Update on the impedance studies of the SPS wirescanners Mauro Migliorati, Benoit Salvant and Carlo Zannini for the impedance team Many thanks to Hannes Bartosik, Nicolas Critin, Heiko Damerau, Bernd Dehning, Jonathan Emery, Federico Roncarolo, Ray Veness, and Christophe Vuitton

Still work in progress, and a lot more studies are needed Many thanks for the very useful comments and suggestions at and following the previous MSWG and LIU-SPS meetings.

Agenda Objectives New design Current design Previous measurements Summary

Objectives Need to find solutions for mechanical design of the new prototype Significant impedance issues found with the proposed design However current design is not transparent either… Which is best? Wire usually breaks during measurement Not so critical in parking position? simulation also of the wire when it gets closer to the beam Does it make sense to try to protect the wire in parking position? Several options were on the table after the last meeting Ferrites as usual All mode damper (Fritz) Change the inner pipe to two plates (Heiko and Hannes after MSWG) Remove completely the inner pipe (Ray and Bernd)

Agenda Objectives New design Current design Previous measurements Summary

New model for SPS CATIA model from Nicolas Critin Model heavily simplified to be able to mesh the geometry Wire and gaps very difficult to mesh Same as LHC but bigger aperture SPS geometry + Use of more realistic 316 LN conductivity (1.35 MS/m) instead of Steel 1008 + clarification of Linac/Circuit convention for both wakefield and eigenmode solver First resonant mode now at Rs=40 kLinacOhm (20 kOhm) in parking position with around f=160 MHz (Q=650) without ferrites. Still significant power loss and impedance contributions in parking position. What about when the wire is scanning the beam?

120 degrees (parked OUT) 60 degrees (crossing beam) 0 degrees (parked OUT) When the wire is moving, the shunt impedances are in the 15 to 20 kOhm range.  >1kW of dissipated power Once the wire is in the pipe, strong fields are close to the wire. Is the wire really protected Shunt impedance in LinacOhm 160 MHz ~400 MHz 900 MHz

New wire scanner Dangerous mode below 200 MHz due to the cut in the pipe. The mode strongly depends on the geometries around the pipe (high sensitivity of the impedance versus small geometrical details) The mode is weakly dependent from wire and fork materials Possible solutions damping of the mode using ferrites  need input from BI to know what is feasible (first proposals rejected) Metal Rods ending on feedthrough (proposed by Fritz: to be simulated, but difficult also with the constraints of wire scanner) Removing the inner pipe?

Removing the inner pipe 120 degrees (parked OUT) 60 degrees (crossing beam) 0 degrees (parked OUT) Factor 2 increase in shunt impedance Strong correlation with angle All fields located close to the wire 300 MHz 650 MHz

Agenda Objectives New design Current design Previous measurements Summary

Current wire scanner More realistic current design: H fork has smaller width Should implement taper Ferrites (random location)

Present wire scanner: longitudinal impedance The longitudinal impedance significantly depends on the wire material. Simulation model account for a carbon wire (with steel wire, the shunt impedance is much larger). There are no ferrites. Measurements F. Roncarolo et al. 2003 Measurement and simulations show a good agreement in the spectral behaviour: strong modes around 300 MHz and 700 MHz.

Beam induced heating on the wire worst case scenario angle=0 (close to parking) Mode number f [MHz] Q R [Ω] Total losses [W] Losses on the wire [W] Mode 1 284 670 3600 224 44 Mode 7 658 2000 8500 196 52 Mode 9 681 1750 22072 433 160 25 ns beam p = 450 GeV 288 bunches N=1.4e11 ppb not excited at injection angle=40 (in beam) Mode number f [MHz] Q R [Ω] Total losses [W] Losses on the wire [W] Mode 1 284 1000 6000 373 73 Mode 2 290 7200 444 81 Mode 7 657 2580 25280 583 155 Mode 10 693 2510 23187 454 172 25 ns beam p = 450 GeV 288 bunches N=1.4e11 ppb not excited at injection  Potentially large losses on the wire A wire breaking occurs with these beam during measurements Strongest heating from mode at ~700 MHz (not excited at injection)

Adding “random” ferrites (still Steel wire and no taper)

With injection beam (Gaussian) With top energy beam (Gaussian) Modes at 700 MHz heavily damped, but not so much the first mode at 300 MHz

Agenda Objectives New design Current design Previous measurements Summary

Observations from 2003 F. Roncarolo et al. DIPAC 2003 Wire heating stronger with shorter bunch length (no ferrite) Indicates that the 2nd mode at 700 MHz is probably very important for the heating

Wire diagnostics (Jonathan Emery) Maximum of the current is later: transient of the dissipation or real? In simulations, larger current inside the wire when the beam is passing through the fork

Current inside the wire We do not use this data usually, not sure we can trust it, requires simple benchmarks Seems to be strongly affected by mesh issues (PEC mesh cells) Due to oscillating Wake fields Beam outside of the fork loop Current in the wire (for bunch of 1nc) Beam inside the fork loop Time in ns Could this explain that there is more heating after the beam is passed? Needs to check if there is voltage data with an IN/OUT scan with beam at injection

Agenda Objectives New design Current design Previous measurements Summary

Where are we? Models and assumptions more refined Need to understand clearly the constraints of the new design Try to see what solution can be implemented: Current ideas: dielectric fork (as BSRT), all mode coupler (better than the wire…), ferrite at appropriate location. Dissipated power predicted in the kW range if no solution is applied Heating of the wire for the current scanner increases with Smaller bunch length Approaching wire Beam inside the fork? Need to implement the taper Bench measurements on the current and prototype scanners with moving fork would help as the meshes are very unstable with thin wire and small gaps.

