Accelerator Design Department Christopher Nantista MIT Lincoln Laboratory
Accelerator Design Department Christopher Nantista MIT Lincoln Laboratory Visit May 17, 2011
What is RF Pulse Compression? RF pulse compression refers to techniques by which pulse width is exchanged for higher peak power. Employed between pulsed high-power RF sources (klystrons) and accelerating structures, it takes advantage of a source’s typical capability of providing a pulse a few times longer than required (Tk >> Tf + Tbeam). Higher than required power can be divided among multiple structures. While longer pulse width improves klystron efficiency by minimizing the impact of unused rise and fall portions of the high voltage modulator pulse, a price is paid in overall efficiency, depending on the pulse compression system and parameters (and any subsequent power division).
Benefit The benefit of rf pulse compression is chiefly realized: • for long linacs of a given design energy – in the reduction of the number of sources required. • for small accelerator systems – in the energy achievable or in the required source peak power capability. Main RF Pulse Compression Techniques: • SLED II • BPC • DLDS
SLED (SLAC Energy Doubler/Development) is an rf pulse compression technique invented in 1973 by P. B. Wilson and Z. D. Farkas and first employed to boost the energy of the SLAC Linac. It employs high-Q resonant cavities, in which energy is stored for most of the pulse and discharged towards the end via a phase reversal. Each SLED includes a pair of such cavities, coupled through a 3 d. B hybrid in such a way that the reflected power is directed away from the source and toward the accelerator. (A circulator would be power limited and add loss. ) The compressed output pulse is exponentially decaying.
SLED (cont. ) unloaded quality factor: cavity coupling: klystron pulse width: compression ratio: effective power gain: compression efficiency: Q 0 = 105 b = 5. 0 Tk = 3. 5 ms Cr = 4. 27 Gp = 2. 60 hc = 0. 610
SLED (cont. ) SLAC S-band structure: structure filling time: Tf = 0. 82 ms group velocity profile: vg = vg 0 (1 – 0. 681 z/L) Energy gain of beam w/ SLED vs. injection time, normalized to energy gain w/o SLED
SLED II is a variation on SLED developed by P. B. Wilson, Z. D. Farkas and R. D. Ruth in 1990 (employing a concept independently published in by A. Fiebig and C. Schieblich in 1988). It replaces the resonant cavities with a pair of resonant shorted delay lines, in order to achieve a flat output pulse. The compressed pulse duration is one round-trip delay time, and the pre-phase flip storage time is a finite multiple thereof. Overmoded circular waveguide operated in the low-loss TE 01 mode is employed to minimize transmission losses. The reflection coefficient of the coupling irises is optimized for the anticipated compression ratio, and motorized plungers are used at the shorted ends for tuning. X-band (11. 424 GHz) SLED-II systems were developed in the NLC program and are still used in the NLCTA and ASTA facilities for high-gradient accelerator research.
SLED II (cont. ) electric field pattern
SLED II (cont. )
BPC (Binary Pulse Compression) Binary Pulse Compression, invented by Z. D. Farkas in 1986, combines the outputs of a pair of sources through a hybrid, using phase patterns encoded on the source outputs to direct the combined power out different ports during the leading and trailing halves. The first half is delayed through a long trombone waveguide (again using low-loss circular TE 01 mode) to make them concurrent. This splitting and stacking can be repeated in multiple stages, with proper phase coding and additional hybrids and delay lines. Working back from the accelerator, the delay of the first line must equal the desired compressed pulse duration, and each additional stage must double in length.
BPC (cont. )
BPC (cont. ) SLAC high-power BPC demonstration experiment using one source (~1990). Components were lossy and primitive compared to what we developed in the ensuing years.
DLDS (Delay Line Distribution System) Binary Pulse Compression is a technique of effective pulse compression invented in 1994 by H. Mizuno and Y. Otake of KEK. It is only applicable to very long linacs, such as those of a linear collider. Like BPC, it requires multiple sources and uses phase pattern encoding to direct the combined power and low-loss transmission waveguide. Unlike with BPC, DLDS waveguides do not double back to the same location; rather, they run upstream to feed a distant set of accelerator structures. The length of each line is reduced by the time-of-flight of the beam. Intervening structures, between feeds of a DLDS are fed by interleaving systems.
