Strategies for achieving sub-10fs timing in large-scale FELs
R. B. WILCOXG. HUANG, L. R. DOOLITTLE, J. M. BYRD
LBNLJ. C. FRISCH, A. R. FRY
LCLSR. HOLZWARTH
MENLO SYSTEMS
FEL 2012NARA, JAPAN
1
Outline
• Overall system design approach for <10fs • Subsystems
– Clock requirements– Stabilized link– Laser oscillator timing control– Downstream monitors and feedback
• Conclusions
2
<10fs pump/probe experiments drive timing requirements• ≤10fs photon pulses from LCLS, SACLA, FLASH…• Want timing uncertainty ≤ pulse width, otherwise…
– Pulse is statistically widened– Timing range is statistically sampled (then “binned” if measured)– Shots are wasted, reducing effective reprate
3valid data range
pumpprobe
detect timing, “bin” data by time
wastedshots
jitterstatistics
Laser synchronization path
• Carry stability of clock down stream– Minimize jitter added by subsystems
4
opticalX-ray
clock
Transmitterlink
stabil-izer
lasercontrol laser
<10fs 10s to 100s of fs currently
5
opticalX-ray
clock
Transmitterlink
stabil-izer
lasercontrol laser
Laser synchronization path
• Carry stability of clock down stream– Minimize jitter added by subsystems– Feedback at maximum bandwidth for each stage
• X-ray jitter assumed to be low (Byrd et al, TUPD29)
<10fs 10s to 100s of fs currently
• CW carrier : continuous signal (fringe counting), no fiber nonlinearity– Analogous to RF clock, but 106 higher
frequency– Δt = Δφ/2πf = 1rad/(2π*200THz) = 0.8fs
Clock distribution methods
6
pulsedlaser
cross-correlator
inter-ferometer
CWlaser
pulsedlaser
pulsedlaser
cross-correlator
pulsedlaser
inter-ferometer
ø mod.
ø mod.
pulsed:
CW:
Linking CW and pulsed lasers
• Timing always transferred via optical phase detection
7
pulsed laser
lock carrier to envelope pulsed laser
lock carrier to envelope
CW laser
lock CW to carrier
lock carrier to CW
offset
interferometerstabilized link
transmitterclock
offset
reprate
link receiver
∫ ∫
CEP stablepulse train
single frequency
Clock and distribution via links
8
How good does the clock have to be?
• Determined by delay difference tD = tA – tB• High frequency: differential noise with period <2tD• Low frequency: phase delay change • Example: 200m fiber
– tD is 1μS– High frequency noise above 500kHz < 1fs– Long term frequency drift < 10-9
clock experiment
9
Δt = tDΔff
Optical clocks are good enough
• RF and optical frequencies, at exact integer multiples
Menlo Systems
Kubina et al, Opt. Expr. 13, 904 (2005)
~10-15 freq. stability
100MHZ 200THz
opticalRF
frequency
ampl
itude
reprate2 3 4 5... 2e6, 2e6+1...
