S2E Study of Linac for TESLA XFEL

1
S2E Study of Linac for TESLA XFEL P. Emma SLAC

• Tracking
• Comparison to LCLS
• Re-optimization
• Tolerances
• Jitter
• CSR Effects

2
LCLS
150 MeV ?z ? 0.83 mm ?? ? 0.10
250 MeV ?z ? 0.19 mm ?? ? 1.8
4.54 GeV ?z ? 0.022 mm ?? ? 0.76
14.35 GeV ?z ? 0.022 mm ?? ? 0.01
6 MeV ?z ? 0.83 mm ?? ? 0.1
Lh L 0.6 m ?rf180?
rf gun
L0
L 9 m ?rf -38
L 330 m ?rf -43
L 550 m ?rf -10
L 6 m
undulator L 120 m
L3
L1
L2
X
...existing linac
BC-1 L 6 m R56 -36 mm
BC-2 L 22 m R56 -22 mm
DL-1 R56 ? 0
DL-2 R56 0
TESLA-XFEL
6 MeV ?z ? 2.0 mm ?? ? 0.1
120 MeV ?z ? 0.5 mm ?? ? 2.0
375 MeV ?z ? 0.1 mm ?? ? 1.4
1.64 GeV ?z ? 0.020 mm ?? ? 0.5
20.5 GeV ?z ? 0.020 mm ?? ? 0.01
Lh L ?1.4 m ?rf -191?
L ? 72 m ?rf ? -40
L ? 16 m ?rf ? -40
L ? 850 m ?rf 0
L 8 m ?rf ? -22
undulator L ? m
L3
L0
L2
3.9
L1
BC-1 L ? 4 m R56 -76 mm
BC-2 L ? 14 m R56 -36 mm
BC-3 L ?18 m R56 -11 mm
(parameters only approximate)
3
Twiss parameters along TESLA-XFEL
undulator
BC1
BC2
BC3
4
Bunch length and energy spread along TESLA-XFEL
sE/E
ss
5
BC1
BC3
BC2
m-bunching exaggerated by noise, but gain may be
large (see modulated beam study below).
6
Longitudinal phase space at end of TESLA-XFEL
m-bunching exaggerated by noise (see modulation
study below)
gex ? 1.3 ? 3.6 mm
7
Slice emittance at end of TESLA-XFEL
8
Slice energy spread at end of TESLA-XFEL
sE/E lt 0.01
9
Sliced Bunch Analysis
gex,y
Ipk
sE/E0
Dl/l
slice 4D centroid osc. amplitude
Twiss slice mismatch amplitude
10
1 mm
Quad alignment tolerances
10 mrad
Quad roll-angle tolerances
11
Longitudinal-only simulation with LiTrack (200k
in 66 seconds)
no CSR
12
Test rf phase sensitivity
Df0 0
Ipk ? 11 kA
Df0 0.2
Ipk ? 6 kA
13
Scan gun-laser timing and charge, monitoring
energy and peak current
DE/E lt 0.1
gun-timing
charge
14
gun-timing
charge
L0-phase
L0-voltage
Dtilt 0.13 ps
DQ/Qlt 4
DV0/V0lt 0.08
Df0lt 0.07
3.9-phase
3.9-voltage
L1-phase
L1-voltage
Dfhlt 0.05
DVh/Vhlt 0.3
Df1lt 0.05
DV1/V1lt 0.21
15
DV2/V2lt 1.6
Df2lt 1.1
DV3/V3lt 0.1
Df3lt 2.2
L2-phase
L2-voltage
L3-phase
L3-voltage
System is very sensitive with large 11-kA spike
at head (T. Limberg)
Note 2nd-order chirp after BC2
This suggests an increase of the 3.9-GHz voltage ?
16
LiTrack with 3.9-GHz voltage raised from 16.6 MV
to 21.0 MV
previous distribution
no spikes
17
With 21-MV 3.9-GHz rf, again testing rf phase
sensitivity
Df0 0
Ipk ? 5.5 kA
Df0 0.2
much less sensitive
Ipk ? 4.5 kA
18
gun-timing
charge
L0 phase
L0 voltage
Df0lt 0.09
DV0/V0lt 0.20
Dtilt 6.0 ps
DQ/Qlt 100
3.9-phase
3.9-voltage
L1 phase
L1 voltage
Dfhlt 0.19
Df1lt 0.24
DV1/V1lt 1.0
DVh/Vhlt 1.0
19
Df3lt 2.2
DV3/V3lt 0.1
L3 phase
L3 voltage
L2 phase
L2 voltage
Df2lt 0.49
DV2/V2lt 1.4
20
adjusted 3.9-GHz
original
3.9-GHz X-band
21
degrees of X-band or 3.9-GHz
22
LiTrack Jitter Simulation of TESLA-XFEL using
jitter budget

