Transcript THOA05 talk

Undulator K-Parameter Measurements at LCLS
J. Welch, SLAC National Accelerator Laboratory
Contributors: R. Bionta, A. Brachmann, F.-J. Decker, Y. Ding, P.
Emma, A. Fisher, Z. Huang, R. Iverson, H. Loos, H.-D. Nuhn, H.
Sinn, P. Stefan, D. Ratner, J. Turner, J. Wu, D. Xiang
This work is supported by the U.S. Department of Energy, contract DE-AC02-76SF00515, and was performed
under the auspices of the U.S. Department of Energy, by University of California, Lawrence Livermore National
Laboratory under Contract W-7405-Eng-48, in support of the LCLS project at SLAC.
August 27,2009
FEL 2009
THOA05
James Welch
[email protected]
Topics
Introduction
Motivation
Diagnostics
Measurements schemes
Calibrations, Checks, Errors
Results
Outlook
August 27,2009
FEL 2009
James Welch
[email protected]
Motivation for in-situ K Measurements
The 130 m long undulator consists of 33, essentially
identical, independently tunable segments.
FEL gain is lost if dK/K (RMS)  1.5x10-4
K Tolerance was well met, we lased right away, but…
Temperature, alignment, position, radiation, can change K.
We have a validation program, whereby segments are
ocassionally removed to the laboratory and tested.
In-situ K measurements will allow timely tuning
correction, and guide segment selection for removal and
validation.
August 27,2009
FEL 2009
James Welch
[email protected]
Diagnostics
K monochromator passes only one
x-ray energy and one angle. It is
not tunable to other energies.
x-rays
K-monochromator
August 27,2009
FEL 2009
SSRL
x-ray energy [eV]
W
Si 111
FWHM 1.2 [eV]
photodiode
Get spectrum by scanning
electron beam energy.
James Welch
[email protected]
Basic Measurement Schemes
One-segment scheme
Compute K difference
from spectrum shift
August 27,2009
FEL 2009
Two-segment scheme
(FEL2006)
Match K of Test to
Reference segment by
minimizing the twosegment bandwidth.
James Welch
[email protected]
One-Segment Method
First, only the REF segment is put online and
a spectrum is measured. The Reference
“inflection point” is determined.
inflection point
Next, the Ref removed and theTest segment
is put online.
Then, we measure a series of spectra for
different horizontal positions the Test
segment and find the match position.
Imager
Test
Test
August 27,2009
FEL 2009
Ref
Ref
undulator segments
(33 total)
K-mono
photodiode
James Welch
[email protected]
Central Ray Determination
Spectrum depends on K and observation
angle .
Insure “Core” radiation for Ref and Test
segments hits detector.

u
1  2 1 K 2 /2   2 2 
2
-15 MeV
Look at image just after K-monochromator
with energy just below pass band energy.
Statistical precision of location of Central
Ray is 0.03 mrad or 3 mm.
August 27,2009
FEL 2009
-10 MeV
James Welch
[email protected]
K Monochromator Transmission
To find electron energy for transmission, aim a bit
high and look at imager. Next search for the
transmission angle.
3 mrad rotation easy to see on imager. (FWHM is ~70
mrad. ) Alternately, scan angle and measure photodiode
signal.
August 27,2009
FEL 2009
James Welch
[email protected]
Single Segment Spectrum
3x3 mm slits for u33 -> +/- 19
mrad.
core size +/-6.7 mrad
Beam energy jitter, 0.04% rms,
typical.
Data is from non-synchronous
acquisition.
Simulation assumes 0.003%
energy resolution based on BPM
resolution and dispersion.
August 27,2009
FEL 2009
James Welch
[email protected]
Errors
Random Errors:
RF phase jitter -> dE/E = 4x10-4.
Wakefield energy loss and peak
bunch current jitter
Photodiode noise
Mitigation….
Dogleg bends bpms provide
3x10-5 relative energy
resolution and freedom from
betatron motion.
Bias electron energy scan to
match K steps.
August 27,2009
FEL 2009
Systematic Errors
Spontaneous radiation
Wakefield energy loss
Temperature differences
Observation angle
Mitigation
3000 A peak bunch current is
normal for FEL operation. Can
easily tune to 500 A
Both bunch current jitter and
wakefield energy loss per meter
are reduced.
James Welch
[email protected]
First Results
Test Segment
Reference Segment
K (Test-Ref/Ref)x104
X match [mm]
4
5
0.5
-0.07
5
6
2.3
-0.34
6
7
-3.8
0.57
7
8
1
-0.15
1.9, 5.6, and 4.7, x 10-4.
8
9
-1.5
0.23
Not implemented in this data
9
10
-0.7
0.11
synchronous acquisition
energy biasing
two-segment method
10
11
0
0
11
12
-1.1
0.17
12
13
-4.7
0.71
13
14
-1.3
0.20
14
15
-2.7
0.40
15
16
-0.3
-0.33
31
32
2
-0.3
32
33
1.9
-0.28
FEL lasing at 0.15 nm means
K’s are in good shape.
Measurement Repeatability
Meas. Ave -0.6
Design
-0.5
August 27,2009
FEL 2009
James Welch
[email protected]
Outlook
Early results from in-situ measurement of K-parameters are promissing,
though somewhat noisy.
Signal levels are good, simulation and measurements are in good general
agreement.
Noise reduction techniques were not fully implemented but are ready.
Measurement parameters (step size, slit settings, gains, integration
times, energy range, harmonic, etc. ) still need to be optimized.
Two-segment method needs implementation.
Systematic effects are small and well in hand.
August 27,2009
FEL 2009
James Welch
[email protected]
Theory of Two Segment Spectrum
Spectral intensity depends on relative detuning
and phase difference
Detuning parameters, 1,2
Phase difference, 
Angle parameter, 
Spectral intensity, I
Includes angle energy
correlation
1 
u
2
2 2
1
K
/2


