforms, the proposed balancing method measures the voltage across each arm rather than measuring every individual capacitor. During each sub-module’s allocated charging slot, the controller compares the arm measurement with a reference and recharges a capacitor when its voltage is below the target. This helps address voltage drift while keeping the sensing arrangement to one voltage sensor per arm. These sensors require high-voltage isolation and sufficient bandwidth for the generic waveform mode.

What the validation demonstrates

MATLAB/Simulink simulations and scaled-down experiments were used to assess the topology. In simulation, three sub-modules per arm and a 1 kV input produced a 6 kV pulse peak at a boost factor of two, and a 9 kV peak at a boost factor of three. The corresponding simulated repetition times were 70 and 100 microseconds. These results illustrate the relationship between boost factor, output voltage, and the time required to energize the inductor.

The experimental setup used a 50 V input and three sub-modules per arm. It produced reported pulse peaks of 300 V at a boost factor of two and 440 V at a boost factor of three. The measured repetition times were 260 and 400 microseconds, respectively. The experimental output at the higher boost factor was accompanied by a longer repetition time, consistent with the additional inductor-energizing interval. These are results from a scaled-down proof of concept, not a specification for a production system.

Practical design considerations

The design offers a way to increase pulse voltage without relying solely on adding sub-modules, and its two-arm arrangement supports bipolar pulses. It can also be extended to multilevel waveforms using arm-level voltage feedback. At the same time, engineers must account for the sequential charging time, switching-device voltage ratings, inductor current ripple, capacitor energy requirements, and the sensing and control needed for voltage balancing.

For system integrators and plant teams evaluating pulse-power equipment, the key implication is that voltage gain is not independent of timing or hardware constraints. Selecting the number of sub-modules, boost factor, pulse duration, and recharge intervals requires considering the required output waveform and repetition rate together.

Conclusion

A boost-mode MMC high-voltage pulse generator uses each sub-module as part of a sequential boost-charging process, then connects charged capacitors in series to produce a rectangular pulse. The architecture demonstrated bipolar operation and a route to multilevel waveform generation with arm-level voltage sensing. Its practical value depends on balancing the desired pulse voltage against charging time, component ratings, capacitor behavior, and controller performance.

Access Official Technical PDF

About The Author