G.D. Wait

Analysis of high power IGBT short circuit failures

The next linear collider (NLC) accelerator proposal at Stanford Linear Accelerator Center (SLAC, Menlo Park, CA), requires a highly efficient and reliable, low cost, pulsed-power modulator to drive the klystrons. A solid-state induction modulator has been developed at SLAC to power the klystrons; this modulator uses commercial high voltage and high current insulated gate bipolar transistor (IGBT) modules. Testing of these IGBT modules under pulsed conditions was very successful; however, the IGBTs failed when tests were performed into a low inductance short circuit. The internal electrical connections of a commercial IGBT module have been analyzed to extract self- and mutual partial inductances for the main current paths as well as for the gate structure. The IGBT module, together with the partial inductances, has been modeled using PSpice. Predictions for electrical paths that carry the highest current correlate with the sites of failed die under short circuit tests. A similar analysis has been carried out for a SLAC proposal for an IGBT module layout. This paper discusses the mathematical model of the IGBT module geometry and presents simulation results.

Measurements on a FET based 1 MHz, 10 kV pulse generator

A prototype pulser, which incorporates thirty-two 1 kV field-effect transistor (FET) modules, has been built and tested at TRIUMF. The pulser has been developed for application in a scheme for pulsed extraction from the TRIUMF 500 MeV cyclotron. Deflection of the beam will be provided by an electric field between a set of 1 m long deflector plates. The pulser generates a continuous, unipolar, pulse train at a fundamental frequency of approximately 1 MHz and a magnitude of 10 kV. The pulses have 38 ns rise and fall times and are stored on a low-loss coaxial cable which interconnects the pulse generator and the deflector plates. The circuit performance was evaluated with the aid of PSpice in the design stage and confirmed by measurements on the prototype. Temperature measurements have been performed on 1 kV FET modules under DC conditions and compared with temperatures under operating conditions to ensure that switching losses are acceptable. Results of various measurements are presented and compared with simulations.

A High Frequency Mosfet Driver for the Titan Facility at TRIUMF

TRIUMFs Ion Trap for Atomic and Nuclear Science (TITAN) Radio Frequency Quadrupole (RFQ) Beam Cooler is a device that cools and collects short-lived isotopes, with half-lives as short as 10 ms, created by an Isotope Separator and Accelerator (ISAC). An RF square wave driver (RFSWD), that must have rise and fall times of less than 125 ns (10% to 90%), performs 2-dimensional focusing of the ion beam within the RFQ, along planes normal to the beams intended trajectory, to confine ion motion along a stable path; hence the ions can be trapped and collected for extraction. The RFSWD, which is based on previous kicker designs developed at TRIUMF, employs stacks of MOSFETs, operating in push-pull, to generate high voltage (HV) rectangular waveforms at a prescribed frequency and duty cycle. Currently a 500 V, 2 MHz drive system is undergoing tests, however, the system configuration allows for operation with higher voltage amplitudes and a repetition rate from 300 kHz up to 3 MHz, continuous. Technical details of the design, operation and performance of the RFQ system, in particular of the drive system, are presented.

Magnetic field in a prototype kicker magnet for the KAON Factory

Kicker magnets are required for all ring-to-ring transfers in the 5 rings of the proposed KAON Factory. The kick must rise from 1% to 99% of full strength during the time interval of gaps (80 ns to 160 ns) created in the beam so that beam extraction losses are minimized. The kick strength must have a uniformity of /spl plusmn/1% over the useful aperture of the magnet. PE2D calculations have been performed to determine the uniformity of the combined electric and magnetic kick in the aperture of a TRIUMF prototype kicker magnet. Measurements of the magnetic field were performed with 50 /spl Omega/ striplines while the prototype magnet was excited with a low voltage 1 MHz sine wave. The predicted and measured results for the magnetic field in the kicker magnet are in good agreement, and are presented. The circuit analysis code PSpice has been utilized to mathematically model the magnet and stripline probe, and the results of the simulations have provided a better understanding of the effect of parasitics upon the measurements. >

Optimization of speed-up network component values for the 30 /spl Omega/ resistively terminated prototype kicker magnet

Kicker magnets are required for all ring-to-ring transfers in the 5 rings of the proposed KAON factory synchrotron. The kick must rise from 1% to 99% of full strength during the time interval of gaps created in the beam (80 ns to 160 ns) so that the beam can be extracted with minimum losses. In order to achieve the specified rise-time and flatness for the kick it is necessary to utilize speed-up networks, comprising a capacitor and a resistor, in the electrical circuit. Speed-up networks may be connected electrically on both the input and output of the kicker magnet. In addition it is advantageous to connect a speed-up network on the input of the resistive terminator(s). A sequence which may minimize the number of mathematical simulations required to optimize the values of the 8 possible speed-up components is presented. PE2D has been utilized to determine inductance and capacitance values for the resistive terminator; this data has been used in PSpice transient analyses. Results of the PE2D predictions are also presented. The research has culminated in a predicted kick rise time (1% to 99%) of less than 50 ns for a TRIUMF 10 cell prototype kicker magnet. The proposed improvements are currently being implemented on our prototype kicker system.<<ETX>>