 

2
2

August 27,2009
FEL 2009
James Welch
[email protected]
Theory - Angle Integration
Two identical segments
Steepest (negative) slope
Most signal comes from
first 7-8 mrad
20 mrad is max angle for 1st
segment (chamber limit
Maximum negative slope for
K measurement doesn’t
depend on angle of
integration much for angles
≈ 7-8 mrad or more.
August 27,2009
FEL 2009
James Welch
[email protected]
Theory - Angle Integrated, 2 Detuned
Segments
Detuning segments
produces slight
slope/linewidth
change
3% slope change
for 0.1% K change
Steepest negative
slope will be used
to track K.
August 27,2009
FEL 2009
James Welch
[email protected]
Radiation Spectrum from Two Undulators
Pinhole Spectrum
Dependence on K
Dependence on N
Dependence on ∆K between
2 segments
Dependence of phase error
between 2 segments
Angle Integrated
Spectrum
Dependence on angle of
integration
Dependence on K
Dependence on N
Dependence on ∆K/K
between 2 segments
Dependence of phase error
between 2 segments
August 27,2009
FEL 2009
James Welch
[email protected]
Measure All Segments: ‘Leap Frogging’
Measure
Adjacent
Pairs
Skip 2
Between
Pairs
...
rms(K1 - K33) ≈ rms(K1-K2) x √33
...
rms(K1 - K33) ≈ rms(K1-K4) x √11
Phase difference introduced by skipping segments can be
adjusted using a closed orbit bump (if 2 or more segments
are skipped).
August 27,2009
FEL 2009
James Welch
[email protected]
Theory - Pinhole, 2 Segments with Phase
Difference
No detuning
Slight shift and
asymmetric distortion
of curve
Max negative slope
change 0.7%.
August 27,2009
FEL 2009
James Welch
[email protected]
Real vs Ideal Undulator Fields
Two identical segments, with a simulated magnetic
field equal to the measured field in the LCLS
prototype, were modeled.
A systematic error of 0.008% was found but is not
understood.
Still within required tolerance 0.015%
August 27,2009
FEL 2009
James Welch
[email protected]
Theory - Pinhole, 2 Detuned Segments
0.1% K detune, no
phase error
-0.09% shift and
Steepest (negative)
slope
slight broadening.
4% decrease in max.
negative slope
August 27,2009
FEL 2009
James Welch
[email protected]
Method
Roll out all but two nearby
segments
Verify pointing angles using slit
scanning to maximize photon
energy.
Precisely measure electron
beam energy jitter, pulse to
pulse
Detect xrays around the first
harmonic using narrow
bandwidth crystal spectrometer
Construct the xray spectrum by
correlating the no. of detected
photons with the measured
energy jitter.
August 27,2009
FEL 2009
Change K of second segment a
known amount by shifting
horizontally.
Obtain another spectrum and
move again (≈ 9 X).
Find steepest slope of each
spectrum.
Fit steepest slopes vs K data to
find position where K’s are
matched.
Advance to next pair of
segments
Repeat until all segments are
measured.
James Welch
[email protected]
Energy Jitter Measurement
(x0, x'0, d)
R=I
x2=R11 x0 + R12 x'0 - d
BPM-1
BPM-2
x1=R11 x0 +R12 x'0 +d
Take BPM reading difference:
x1 - x2 = 2d
Get clean relative energy signal:
d = (x1 - x2) / 2
Error, sd , is BPM resolution, sx :
sd = sx ⁄ √2
August 27,2009
FEL 2009
 = 125 mm, sx ≈ 5 mm 
sd ≈ 3 x 10-5
James Welch
[email protected]
Phase Difference
Phase difference between
segments distorts shape of
spectrum.
Effect is easy to identify and
if necessary data can be
excluded from fit for
steepest slope determination.
Effect of 70 degrees of phase
difference between segments. (LCLS
spec. is max of 20 degrees)
August 27,2009
FEL 2009
James Welch
[email protected]
Theory - Pinhole
u
1  2 1 K 2 /2   2 2 
2
One segment K
dependence
Simple frequency
(photon energy)
shift of spectrum
Higher K means
lower frequency

