1. GINGER Results for the NEW LCLS Undulator Configuration William M. Fawley Lawrence Berkeley National Laboratory Presented to LCLS Undulator Parameter Workshop 24 October 2003

2. Outline of GINGER Study ● First determine best K for peak gain for monochromatic cases at =1.5, 0.15, and 0.1 nm ● Examine taper performance for SASE runs at 0.15-nm, 11.47 GeV base case ● SASE performance at 0.1 nm with E=14.04 GeV ● Study of taper sensitivity for 1.5 nm case ● No wake effects examined --- need ELEGANT time- dependent beam parameters ● Some additional S2E SASE results for ICFA03 study – envelope reconstruction looks surprisingly good

3. Study Parameters/Bottom-Line Results  F  L  L gain Z sat P sat P max 0.114.04 GeV4.27E-48.41163.5 GW7 GW 0.1511.465.23E-46.2926.0 GW48 GW 1.53.631.6E-31.863012 GW84 GW SASE results; best taper for 0.15, 1.5 nm

4. Effects of Linear K Taper on LCLS SASE Output Power at =0.15nm ● GINGER SASE runs ● New drift space/undulator configuration ● Quadrupole strengths & Twiss parameters from H.-D. Nuhn ● Taper begins at z=75 m; simple linear decrease with z (including drift spaces) ● Max power obtained around 0.3 to 0.4% taper; excessively large tapers appear to lead to rapid debunching with z and thus reduced gain

5. Bunching and Inverse Bandwidth vs Z: 0.15-nm LCLS GINGER SASE runs for new LCLS drift space/undulator configuration

6. SASE Results at 0.1-nm Wavelength No taper “1 st ” saturation at ~110 m Output power ~6 GW No obvious anomalies --- but little margin for any beam degradation

7. 1.5-nm Taper Results 1.5 nm option is a “cake-walk” for LCLS parameters “1 st ” saturation at ~30 m; simple linear tapering begins at z=25 m Tapering increases power over 6-fold to > 80 GW 60 m of undulator gives most of output power More “intelligent” tapers probably could increase power to >100 GW

8. New GINGER Results for ICFA03 “2nd-Order” Simulation Study Extension of results for ICFA03-Zeuthen S2E study for LCLS Full SASE simulation extended over full beam head region - results low-pass filtered in time (original res- olution = 12 attoseconds) In regions where 5D distribution is “simple”, full SASE and envelope reconstruction agree surprisingly well Similar runs underway for wake case (CSR but no wake fields) Output P for SASE; peak P(z) for no slippage cases

9. Where might we go from here? ● Need ELEGANT runs with/without CSR effects to produce time-dependent 5D distributions at undulator entrance ● Examine temporal sensitivity of P(z) to taper ● Examine “ “ to wakes ● One optimization criterion is maximizing product of power times the inverse bandwidth (at least for experiments in which monochromatization will be done) ● See if results with taper are more sensitive to undulator errors, beam offset/pointing errors – Perhaps greater sensitivity to phase jitter but does deeper ponderomotive well help? ● Develop taper algorithm for undulator with drift spaces & consider effects of “spiky” SASE P(t)