Current SPS model

Remaining issues Can we trust the simulated current curves?

New Model (SPS) CATIA model from Nicolas Critin Model heavily simplified to be able to mesh the geometry Only one longitudinal gap (2.3 mm) Wire and gaps very difficult to mesh

Solutions with ferrite? With ferrite in green With less ferrite in red Re(Zlong) in Ohm Re(Zlong) in Ohm Frequency (GHz) Frequency (GHz) To be checked if these solutions can be implemented To be compared with the current wire scanner design

Proposal of Heiko and Hannes Frequency in GHz Longitudinal impedance in Ohm  Better than the initial solution, still worse than the current design  However, solution not feasible for BI (“the fork should rotate completely around the axis”)

Agenda New design Current design Power loss Mode geometry Previous measurements Comparison with wall current monitor Summary

Current design (also simplified) Wakefield simulations Very high Q resonances, solver Resonant frequencies are much lower for the new design

Subtleties of the current design Taper in 416 H vertical fork much wider than horizontal fork  resonance frequency change 416H Wire out 416 Wire in

Agenda New design Current design Power loss Mode geometry Previous measurements Comparison with wall current monitor Summary

Comparison for the resonant modes (no ferrites) Frequency Shunt impedance (kOhm) Q Fields close to the wire? Power loss at 450 GeV (from LHC spectrum) 140 MHz 250 2500 No 60 kW 470 MHz 15 3500 Yes (partly H) 0.3 kW New design Frequency Shunt impedance (kOhm) Q Fields close to the wire? Power loss at 450 GeV (from LHC spectrum) 290 MHz 18.5 1660 Yes 1.3 kW 658 MHz 55 7000 0.8 kW 682 MHz 146 5500 2.4 kW 1020 MHz 210 12000 0.5 kW Current design  Hopefully there are ferrites to partially damp these modes!  Need to do that also for the SPS spectrum at injection energy?

Spectrum at injection From C. Zannini 72 8 72 8 72 8 72 25ns buckets First lobe of the beam spectrum after around 700 MHz

Comparison for the resonant modes (no ferrites) modes also significant at injection? Frequency Shunt impedance (kOhm) Q Fields close to the wire? Power loss at 450 GeV (from LHC spectrum) 140 MHz 250 2500 No 60 kW 470 MHz 15 3500 Yes (partly H) 0.3 kW New design Frequency Shunt impedance (kOhm) Q Fields close to the wire? Power loss at 450 GeV (from LHC spectrum) 290 MHz 18.5 1660 Yes 1.3 kW 658 MHz 55 7000 0.8 kW 682 MHz 146 5500 2.4 kW 1020 MHz 210 12000 0.5 kW Current design  Hopefully there are ferrites to partially damp these modes!

Simulated power loss with 50 ns SPS beam with ferrite as a function of bunch length Carlo Zannini Still between 50 and 100 W in the ferrite Is that ok with the duty factor of LHC beams in the SPS? Would cooling be needed and is it possible to cool there?

Power loss (with ferrites) HL-LHC New design modified design (Heiko) Current design Predicted for post LS1

Agenda New design Current design Power loss Mode geometry Previous measurements Comparison with wall current monitor Summary

New design 472MHz 138 MHz H field H field E field E field

Old design 293 MHz 658 MHz 681 MHz 1018 MHz All fields for all modes are located around the wire forks Ferrite location not ideal.

Agenda New design Current design Power loss Mode geometry Previous measurements Comparison with wall current monitor Summary

Previous measurements by Federico et al in 2003 Cavity Mode Related Wire Breaking of the SPS Wire Scanners and Loss Measurements of Wire Materials" F. Caspers, B. Dehning, E. Jensen, J. Koopman, J.F. Malo, F. Roncarolo - Proc DIPAC 2003 Simulation results (from eigenmode, 2013) Frequency Of main modes Shunt impedance (kOhm) Q 290 MHz 18.5 1660 658 MHz 55 7000 682 MHz 146 5500 1020 MHz 210 12000  Modes around 700 MHz To be checked with the actual location of shielding and ferrite To be checked with simulations of transmission measurements

Agenda New design Current design Power loss Mode geometry Previous measurements Comparison with wall current monitor Summary

Example of Wall current monitor (J Example of Wall current monitor (J. Belleman - BI) impedance reduction of the pipe interruption with coaxial line  This reduction of the impedance of the gap does not seem possible with the wire scanner

Agenda New design Current design Power loss Mode geometry Previous measurements Comparison with wall current monitor Summary

Outlook New design of wire scanner seems very unfavourable from the impedance point of view due to the abrupt unmatched beam pipe interruption Interesting solution proposed by Heiko and Hannes after MSWG helps reducing impedance, but not feasible for BI. Putting ferrites reduces efficiently the heat load, but: Is it possible to put ferrites there? If needed, is it possible to cool the ferrites? Work still on-going to: Find acceptable solutions Compare with previous measurements of Federico See the impact of the position of the wire in the tank