DLDS (cont. ) 1 2 3 4 … beam … … … Combined power is directed sequentially along different paths via relative phasing.
Variations and Development Through the 90’s and into the new millenium, a number of variations on and adaptations of these basic schemes were explored. In the process of this R&D, an arsenal of novel high-power waveguide components were developed, including mode converters, bends, phase shifters, hybrids, etc.
Questions Specifically, we're hoping to understand what the differences are between SLED, SLED II, Binary Pulse Compression, and Delay Line Distribution Systems. This has been addressed in the preceeding slides. What are the limitations of each in terms of pulse shape, efficiency, and maximum peak power? Basic SLED is the only technique with intrinsic pulse shape issue, though it can be addressed via amplitude (or phase) manipulation, at the expense of efficiency. What results have been achieved using these methods? We have produced a 500 MW compressed pulse with our dual-moded SLED-II. Is this capability tunable over a bandwidth? Our development has all been aimed at fixed frequency. Energy storage cavities/delay lines can be tuned. Bandwidth of components, e. g. hybrids, may be the limiting factor. How consistent is this process if we're considering two separate klystron / compression systems? I’m not sure what is meant here. Many klystrons/SLED’s are used in the SLAC linac, and three SLED-II sytems are used in the NLCTA.
1. 2 ~22. 5 m 300
Dual-Moded 8 x 8 DLDS The current rf system design achieves an effective pulse compression ratio of 8 with an 8 -feed multimoded DLDS utilizing TE° 01 and TE° 12. Each feed powers six consecutive 0. 9 m structures.
Dual Moded SLED II Input Stepped Taper and End TE 01/TE 02 Mode Converter S=
Demonstration of Multimoded Reflective Delay Line TE 01 TE 02 TE 01 Measured Delay through 75 feet of WC 475 waveguide terminated with a flat plate. The round trip delay time is 154 ns. Measured delay through 75 feet of WC 475 waveguide terminated with a TE 01 -TE 02 Mode converter. The round trip delay time is 320 ns. S. Tantawi
3 Options for X-Band Pulse Compression: SLED-II Tk=1. 5 ms, Tp=150 ns Cr=10 D = 4. 750”, L = 71’ 1. 5” (21. 679 m) Q 0= 1, 291, 300, Td = 150 ns, s = 0. 7644 G = 5. 417, h = 0. 542 3 -Bin SLED-II Compressed pulse is 3 bins wide, flattened by amplitude modulation. D = 4. 750”, L = 23’ 8. 5” (7. 2263 m) Q 0= 1, 135, 700, Td = 50 ns, s = 0. 925 Last 3 bin amplitudes: [-. 5388 -. 7608 -1] G = 4. 574, h = 0. 457 Amplitude Modulated SLED TE 0, 1, 10* cavities w/ D = 5. 000”, L =5. 3381” (13. 559 cm) Q 0= 109, 169 b = 4. 73, Tc = 530. 86 ns Flipped amp. swept from -0. 349 to -1. 000 G = 3. 498, h =0. 350 * SLAC uses TE 0, 1, 5 @ S-band, and Spring-8 uses TE 0, 1, 15 @ C-band.
X-Band Pulse Compression Redux 473 k. V, 0. 92 k. A, 600 ns 4 65 MW 600 ns CP 3 Cr = 3 hi = 1 2 1 Dual-moded transmissive delay line. →No tuning needed (use LLRF phase). → Intrinsic efficiency is 100%. TE 12 (260 MW) TE 01 (520 MW) 1 unit delay (200 ns) ~30 m 3 260 MW 200 ns • Four klystrons are combined through a cross potent superhybrid. • During the 1 st time bin, power combines through path 1 into the delay line. • Upon returning through the second pipe of the delay line, the power is combined with that of the 2 nd time bin, coming along path 2, and sent through the delay line again in a different mode. • The returning power is then split and feeds the accelerator simultaneously with the 3 rd time bin power directed along path 3. C. Nantista ’ 08
- Slides: 22