BCD-Main Linac

1. BCD-Main Linac RF-Source: Modulator, Klystron Cavity: High Power Input Coupler, Frequency Tuner Cryomodule: Configuration RF Distribution: Configuration Cryogenic System: Configuration Tunnel: Curvature, Layout

2. Modulator • Baseline • To power the 10 MW MBK’s, the baseline choice is to use the bouncer-compensated pulse transformer modulator that was developed initially by FNAL and industrialized by DESY. • The units perform well, however, expensive, multi-ton, oil-filled transformers and susceptible to single-point failures. • FNAL recently designed a lower-cost, more-reliable, pulse-transformer modulator that has 5 ms pulse capability for the Proton Driver projenct. • DESY continues to work with industry to improve the reliablity of their modulator design for the XFEL. • Alternatives • The main alternative is a Marx-style generator which is being developed at SLAC. • This modulator has the design with built-in redundancy and it shoud be easier to mass produce and repair compared with the baseline modulator. • Other approaches are a direct-switch by DTI and a DC-to-DC converter by LANL. • If a TDR-like tunnel layout were adopted with the modulator separated from the klystron by up to 2.5 km, the transport and impedance matching(cable to klystron) of the 120 kV pulses would require further development. (Marx-style is the easiest way.)

3. Klystron • Baseline • The 10 MW Multi-beam klystrons (MBK’s) being developed by Thales, CPI and Toshiba are the baseline choice. • If the MBK’s do not meet availability requirements, the commercial, single beam, 5MW tube from Thales could be used. This has been the ‘work-horse’ for L-band testing at DESY and FNAL. • Alternatives • 10 MW sheet-beam klystron by SLAC to reduce cost • 5 MW inductive output tube by CPI to improve efficiency • 10 MW, 12 beam MBK by KEK to reduce the modulator voltage and the modulator plus klystron cost.

4. High Power Input Coupler (1) • BCD Choice • Twin cylindrical window architecture of the TTF-III coupler Pro: • Lengthly experience on TTF: (~100,000 coupler hours) . • Demonstration of use with a cavity at 35 MV/m on CHECIA. • Tested at a power of 1MW, 1.3ms pulse in TW mode. • In principle, then this power coupler would be sufficient even if the cavities were to be run at 35 MV/m and would meet, at least in TW mode, the needs of 29 cell superstructure at 35 MV/m. Con: • The present unit cost is prohibitive. However, the couplers have only been built in small numbers. • The experience with conditioning indicates that the conditioning time is rather long. The conditioning issue is currently under study at Orsay in the DESY-LAL collaboration. A period of ~two years will be necessary to complete the conditioning and industrialization study. Potential Modification • Increase in the diameter of the cold assemble (from 40 mm to 62 mm). Technical benefit of pushing multipactor levels to higher powers Of interest in case of a choice of higher gradient (~45 MV/m)

5. High Power Input Coupler (2) • Alternatives • The coupler designs incorporating two disk type window coupler • The “capacitive” disk window coupler • The “TRISTAN” like window coupler • The TW60 coupler • The AMAC window coupler Pro: • Relatively free from multipactor. • Mechanically easy to fabricate and therefore cheaper. • Thin disk windows can be positioned at low value of the SW electric field. • Disk windows should be relatively easy to braze into the coupler. Con: • The current version of the capacitive coupler cannot be DC biased. • The present capacitive and TRISTAN like couplers have no possibility to have their external Q variable. • Disk ceramic are in the “line of sight” of the cavity beam pipe. Seen to be a problem on the early CEBAF linac design The co-axial version may be less problematic because of the reducing the solid angle presented to the x- ray or electrons which might impinge on the ceramic. • Too early to estimate the cost impact.

6. Frequency tuner (1) • ILC requirements • Coarse tuning range:500kHz (1.6mm at 315 Hz/mm) • Coarse resolution:<5Hz • Fast tuning range (static at 2K): f = 2  K  E^2 (factor 2 for dynamic operation overhead) 2.5kHz (for K=1 Hz/(MV/m)^2 cavity at 35 MV/m) 3.2kHz (for K=1 Hz/(MV/m)^2 cavity at 40 MV/m) 4.0kHz (for K=1 Hz/(MV/m)^2 cavity at 45 MV/m) • The requirement on the fast running range is not well known at this point. • Dedicated experiments are needed to define the actually required fast tuning range. • Options under consideration: • Original Saclay/TTF tuner Used in TTF/VUVFEL for several years. No fast tuning element, though a piezo actuator added for proof-of-principle tests of Lorentz-force detuning. Range of the fast tuner <500Hz. • Modified Saclay tuner Similar to the original Saclay tuner. Located at one end of cavity. Incorporated piezo actuators (~1kHz tuning range). • INFN/DESY blade tuner No fast actuator(1st version). Tested at TTF with the superstructure. Located around the LHe vessel. The recent version includes piezo actuator(~1kHz). • TJNAF Renascence tuner Incorporated piezo actuator(~1kHz). • KEK slide jack tuner Designed for the 35MV/m cavity. The motor is placed at room temperature outside the vacuum vessel. • KEK coaxial ball screw tuner Designed for the ICHRO cavity. The motor and piezo are placed at intermediate temperature inside the vacuum vessel.

