Analysis of ATF EXT/FF Orbit
Jitter and extrapolation to IP
(Data of 2014.11.18)
ATF2 Project Meeting 2015. 2
K. Kubo
Data taking
• Data taken in November 18 (2014) swing shift.
• After IP beam size tuning, observing modulation with
IPBSM 174 degree mode. (10x1 optics)
– Positions of the BPM reference cavity and OTR2 chanber were
optimized.
• Data of EXT/FF BPM are saved with several conditions.
(See Log Note)
• This report used data of
– One file (2000 pulses) with bunch intensity
intentionally changed by hand.
– 3 files (1000 pulses), each is for fixed bunch intensity
setting. (N~8.5E9, 4.9E9, 1.8E9)
BPM reading vs, pulse number
(MQD10BFF: Large beta_y)
600
500
y
400
400
300
200
200
0
100
-200
0
-400
-600
-100
0
500
1000
1500
2000
pulse number
In horizontal, drift is larger compare with pulse to pulse jitter.
In vertical, pulse to pulse jitter is more significant.
(It may be only in this particular case. (?))
y (m)
x (m)
x
Analysis using SVD
 BPM1x - pulse1 BPM1y - pulse1 BPM2x - pulse1  


 BPM1x - pulse2 BPM1y - pulse2 BPM2x - pulse2  
M 
BPM1x - pulse3 BPM1y - pulse3 BPM2x - pulse3  







monitor
(Average is subtracted for each monitor)
(Data far from average (>5-sigma) removed. BPM noise, etc.)
M  UWV
W: Diagonal matrix of singular values
 w1