Prototype studies of a 1 MHz chopper for the kaon factory

The 1.025-MHz beam chopper will create 108-ns gaps in the 1GeV/c H/sup -/ beam to allow enough time for the magnetic field to be established in the kicker magnets in each of the 5 accumulator rings of the kaon factory at the Tri-University Meson Facility (TRIUMF). The required deflection of 1 mrad can be achieved with a set of plates 5-cm apart in which the product of voltage difference and plate length is 37.7 kV-m. The kick must have a rise and fall time of less than 39ns and a flat top of 49ns, and 92ns on alternate pulses. The 1-MHz chopper concept involves an energy storage system where the electric pulses are stored in a large-diameter low-loss coaxial cable. Measurements of the performance of a high-voltage prototype are presented. Results show that this novel design can be implemented successfully for the kaon factory.<<ETX>>

Results of calculations on the beam deflection due to the 1 MHz chopper for the kaon factory

Deflection rise time is a function of beam transit time through the deflector plates and the rise time of the stored voltage pulse. The authors present the results of time-domain mathematical simulations to assess the relationship between these quantities: the results of these simulations allow an accurate determination of the required rise-time of the stored voltage pulse. The representation of the deflector plates is modified so that linear displacement of the beam, as well as angular deflection, may be assessed. Simulations have also been performed to assess the attenuating effect of the deflector plates upon both angular deflection and linear displacement of the H/sup -/ beam caused by voltage ripple. A measured voltage pulse is simulated as driving the deflector plates, and beam deflection is predicted.<<ETX>>

Precision 25 kV Pulse and RF Suppression in a 75 kHz Kicker

An international collaboration plans to measure the lifetime (~2 mus) of the muon to a precision of 1 ppm. The MuLan experiment is in progress at the Paul Scherrer Institut (PSI) in Switzerland. This experiment requires a fast beam line kicker, which can turn the beam on and off, to invoke an artificial time structure on the continuous beam which has a 50.6 MHz time microstructure. The kicker was designed and built at TRIUMF and installed at PSI in June 2003. Tests at PSI showed that RF from the kicker interfered with experimental detectors. Also it was realized that the specification for the flattop of the voltage pulse was not adequate for this precision experiment. Significant reduction in the RF and improvement in the pulse flattop were required. The kicker was returned to TRIUMF, and extensive changes were made, based on detailed measurements and PSpice simulations. After the changes the RF was reduced by a factor of more than 5000. The flattop variation, for a 22 mus pulse duration, was reduced from 0.18% to less than 0.001%: in addition the pulse overshoot was reduced from 10% to 2%.

Longitudinal impedance of a prototype kicker magnet system

The longitudinal impedance of a kicker magnet system for the proposed KAON factory has been measured from 0.3 to 200 MHz. The measurement was done by transforming the kicker magnet under test into a coaxial line in order to measure the transmission parameter S/sub 21/ through the line. The measurement was performed in two frequency ranges. From 0.3 to 50 MHz the magnet was transformed into a 50 /spl Omega/ coaxial line, and from 45 to 200 MHz into a 180 /spl Omega/ coaxial line. Resonances in the longitudinal impedance spectrum are due to the electrical resonant modes of the kicker magnet system. The effect on the longitudinal impedance of a speed-up network and a saturating inductor, installed on the input to the kicker magnet to improve its kick performance, was determined. The speed-up network can damp some of the resonances whereas the saturating inductor can eliminate the resonances due to the input cable of the kicker magnet. Above 45 MHz where attenuation in the LC cells of the kicker magnet is strong, external components connected to the kicker magnet have negligible influence on the longitudinal impedance. Hence the longitudinal impedance spectrum of the kicker magnet in the 45 to 200 MHz frequency range does not depend on external components such as the speed-up network, the saturating inductor and the input and output cables.<<ETX>>

Algorithm for the deflector plates of the 1 MHz chopper for the KAON Factory

Results are presented of time-domain mathematical simulations to assess errors introduced by approximating the deflector plates using a finite number of sections. In order to validate the mathematical model of the deflector plates the predictions are compared with analytical equations for angular deflection for the situation where center-fed plates are energized by a step-function. Predictions of angular deflection are presented for four configurations of deflector plates in order to confirm that center-feeding is the best option considered.<<ETX>>