Laboratory Report Rostock University

Transcription

1 Laboratory Report Rostock University TESLA Collaboration Meeting DESY, September 15-17, 2003 Karsten Rothemund, Jelena Maksimovic, Viktor Maksimovic, Gisela Pöplau, Dirk Hecht, Ulla van Rienen Partly supported by Pulsar Physics Ursula van Rienen, Universität Rostock, FB Elektrotechnik und Informationstechnik, AG Computational Electrodynamics

2 Overview Q values for modes in the third dipole passband of the 2x7 Superstructure K. Rothemund, C. Schmidt*, D. Hecht; joint work with H.-W. Glock** Status of the 3D Space Charge Routine in GPT with Multigrid Poisson Solver Gisela Pöplau; joint work with Pulsar Physics, NL CSR Calculations in Bunch Compressor Jelena and Viktor Maksimovic; joint work with M. Dohlus, E. Schneidmiller, DESY *graduate student, **Univ. Rostock until April 2001, free coworker since May 2001

3 2x7-Superstructure 7 Cell TESLA Cavity Input-Coupler HOM-Coupler Radius Adapter e - S-parameter: 5 modes taken into account f f f TE11 co TM01 co TE21 co = GHz = GHz = GHz TE 11 (deg.) TM 01 TE 21 (deg.) Images: I.Ibendorf

4 Results transmission: Frequency range: GHz (in 3 rd dipole passband) Resonances with significant Q-values S.. /db f/ghz f/ghz Fit S-parameter to: S( f) = Q k = N k = 1 i Im 2 Re a k 2π f pk { p k } { p } k

5 Results Q Fit of poles with low Q-value is difficult fitted curve orig. data S.. /db All Q-values are 5 well below 10 f/ghz

6 Summary of Q Value Determination Q values of 2x7 TESLA-Superstructure have been calculated (an open structure) with CSC 5 modes have been considered in the structure S-parameters of subsections were computed with CST-MicrowaveStudio TM (coupler sections, 3D) MAFIA (TESLA cavity, 2D-rz-geometry) analytically (shifting planes, rotation) Q values of resonances were determined all Q values are well below 10 5

7 Status of Poisson Solver in GPT GPT (General Particle Tracer) is a widely used tracking code Multigrid Poisson Solver for fast 3D space charge calculations implemented in GPT Improvements of the 3D space-charge routine A lot of technical issues speed up the performance Study of test cases: cylindrically shaped bunches ( pancake cigar bunches) optimal simulation parameters Improved memory management: block memory management Simulations for TTF2: Frank Stulle, DESY (February 2003) Speed up & stabilization

8 Simulations for TTF2 Test bunch from TTF2 simulations (Frank Stulle) Bunch after compression 125 MeV Particle distribution (10,000 particles) x Improve GPT elements (Pulsar Physics) Correct behavior for test-bunch by changing bounding box Drift problem solved Energy dependent quadrupole z E x x

9 Why Multigrid? Other Poisson solvers are much easier to implement They slow down considerably on non-equidistant meshes bunch with Gaussian distribution in a sphere equidistant mesh Discretization of the TTF2 bunch non-equidistant mesh

10 Analytical Test-Case Number of of meshlines meshlines Improvements: RMS Field (red=10%) RMS field error (red=10%s) log(aspect ratio) GPT Log(aspectratio) Robust and fast performance over broad parameter range Stable open boundaries Removed: noise, strange spots Improved memory management Number of meshlines Cylindrically shaped bunches: cigar-shape -> pancake-shape < A =R/gL< 1000 Worst-case test scenario: Number of meshlines Hard-edge Extreme aspect ratios CPU-time (red=3s) log(aspect GPT Log(aspectratio) Settings: fn=0.5, number of particles=10,000

11 Speed Up and Stabilization Tracking of a pancake shaped bunch R=1mm, L=0.1mm, g=5, total charge 1nC Tracking time 100ps Simulation time (50,000 particles) Old: 289 s New: 170 s

12 Cathode Non-Uniformity 3D space charge simulation: Length: 100 fs 10 MeV acceleration in a 100 MV/m uniform field Solenoid: z=0.15 m RMS radius [mm] Modulation= z [m] Final phase-space at z=0.6 m x-velocity/c x [mm] Simulation: Pulsar Physics

13 1D and 2D CSR Model Calculate retarded position of the bunch tail by bisection method. 1D bunch: no change of longitudinal profile - rigid bunch linear charge density moving with const. velocity along the arbitrary path. 2D Gaussian bunch 0.2 y[m] Test example: Bunch compressor s1ret s path x[m] Trajectory ret Bunch tail s 1 and retarded bunch tail s 1 in the arc 0.2 y[m] y[m] s1ret s1 x[m] path x[m] relativisticfactor g= 1000 relativisticfactor g= 10

14 1D and 2D CSR Model Retarded charge density distribution in translatory motion Retarded charge density distribution in circular motion Basic software structure