0
W 
0

 
pulse
0
0
w2
0
0
w3








U_ij represents amplitude of
mode j in pulse i
V_kl represents response of
monitor l to mode k
Result of SVD: First 4 modes vs. BPM x,y
Circles: amplitude at each BPM, Lines: Fitting by injection orbit + dispersion (model)
150
150
x data
y data
x fit
y fit
RMS Amplitude (m)
100
50
x data
y data
x fit
y fit
mode 2
100
RMS Amplitude (m)
mode 1
0
-50
-100
50
0
-50
-100
-150
-150
0
10
20
30
40
0
10
BPM Number
30
40
BPM Number
40
40
x data
y data
x fit
y fit
20
mode 4
30
RMS Amplitude (m)
mode 3
30
RMS Amplitude (m)
20
10
0
-10
-20
x data
y data
xfit
yfit
20
10
Dispersion
0
-10
-20
-30
-30
-40
0
10
20
BPM Number
30
40
0
10
20
BPM Number
30
40
Result of SVD: other 2 modes vs. BPM x,y
25
10
15
RMS Amplitude (m)
RMS Amplitude (m)
x data
y data
x fit
y fit
mode 5
20
10
5
0
-5
x data
y data
x fit
y fit
5
0
-5
-10
-10
-15
0
10
20
30
40
BPM Number
Intensity dependence
Only seen in data with large intensity change
0
10
20
30
BPM Number
y position at IP phase
40
Standard deviation at each BPM in EXT/FF
(not from SVD)
(High intensity N~8.5E9)
200
x stdv
0.3*sigmax
y jitter (m)
150
Horizontal: ~20%
of beam size
(emittance 2 nm assumed)
100
50
0
0
10
20
30
40
BPM number
140
y stdv
0.3*sigmay
y jitter (m)
120
Vertical: ~40%
of beam size
(emittance 12 pm assumed)
100
80
60
40
20
0
0
10
20
BPM number
30
40
Effect to IP position of some first modes -1
• 3 Different intensities (about 1000 pulses for each)
– All BPMs used for fitting
Mode
1
2
3
8
4
Total of
mode 1 - 10
Vertical Jitter propagation to IP (nm)
N~8.5E9
N~4.9E9
N~1.8E9
3.9
12.5
17.3
16.6
9.4
7.1
1.1
4.4
2.0
6.1
7.3
6.6
0.5
0.8
0.1
18.1
17.9
No intensity dependence.
20.3
Effect to IP position of some first 10 modes -2
Compared
• Using All BPMs
• Using only downstream BPMs
Extrapolated to IP
Used BPM
Vertical Jitter propagation to IP (nm)
N changed
N~8.5E9
N~4.9E9
N~1.8E9
All
19.9
18.1
17.9
20.3
From QD10AFF
1272
1258
1098
1139
From QF7FF
1000
974
863
882
Too large jitter using downstream BPMs only
 check waist shift (May be from optics error???)
Fitted orbit around IP (s=0)
Y (microm)
All BPMs used
s (m)
There is waist shift.
(Note different scales)
Y (microm)
BPM from MQD10AFF used
s (m)
Effect to position at waist of first 10 modes
• Using All BPMs
• Using only downstream BPMs
Extrapolated to waist (not exactly IP)
Used BPM
Vertical Jitter propagation to IP (nm)
N changed
N~8.5E9
N~4.9E9
N~1.8E9
All
8.5
6.4
9.1
8.0
From QD10AFF
233
283
184
183
From QF7FF
251
213
208
210
Still large using BPMs only in downstream region
Simulation of SVD analysis
• Create sets of EXT-FF BPM data
– Orbit jitter is given
• At the beginning of the extraction line (injection error)
• Or position jitter of QD10BFF magnet
• Or position jitter of All Q-magnets
– Beam position at every BPM is calculated by SAD
– BPM resolution assumed
• 1 micron, or 100 nm for cavity BPMs
• 10 micron for stlipline BPMs
• Apply the same analysis program for the real data.
– Look at evaluated (extrapolated) y position jitter at IP by SVD
– Compare with “real” jitter at IP
Simulation result of SVD analysis -1
injection jitter
Extrapolated y jitter at IP by SVD (nm)
Evaluated jitter (all BPMs used) vs. Real jitter at IP
Real jitter was changed by changing amplitude of injection jitter.
Evaluated jitter include 10 SVD modes
40
Resolution
CavBPM: 1 um
Stlipline: 10 um
35
30
25
20
15
10
Good resolution for injection jitter,
using all BPMs
5
0
0
5
10
15
20
25
30
real y jitter at ip (nm)
35
40
Simulation result of SVD analysis -2
jitter induced at QD10BFF
160
140
Fitted y jitter at IP (nm)
Evaluated jitter vs. Real jitter at IP
“Real” jitter amplitude was changed by
changing position jitter amplitude of
QD10BFF.
Evaluated jitter include 10 SVD modes
6 cases of used BPMs (and resolution)
Use All BPMs
from QD10AFF (1 um and 100 nm)
from QF7FF (1 um and 100 nm)
Good estimation using BPMs
downstream of jitter source only
All BPM used
BPM from QD10AFF
(resolution 1m)
BPM from QD10AFF
(resolution 100nm)
BPM from QF7FF
(resolution 1m)
BPM from QF7FF
(resolution100 nm)
120
100
80
60
simulation, orbit by
QD10BFF offset
40
100 nm BPM resolution: good estimation
1 um BPM resolution: over estimate jitter
at IP by several x 10 nm.
20
0
0
20
40
60
80
100
Real y jitter at IP (nm)
120
140
Simulation result of SVD analysis -3
jitter induced at all Quads
140
120
Fitted y jitter at IP (nm)
Evaluated jitter vs. Real jitter at IP
“Real” jitter amplitude was changed by
changing position jitter amplitude of all
Quads in EXT-FF.
Evaluated jitter include 10 SVD modes
All BPM,
CavBPM rtesolution 100 nm
All BPM,
CavBPM resolution 1 m
BPM from MQD10A,
resolution 100 nm
BPM from MQD10A,
resolution 1 m
100
80
Use All BPMs and use from MQD10AFF
CavBPM resolution 100 nm and 1 um
Orbit by
all Quads' vibration
60
Always underestimate jitter at IP
40
Bad BPM resolution makes
estimated jitter large
20
0
0
20
40
60
80
100
Real y jitter at IP (nm)
120
140
Fitted jitter depends on BPM resolution
jitter induced at all Quads
Fitted y jitt at IP (nm)
1000
Orbit by vobration of all Q 20 nm
Real jiier at IP = 35 nm
use BPMs from MQD10AFF
100
10
0.1
1
10
Cav BPM resolution (m)
BPM resolution about 5 um may explain the large fitted jitter using
BPMs in downstream part of FF line only. (about 200 nm at waist)
Extrapolated position jitter at IP
• About 20 nm (RMS) vertical jitter at IP using all BPMs (57% of beam
size)
•
•
Using BPMs in downstream part of FF line only, large extrapolated position
jitter (about 1 um at IP, 200 nm at waist)
– Orbit jitter induced in FF line so large? (But measured beam size was
~60nm a few hours before these data taken.)
– Or, misunderstood BPM data, or assumed wrong optics in FF, , , , , ,?
Simulations were performed to check the analysis.
– If BPMs downstream of jitter source are used, the analysis can
reproduce jitter at IP.
– Including BPMs upstream of jitter source, the analysis tends to
underestimate jitter at IP. (This is reasonable.)
– Bad BPM resolution overestimate jitter at IP (This is reasonable too.)
• 200 nm jitter from the analysis of the experimental data may be
explained if BPM resolution is about 5 um. But it seems too large.
No clear conclusion on jitter induced in the beam line
NOTE: BPMs in downstream FF line is not so sensitive to
Position-at-IP phase orbit
4
5 10
1000
orbit y-at-IP phase
y = 37 nm, y' = 0
IP
orbit y-at-IP phase
y = 37 nm, y' = 0
IP
IP
IP
y (nm)
y (nm)
500
0
0
-500
4
-5 10
-1000
0
20
40
s (m)
60
80
50
55
60
65
70
75
80
85
s (m)
Using IPBPM will be probably necessary for reliable estimation of
orbit difference induced in the FF line.
90
SUMMARY
•
•
•
•
•
Data of EXT/FF BPM taken in November 18 (2014) swing shift were
analyzed using Singular Value Decomposition method.
Vertical jitter at FF line BPMs (angle-at-IP phase) is a bout 40% of beam
size.
Position jitter at IP (or waist) estimated extrapolating orbits calculated from
SVD modes.
– About 20 nm jitter from injection (or from upstream part of EXT) jitter.
– No intensity dependence.
– Estimation of jitter induced in FF beam line was tried without any clear
results.
With more data, better optics information and careful BPM calibration, some
improved results may be possible. But,
Using IPBPM is simple for reliable estimation of position jitter at IP.