• (DE/E0)rms
• 0.09
• Dl/l
• 0.18

energy
energy spread
6.7 minutes _at_ 5 Hz (no CSR)
(DI/I0)rms ? 13
(Dt)rms ? 0.2 ps
peak current
arrival time
23
Now test re-optimized setup with full 6D tracking
(Elegant)
No CSR
24
Elegant tracking with CSR (and increased 3.9-GHz
voltage)
4 keV injector slice energy spread
m-bunching exaggerated by noise, but gain at l ?
3 mm may be large (see modulation study below)
gex ? 1.3 ? 2.4 mm
25
Elegant tracking with CSR and slice energy spread
6 from gun
23 keV injector slice energy spread
gex ? 1.3 ? 2.0 mm
m-bunching damped by large intrinsic energy
spread (23 keV or ? 10-4 at undulator)
26
Full 6D Elegant tracking with increased 3.9-GHz
voltage and 23 keV
slice ge
slice sE/E lt 0.01
b-tron oscillation induced by CSR energy loss
gex might be affected
27
Add modulation on density and energy profile
at 120 MeV
l 500 mm A ?0.5
Use 106 macro-particles and quiet-start bunch
population in x, x?, z, DE/E
28
CSR m-bunching in full TESLA-XFEL N 106, bins
500, transient 1D model, linear optics, matched
bs, Q 1 nC, gex 1 mm, Ipk /Ipk0 ? 100, sE0
4 keV 23 keV
120 MeV
20.5 GeV
?10-2
Ipk ? 50 A
linear optics
sE/E ? 10-4 at 20 GeV after BCs
Ipk ? 5 kA
CSR off
29
Track full XFEL in 4D (x, x?, z, DE/E) from
pre-BC1 at 120 MeV to just past BC3 at 1.64 GeV
using CSR_calc (PE) and linearly re-matching to
proper b,a and energy chirp prior to each BC.
BC2
BC3
BC1
re-match point
re-match point
30
sE0 4 keV
post-BC1 l ? 123 mm, A ? 0.5
injector l 500 mm, A 0.5
gain ? 100
post-BC2 l ? 20 mm, A ? 6.0
post-BC3 l ? 6.6 mm, A ? 50
31
sE0 23 keV
post-BC1 l ? 123 mm, A lt 1.0
injector l 500 mm, A 0.5
gain ? 6
post-BC2 l ? 20 mm, A lt 1.0
post-BC3 l ? 6 mm, A ? 3
32
l 250 mm, A 0.5
sE0 4 keV
sE0 23 keV
gain ? 150
gain 1
33
TESLA-XFEL CSR Compound Gain Curve (no LSC)
starting at 120 MeV
34
Final Comments

• Large m-bunching gain, even without longitudinal
  space charge adding energy spread is very
  helpful
• Charge jitter in XFEL much looser than LCLS
• Some rf phase tolerances tighter than LCLS
• Lack of longitudinal wakefield allows very linear
  compression, producing nearly uniform current
  profile not possible in LCLS
• Possibly better performance if BC3 were
  integrated into BC2?
• Thanks especially to Yujong, Jean-Paul, and
  Torsten