LEP Couplers..a Troubled Story of a Success HPC2002, Jefferson Lab, October 30 th, 2002 R. Losito, CERN 1 1
Overview & development: specifications, problems, solutions Operation: field equalization, trip statistics. Conclusions & Acknowledgements HPC2002, Jefferson Lab, October 30 th, 2002 R. Losito, CERN 2
First attempt: 500 MHz coupler directly in the cell, magnetic coupling (loop), Problems: multipacting, field emission, (test of a cavity in PETRA, Desy). Performance: up to 70 kw, (3.4 MV/m in the cavity) HPC2002, Jefferson Lab, October 30 th, 2002 R. Losito, CERN 3
HPC2002, Jefferson Lab, October 30 th, 2002 R. Losito, CERN 4
HPC2002, Jefferson Lab, October 30 th, 2002 R. Losito, CERN 5
Initial specifications: Qext=1.75 10 6 : matched for 6 MV/m, 3.4 mamp/beam, Φs=-32.8º Power: ~60 kwatts CW HPC2002, Jefferson Lab, October 30 th, 2002 R. Losito, CERN 6
Starting point: Warm LEP cavity power coupler: 103 mm diameter cylindrical window, nominal power 120 kw; No harmful multipacting up to the nominal power of 120 kwatts. Conditioning on room temperature cavities up to 180 kw without special problems and no degradation of performance during operation HPC2002, Jefferson Lab, October 30 th, 2002 R. Losito, CERN 7
First design at 350 MHz open-ended coaxial coupler Characteristic impedance: 50 Ω; fixed coupling; Outer/inner conductor diameter: 103/44.8 mm Outer conductor "extended", to bring the ceramic outside the cryostat Extension cooled with Ghe; HPC2002, Jefferson Lab, October 30 th, 2002 R. Losito, CERN 8
FOTO COUPLER FISSO 50 OHM HPC2002, Jefferson Lab, October 30 th, 2002 R. Losito, CERN 9
FOTO COUPLER FISSO 50 OHM HPC2002, Jefferson Lab, October 30 th, 2002 R. Losito, CERN 10
FOTO COUPLER FISSO 50 OHM HPC2002, Jefferson Lab, October 30 th, 2002 R. Losito, CERN 11
Meanwhile, specifications changed, and a new variable coupler was designed to cope with: Qext=1.75 10 6 : matched for 6 MV/m, 7 mamp/beam, Φs=-32.8º Power: ~125 kwatts CW HPC2002, Jefferson Lab, October 30 th, 2002 R. Losito, CERN 12
Advantages of a variable coupler: Possibility to work easily under very different conditions (Qext = 3 10 9 5 10 5 ) : perfect match to all the beams, possibility to measure the Q(Eacc) characteristic of the cavity at any time, Possibility to compensate random errors in the distribution of Qext seen with fixed couplers. (equalization of fields for the 8 cavities per klystron). HPC2002, Jefferson Lab, October 30 th, 2002 R. Losito, CERN 13
CHARACTERISTICS OF THE VARIABLE COUPLER: 50 Ω, λ/4 choke as an internally folded 25 Ω stub, due to limited space in the tunnel, All the other dimensions unchanged HPC2002, Jefferson Lab, October 30 th, 2002 R. Losito, CERN 14
HPC2002, Jefferson Lab, October 30 th, 2002 R. Losito, CERN 15
HPC2002, Jefferson Lab, October 30 th, 2002 R. Losito, CERN 16
HPC2002, Jefferson Lab, October 30 th, 2002 R. Losito, CERN 17
HPC2002, Jefferson Lab, October 30 th, 2002 R. Losito, CERN 18
HPC2002, Jefferson Lab, October 30 th, 2002 R. Losito, CERN 19
Kovar ring 10 kω Ti layer 10 MΩ Ti layer HPC2002, Jefferson Lab, October 30 th, 2002 R. Losito, CERN 20
Results (in few words): During a test on a cavity at 4.5K, an important leak on cavity vacuum appeared; Inspection revealed a hole on the extension with evidence of burnings. HPC2002, Jefferson Lab, October 30 th, 2002 R. Losito, CERN 21
Investigations on causes: Main suspect: gas discharge from cold surfaces Measurements of multipacting levels where refined adding pickup for electron current measurements and flange for vacuum gauge; HPC2002, Jefferson Lab, October 30 th, 2002 R. Losito, CERN 22
HPC2002, Jefferson Lab, October 30 th, 2002 R. Losito, CERN 23
Investigations on causes: Main suspect: gas discharge from cold surfaces Measurements of multipacting levels where refined adding pickup for electron current measurements and flange for vacuum gauge; Multipacting simulations confirmed levels measured on cavities; Harmful levels coming from one point multipacting on outer conductor. On variable coupler, also 2-point multipacting in the 25Ω choke. HPC2002, Jefferson Lab, October 30 th, 2002 R. Losito, CERN 24
Actions to reduce multipacting: review of procedures: Pre-conditioning on warm cavity; HPC2002, Jefferson Lab, October 30 th, 2002 R. Losito, CERN 25
HPC2002, Jefferson Lab, October 30 th, 2002 R. Losito, CERN 26
Actions to reduce multipacting: review of procedures: Pre-conditioning on warm cavity; assembly in clean room; HPC2002, Jefferson Lab, October 30 th, 2002 R. Losito, CERN 27
Actions to reduce multipacting: review of procedures: Pre-conditioning on warm cavity; assembly in clean room; Conditioning on cold cavity; HPC2002, Jefferson Lab, October 30 th, 2002 R. Losito, CERN 28
HPC2002, Jefferson Lab, October 30 th, 2002 R. Losito, CERN 29