Observation angle
 can only shift
spectrum lower
August 27,2009
FEL 2009
James Welch
[email protected]
Detector
Noise effects that add error
to the number of detected
photons or the frequency -->
August 27,2009
FEL 2009
James Welch
[email protected]
Finding ∆K=0
Scan K of one segment and
find value that maximizes
the steepest slope
Neglecting small energy
loss between segments,
the extremum value is
when the segment K values
are identical.
Simulation shows
resolution of ∆K /K of
0.004% rms
August 27,2009
FEL 2009
James Welch
[email protected]
Inflection Point Determination
Steepest slope depends on K
difference, but not on spectrum
absolute shift
Third order polynomial fit to
truncated spectrum data easily
yields steepest slope
N  N 0  a( / )  b( / ) 2  c( / ) 3
 dN 
b2

  a 
3c
( / ) max
August 27,2009
FEL 2009
James Welch
[email protected]
Two-Segment Spectrum Makes No sense!
Good general
agreement with
simulation
Way too little
slope compared
with one-segment
Again, excess
noise
Measurement
Simulation
some amplitude
noise
August 27,2009
FEL 2009
James Welch
[email protected]
Error sources
Beam energy jitter, 0.1% rms.
Detector is assumed to be narrow bandwidth ( << 1/N), high efficiency, Si
crystal, Bragg diffraction
Measure each pulse to 3x10-5 and use to reconstruct the spectrum
Natural beam energy jitter is sufficient to sample region of steepest slope.
Phase differences between segments
Shown to be neglible
Alignment/Pointing errors
More than about 8 mrad beam angle will scrape core SR on the vacuum
chamber and distort the high energy edge of the measured spectrum.
August 27,2009
FEL 2009
James Welch
[email protected]
More Disclaimer
All measurements are
preliminary - not
credible.
Only <1 shift of
reasonable looking data
was obtained
No verification using
Two-segment technique
August 27,2009
FEL 2009
James Welch
[email protected]
Random Error Mitigation
Measure energy
deviation of each pulse
in dispersive region.
Dogleg bends bpms
provide 3x10-5 relative
energy resolution and
freedom from betatron
motion.
August 27,2009
FEL 2009
Run at low bunch
3000 A peak bunch
current is normal for FEL
operation. Can easily tune
to 500 A (longer bunch).
Both bunch current jitter
and wakefield energy loss
per meter are reduced.
James Welch
[email protected]
2-Segment Scheme
Measure synchrotron radiation spectrum produced by two
undulator segments, and scan K of one segment
Other schemes compare spectra from individual segments.
(Pinhole technique, angle-integrated edge measurement,
reference undulator)
K’s are matched when spectrum has the steepest slope on
high energy side of 1st harmonic peak.
Match segments pairwise until all segments are measured.
undulator segments (33 total)
Test
August 27,2009
FEL 2009
Ref
James Welch
[email protected]
K Adjustment Mechanism
Effective K varies
linearly with
horizontal position,
K/K = -2.68x10-3
mm-1
Segment
Horizontal Slides
Canted Poles
August 27,2009
FEL 2009
James Welch
[email protected]
Calibrations and Checks
Alignment of Central Rays
K-monochromator transmission angle and
energy
One-segment spectrum
Measurement details
August 27,2009
FEL 2009
James Welch
[email protected]
Measurement Details
Real Data
Inflection point can be sensitive
to range of data used for fit
when data is noisy.
Inflection Point
Biasing the electron energy scan
range avoid biasing the fit.
One measurement takes about 5
minutes. (Slow stage travel.)
August 27,2009
FEL 2009
James Welch
[email protected]