August 5, 2005  


Marx Modulators are a Potential Major Benefit to ILC

By Albe Larsen

When Greg Leyh (ESD) asked for approval to design a Marx-type solid-state modulator, SLAC was in the thick of the NLC program and was cranking up the 8-pack test facility. However, a Marx modulator based on IGBT technology seemed a strong possibility to replace the solid-state modulator developed for the NLC 8-pack—a short-pulse machine using a novel coaxial transformer to achieve 500 kilovolts (kV). Technical Division R&D funds supported this development, and this effort started apart from mainstream NLC R&D about three years ago.

The first test cell for the ILC Marx Modulator designed by Greg Leyh and built by Greg Leyh and Piotr Blum
(Photo courtesy of Ray Larsen)

Why, then, was this development important enough to commit special Technical Division R&D funds? While incorporating solid-state IGBT technology exploited for the NLC 8-Pack modulator, it would be much smaller, could avoid the complex magnet system and could also be significantly less expensive to build—very important for an instrument with the price tag associated with the linear collider. There were plenty of challenges, chief of which were very rapid switching and compensating for voltage droop when the modulator fires.

Development started and Leyh, with Piotr Blum (ESD) providing mechanical design and technical assistance, built and tested an 18 kV, 550 ampere (A) Marx Cell. The project status was reported at the 2004 European Particle Accelerator Conference.

The ICFA decision to use superconducting L-Band cavities for the ILC (August 2004) came while plans were in process to build a 30-cell unit to drive two 75 kW klystrons.

What then of the Marx Modulator? The parameters for a superconducting rf system are very different from those for the warm NLC. With the same nimbleness of the whole NLC team in transition to an ILC R&D plan, Leyh assessed the new parameters and determined that we could indeed build a Marx modulator that would work. This would also have at least one additional large advantage at the lower voltage of 120 kV—no need for oil cooling. Leyh also projected higher modulator efficiency (~98 percent vs. ~60 percent for the 8-Pack design) and lower switching losses.

The biggest challenge was the much longer pulse length—now 1.6 milliseconds (ms)—with a good flat top. This is an issue because the energy storage capacitors in the Marx will have significant ‘voltage droop’ at long pulse lengths. To solve this problem, Leyh designed a novel compensation scheme in which a few of the cells (large, multistage IGBT boards) are dedicated to pumping additional charge in short ramps across the long pulse to keep the pulse top flat (±0.5 percent). Other output parameters are 120 kV at 23 kilojoules per pulse (23 kJ is about the energy of a 20 mm cannon shell) delivered to the klystrons at 5 pulses per second. The klystrons convert these 120 kV pulses into 10 megawatts (MW) of peak rf power for the accelerating structures.

Compare this with NLC parameters of 500 kV delivering 440 joules (equivalent to a .38 caliber pistol) over a 1.6 microsecond (µs) time, at a rate of 120 pulses per second—a much higher voltage and current and a much shorter, faster pulse. This was a huge change.

The current design includes a 16-cell modulator in which 10 cells are active and 4 cells are used to compensate for voltage droop in the output pulse. In actuality, only 14 of the cells are used and two act as spares. This allows up to two active cells to fail and the two on standby take over without impeding operation. In a machine such as the ILC with 576 modulator-driven klystrons, minimizing failure and maximizing both reliability and availability is crucial to successful operation. The modulator is completely modular so that failed cells can be replaced while the modulator is still operating, using either a remote hand-operated machine or by an overhead robotic service platform. This, too, increases availability.

So far the greatest effort has gone into designing the 12 kV IGBT switch, since it has the greatest technical risk. The first prototype cell has been completed and tested successfully. A 6-cell ‘short stack’ is well toward completion and will be tested by the end of the year.

Gerry Dugan, U.S. Deputy Director of the ILC Global Design Effort says, “Successful development of this Marx Modulator should be considered the SLAC ILC team’s first priority.”

Industry Involvement

While the SLAC ILC Power Conversion group has made great progress, interest in the Marx modulator has been stimulated in the broader HEP community. This led to three DOE Office of Science Small Business Innovation Rewards (SBIRs) for Marx Modulator work. Diversified Technologies of Bedford, MA is now working on a Marx modulator with parameters similar to the SLAC device but with very different packaging. ISA Corporation of Dublin, CA has proposed development of a similar Marx modulator that might also be used as an economic alternative in the radiation processing of materials and in food irradiation. Stangenes Industries of Palo Alto is also beginning work on a solid-state ILC Marx modulator, with other applications being cargo inspection devices and medical accelerators.

For more information on Marx Bank, see:
NLC design:
ILC design:





The Stanford Linear Accelerator Center is managed by Stanford University for the US Department of Energy

Last update Tuesday August 09, 2005 by Topher White