Actions to reduce multipacting: review of procedures: Pre-conditioning on warm cavity; assembly in clean room; Conditioning on cold cavity; Result: De-conditioning: reappearance of levels already conditioned when going back to lower power. Complete de-conditioning after switching off RF for several hours HPC2002, Jefferson Lab, October 30 th, 2002 R. Losito, CERN 30
Stronger actions to reduce multipacting were needed: 1 st idea: Multipacting levels depend on voltage seen by electrons; for the same incident power, the voltage (electric field) at the outer conductor is reduced by increasing the characteristic impedance. 1 st action: go to 75Ω antenna (diameters 103/30 mm) HPC2002, Jefferson Lab, October 30 th, 2002 R. Losito, CERN 31
HPC2002, Jefferson Lab, October 30 th, 2002 R. Losito, CERN 32 LEP Couplers...a Troubled story of a success TEM r1 r2 = = 1 2 0 1 2 0 ) ( 0 60ln ; ln 1 r r Z r r E e r E E kz j ωt
Stronger actions to reduce multipacting were needed (cont'd): 2 nd idea: remove low impedance choke: 25Ω line introduced new multipacting levels below 40 kwatts; 2 nd action: return to a fixed coupler, even if this meant loss of flexibility. HPC2002, Jefferson Lab, October 30 th, 2002 R. Losito, CERN 33
HPC2002, Jefferson Lab, October 30 th, 2002 R. Losito, CERN 34
Stronger actions to reduce multipacting were needed (cont'd): 3 rd idea: Multipacting is originating from gases adsorbed on the outer conductor wall; conditioning only displaces these gases, that are re-adsorbed on the cold surfaces: conditioning could be never-ending; 3 rd action: find and suppress main sources of adsorption HPC2002, Jefferson Lab, October 30 th, 2002 R. Losito, CERN 35
Stronger actions to reduce multipacting were needed (cont'd): 3 rd action: find and suppress main sources of adsorption 1 st source: Ceramic bake-out "in-situ"; 2 nd source: copper-plated extension: refinement of the surface (suppress weldings forging: smoother surface); copper plating (10µm) by sputtering instead of galvanic deposition (titanium layer in between, 0.5µm). (NEG on extension?) HPC2002, Jefferson Lab, October 30 th, 2002 R. Losito, CERN 36
Stronger actions to reduce multipacting were needed (cont'd): 4 th idea: DC Bias: According to simulations both positive and negative DC bias are effective. For LEP, -2.5 kv were applied to inner conductor, with a nearly "magic" result. HPC2002, Jefferson Lab, October 30 th, 2002 R. Losito, CERN 37
HPC2002, Jefferson Lab, October 30 th, 2002 R. Losito, CERN 38
HPC2002, Jefferson Lab, October 30 th, 2002 R. Losito, CERN 39
Operation Operation conditions: Initial Eacc: 6 MV/m (10 MV voltage), year 2000, up to 9 MV/m Current: 3 to 8 mamps Ultimate power: >100 kwatts CW Cavities equipped: 288 HPC2002, Jefferson Lab, October 30 th, 2002 R. Losito, CERN 40
Operation Operation was complicated by: 8 cavities/klystron Fixed couplers, with Qext spread of ±20% around nominal value; Spread in field for cavities on the same klystron HPC2002, Jefferson Lab, October 30 th, 2002 R. Losito, CERN 41
Operation Field distribution in waveguides different for different beams: Decision to reduce the effect on worst cases with fixed λ/4 stubs in waveguides; Fine tune of waveguide length (less than 3º of RF wavelength of difference from nominal length); HPC2002, Jefferson Lab, October 30 th, 2002 R. Losito, CERN 42
Operation Protection of Main Coupler (interlocks): MC vacuum Arc Detector Electron current monitor Monitor of cooling air flow Temperatures on window and extension HPC2002, Jefferson Lab, October 30 th, 2002 R. Losito, CERN 43
Operation Trip statistics: Only MC vacuum significative: all the others often coming from electronic noise: In average, MC vacuum was responsible of 4% of downtime of the total downtime due to RF (~25% of total for machine). Any outburst was a big risk, but finally there was not even one window breakdown. HPC2002, Jefferson Lab, October 30 th, 2002 R. Losito, CERN 44
Conclusions At the end, Couplers for LEP were a success, but they could have led the whole project to disaster. Solutions to problems were found resorting to experts from different disciplines (RF, Vacuum, metallurgy, thin films etc ). At the end the couplers could easily withstand the nominal power (120 kw), and several were tested up to 4 times this power. Main problem during operation: field spread in cavities due to distribution in values of Qext HPC2002, Jefferson Lab, October 30 th, 2002 R. Losito, CERN 45
Acknowledgents Ernst Haebel, Joachim Tuckmantel,Sergio Calatroni and Olivier Brunner told me the story. A lot of people worked on LEP couplers ( Mircea Stirbet, HP Kindermann, Daniel Boussard, Gunther Geshonke, Jean-Pierre Boiteux, Pete Brown) from RF groups, but also from several other groups (in particular from the actual LHC/VAC and EST/SM group). There is a huge quantity of written material, thanks to those who wrote everything about their work. HPC2002, Jefferson Lab, October 30 th, 2002 R. Losito, CERN 46