<0.1fs jitter above 500KHZ
J. A. Cox et al, Opt. Lett. 35, 3522 (2010)
Lock CW to clock• Optical phase-locked loop
– Beat CW with nearest comb line
• 8kHz bandwidth piezo tuner– 1.8fs RMS jitter
11
1.8fs RMS
pulsed laser CW laserjitter spectrum
output
• Optical phase-locked loop– Beat CW with nearest comb line
• 8kHz bandwidth piezo tuner– 1.8fs RMS jitter
• >1MHz EO phase modulator– 320as RMS jitter
12
320as RMS
pulsed laser CW laserEO ø mod. • Fast extracavity
control becomes possible when jitter is <10fs
Lock CW to clock
Interferometric link stabilizer
13
tune
CW laser
opticalfrequencyreference (Rb) RF controller
frequencyshifter
outputmirror
Integrated jitter1.5fs unlocked50as locked
150m
• No moving parts• Similar device demonstrated
over 500km fiber (Science 336, 441 (2012))
• Tracks phase continuously– Nanosecond error range– Sub-fs precision
Link transmission jitter
• Transmission over 100m fiber adds ~400as to optical jitter14
pulsed laser
CW laser
Interferometer
jitter A0.95fs
jitter B1.1fs
100mjitter B-A
pulsed laserpulsed laser
Control of laser timing
15
Lock pulsed laser to CW
• Control reprate based on CW laser phase
• Increase control BW with EO• Pulsed laser not CEP stable
16
CW laser EO ø mod. pulsed laser
piezo
piezo tune only, 0.52rad
EO mod added, 0.34rad
In-loop: 280as RMS
CEP-stabilized fiber laser locked to CW
• 0.138 radian envelope-to-CW phase (113as at 1550nm)– With intracavity electro-optic modulator
• Envelope synced to CW phase with 113as RMS jitter
17
Baumann et alOpt. Lett. 34, 638 (2009)
Cross-correlation of locked lasers
• <8fs RMS difference, 10Hz to 1MHz
• 8fs/√2 = 5.6fs• Most noise below 100kHz• Limited control bandwidth
– Increase BW to 1MHz with EO crystal in cavity
18
crosscorrelator
pulsed CEP stab. laser 2
CW laser
pulsed CEP stab. laser 1
jitter spectrum
integrated jitter
Need to first “coarse” lock at 10s of fs
• RF clock controls remote oscillator • ~10fs is the limit
– 0.01 degree phase error– 10fs at 3GHz
• Currently used in LCLS and Fermi@Elettra
19time, hours
dela
y er
ror,
fs
8.4fs, 20 hours to 2kHz (loop BW)
Out-of-loop resuts:
Rbref
AMCWlaser
FS
RF phasedetect,correct
opticaldelay
sensing
ωRF
transmitter receiver ωRF
Controlling VCXO, 200m fiber
VCO or laser
ωRF
Optimizing RF lock for ti:sapphire laser• Determine open loop transfer function• Add filter to prevent oscillation with high gain (30kHz LPF)
20
Transfer function:
amplitude
phase
39kHzresonance
laser
DAC
stepresponse
ADC
RF locking results with tisaf• In-loop measurement compared with difference between
two externally referenced measuements
21
21fs RMS1Hz to 170kHz
FFT of noise
Jitter spectral densityof laser and reference
control bandwidth26fs RMS
30Hz to 170kHz
Integrated RMS jitter
In-loop:
Out-of-loop:
Downstream monitors and feedback
22
Noise measurement and control depends on repetition (sample) rate• High reprate enables high bandwidth feedback
– Control BW ≈ sample rate/10• Integrated jitter above sample rate is “shot to shot”
23
100kHz
100Hz
Effect of amplifiers on CEP
• CEP thru example optical parametric amp, 240as long term – Measured with a nonlinear interferometer after the amp– Variations due to air turbulence in compressor?
24
Schultze et al,Opt. Exp. 18, 27291 (2010)
3μJ6fs100kHz
88as 240as
Optical pulse timing detector
• Same heterodyne scheme at low reprate• Beat frequency is down at ~20kHz
– More noise, narrow bandwidth 25
100kHzpulse train
choppedCW
100kHztime: optical frequency:
1ns
10μs
1GHz (104 lines)<1kHzlinewidthchop
CW from link pulse trainto be timed
heterodyne beat
Cross correlating with optical radiation
• 2-color or “optical afterburner” concept– Saldin et al, PRST AB 13, 030701 (2010)– Optical synced with X-ray to 30as (10^-4 energy jitter)– Cross correlate with experiment laser, sub-fs uncertainty
• X-ray/optical cross-correlators 5-10fs currently• Use to correct long term drift
– No need for passive thermal stability in other subsystems 26
cross-corr.
expt.laser
Timing system block diagram
freq.converter
exptlaser
informationprocessing
BAMlaser
seedlaser
lock lock
transmitter
clock laser
lockoptional
data binning
linac
lock
RFplants
timing sensor
cathode, heater
27
FEL BAM cross-corr.
Conclusion: robust <10fs sync is likely
• All subsystems capable of ≤1fs short term jitter– Optical phase lock provides fine timing sensitivity– Well developed technology for optical metrology
• Slow drift corrected based on cross-correlation at experiment
• Experiments to further demonstrate capabilities are ongoing
28