7. Frequency tuner (2)

8. Frequency tuner (3) • Risk and reliability • All but KEK slide jack tuner have a cold drive motor inside the vacuum vessel. • The main risk is a failure of the motor, the fast actuator or the gearing. • All designs with cold drive can use the same type of motor, gearing and fast actuator, so that there is no principle difference in risk and reliability between these tuner designs. • The reliability of a single cold drive might not be sufficient. The tuners will either have to be made very reliable (probable via redundancy) or their failure prone components made replaceable without warm-up. • Options for improving the reliability • Redundant motor and piezo, if inside of vacuum vessel • Improved design with highest reliability for motor and/or piezo, if inside of vacuum vessel • Warm motor • Fast actuator • Piezo actuator • Detailed studies have been done to verify pulsed cryogenic operation in radiation environment. • Magnetrostrictive actuator • Similar size to piezo, so can be used instead of piezo. • Has a significant larger stroke than a piezo at 5K, produces less heat and might have a higher life time and higher tolerance for preload change than piezo. • Base line • No existing tuner design fulfill the specification on fast tuning range above 30 MV/m. • The tuner needs to provide 500 kHz slow tuning range and more than 3kHz fast tuning range. • Required R&D • Tuner design for 40+MV/m operation and prototype tests including of Lorentz-force detuning at highest fields with BCD cavity. • Reliability(MTBF) studies of motor/gearing/piezo/magnetostrictive actuator, including failure mechanisms and improved estimate of requirements. • Performance of magnetrostrictive actuator. • Cavity design with small Lorentz-force detuning. • Cost estimation for external motor.

9. Frequency tuner (4) • Close to BCD • Modified Saclay tuner Pros: • Relative simple and compact design • Redundant design for piezo element • Original Saclay tuner was tested in detail Cons: • Maybe difficult to increase fast tuning range • Redesign needed with increased fast tuning range • Poor maintainability of stepping motor and fast actuator R&D necessary: • Design with increased fast tuning range • Fast actuator R&D • Prototype tests with Lorentz-force compensation at 35 MV/m • Verification of sufficient MTBF for cold motor • INFN blade tuner Pros: • Compact design (not at cavity end) • High stiffness • Tested (without fast actuator) • Relative easy to increase fast tuning range Cons: • Redesign needed with increased fast tuning range • Poor maintainability of stepping motor and fast actuator R&D necessary: • Design with increased fast tuning range • Fast actuator R&D • Prototype tests with Lorentz-force compensation at 35 MV/n • Verification of sufficient MTBF for cold motor

10. Frequency tuner (5) • Alternatives • TJNAF renascence tuner • KEK slide jack tuner Pros: • Motor outside of vacuum vessel (inexpensive motor) • Piezo can be replaced (cryostat warm-up required) • High stiffness Cons: • Feed-through to outside needed (penetration of shields and vacuum vessel) • Some static losses (0.05 W?) • Redesign needed with increased fast tuning range • Poor maintainability of fast actuator; no redundancy R&D necessary: • Design with increased fast tuning range • Fast actuator R&D • Prototype tests with Lorentz-force compensation at 35 MV/m • KEK coaxial ball screw tuner Pros: • Wide tuning range • Compact design with common technology • Cost effective • High stiffness • Maybe access to piezo (warm-up required; need to pass through all thermal shields) Cons: • Poor maintainability of stepping motor • Poor maintainability of fast actuator; no redundancy • Heavy weight • Some static losses • Redesign needed with increased fast tuning range R&D necessary: • Choice of coating material for balls • Weight reduction • Design with increased fast tuning range • Fast actuator R&D • Prototype tests with Lorentz-force compensation at 35 MV/n • Verification of sufficient MTBF for cold motor

11. Gradient • Baseline • The WG5 recommendations call for TESLA-like cavities to be used. • Qualified to operate at a gradient of at least 35 MV/m with a Q>0.81010 in CW tests. Cavities not meeting these requirements would be rejected or reprocessed. • Only a small fraction of cavities and cryomodules would be pulsed-power test. • With such screening, a 31.5 MV/m gradient and Q of 11010 would be achieved on average in a linac made with eight-cavity cryomodules. • For a future upgrade, the WG5 recommends that cavities of the low-loss or reentrant type be used and that they be qualified to at least 40 MV/m with Q>0.81010 in order to achieve 36 MV/m and Q= 11010 on average in the linac. • Alternatives • The linac cost is a weak function of gradient in the 30-50 MV/m range, and operating close the ultimate 45-50 MV/m gradient limit would prevent extending the machine energy by lowering the beam current (and depending on the cooling overhead, lowering the machine repetition rate). Thus, a better strategy would be to design for a gradient around 30 MV/m, and if the cavities that are eventually installed perform better than the initial requirement, use this capability to extend the machine energy reach (e.g., up to 750 GeV if 45 MV/m operation is eventually achieved). • The WG5 ACD gradient recommendation for 500 GeV operation is the same as given for the BCD upgrade.

12. Cryomodule: Configuration • Baseline • A reasonable baseline rf unit is a 10 MW klystron driving 24 TESLA cavities. • The cavities would be divided into three cryomodules instead of two. • This is the configuration that has been used and will continue to be used for several years. • There is no significant cost savings with longer cryomodules. • The cavity gradient variation can be more efficiently dealt with if there are less cavities per cryomodule. • Every fourth cryomodule in the linac would include a cos(2q)-type quadrupole with horizontal and vertical corrector windings. • The quad helium vessel would be supported above by the 300 mm diameter gas return pipe. • The quad would be located below the center (fixed) post, and attached to its upstream face would be a BPM. • The TDR cavity spacing of 283 mm was optimized based on the flange connection and bellow scheme. • The linac packing fraction would be about 75 %. • For 500 GeV operation, 328 such rf units would be required. Overall, 71 % of the peak rf capability would be transformed into beam power. • Alternatives • Having up to 12 cavities per cryomodule instead of 8 to reduce the number of inter-connections. • Shortening the distance between cavities to 250 mm like JLab, or as close as possible to the 180 limit from cavity cross talk and heat losses at the transitions. • Instrumenting the HOM readouts to provide a measure of the average beam position in the cryomodule. The cryomodules would either be moved manually during down periods or equipped with remote-controllable mover for beam centering. • Putting the quad and BPM in a separate cryo-section for better stabilizing them vibrationally. No corrector magnets for moving quad and BPM independently from cavities. • Putting the movers on the middle support post of the gas return pipe to allow adjustment of the quad and BPM. • Reducing the quad aperture by half (~35mm) to allow the use of superferric quads, which will likely have more stable magnetic centers with respect to quad shunting.

13. RF-distribution: Configuration • Baseline • TDR-like distribution system that includes a circulator in each cavity feed followed by a tuner(three-stub or E-H type) to allow control the cavity phase and Qext. • DESY uses off-the-shelf components for the distribution system. • Alternatives • The circulator is the largest cost component at about 25 % of the rf distribution cost. • There are several alternative distribution scheme that eliminate the circulator. • More precise cavity-to cavity phasing. • Harder to deal with the variation in the maximum cavity gradients. • In the event of rf breakdown in a coupler or cavity, these schemes would allow some fraction of the reflected power to propagate to the other cavities. • KEK plans to test such a distribution scheme at the STF in the next year.

14. Cryosystem: Configuration • Baseline • The refrigerator spacing would depend on the choice of operating gradient (the spacing is about 5 km in the TDR). • The length of the 2K, two phase lines depends on the tunnel slope. • For the slopes up to 0.3 mrad, the 167 m long, 8.5 cm diameter lines specified in the TDR could be used. • For larger slopes, shorter lines. • For 1-4 mrad, a canal-like system. A laser-straight tunnel would have a 3 mrad slope. • The maintenance length is half of the refrigerator spacing. • The maintenance length is the length that would need to be warmed up to repair a cryomodule. • Alternatives • The thermal cycling long string of cryomodules will be slow and may cause vacuum leaks. • To reduce the maintenance length, U-tubes and turnarounds can be included. • If such sections (each 1.5 m long) were installed every 500 m, the number of cryomodules thermally cycled would be reduced by a factor of five. • The warm-up time is reduced by a factor of two. • The cool-down time is reduced by a factor of 10.

15. Tunnel: Curvature • Baseline • Until on-going beam dynamics simulation show otherwise, the linac will follow the curvature of the earth, unless a site-specific reason (cost driven) dictates otherwise. • Alternatives • Laser-straight linac or one constructed from straight-line segments. • Final choice depends on the cost consideration. • Required R&D • Aggressive beam dynamics simulations of emittance preservation (beam-based alignment) for the continuous curved linac, including tolerance studies. • Studies of viable and cost effective cryogenic solution for a tilted linac.

16. Tunnel: layout • Baseline • The rf sources are to be distributed along a second tunnel (or surface gallery) that runs parallel and nearby to the beam line tunnel. • The sources would not be subjected to radiation and could be accessed for repairs while the machine is running. • Alternatives • The beamline is in a near-surface tunnel (<30 m deep) and the modulators, sans transformers, are clustered in surface building located every 5 km (the beamline tunnel contains the modulator transformers and klystrons). • For both case, locating the beamline near the surface would allow easier access and shorten the power and cooling distribution lines that connect to the surface. • The main disadvantages would be larger ground motion and limited site availability.