Injection Transient Observation via Bunch-by-bunch Beam ...accelconf.web.cern.ch › AccelConf ›...

27
Injection Transient Observation via Bunch-by-bunch Beam Size Measurement System Hanjiao Chen, Jie Chen, Bo Gao, Yongbin Leng SSRF BI Group FLS 2018 Future Light Source Shanghai, China, 5-9 March 2018

Transcript of Injection Transient Observation via Bunch-by-bunch Beam ...accelconf.web.cern.ch › AccelConf ›...

Page 1: Injection Transient Observation via Bunch-by-bunch Beam ...accelconf.web.cern.ch › AccelConf › fls2018 › talks › tua1wb03_talk.pdf · Injection Transient Observation via Bunch-by-bunch

Injection Transient Observation via Bunch-by-bunch Beam Size

Measurement System

Hanjiao Chen, Jie Chen, Bo Gao, Yongbin Leng

SSRF BI Group

FLS 2018Future Light Source

Shanghai, China, 5-9 March 2018

Page 2: Injection Transient Observation via Bunch-by-bunch Beam ...accelconf.web.cern.ch › AccelConf › fls2018 › talks › tua1wb03_talk.pdf · Injection Transient Observation via Bunch-by-bunch

Outline

2

• Bunch-by-bunch study background

• System Setup

• Application: Injection transient study

• Next Works

• Summary

2018/3/6

Page 3: Injection Transient Observation via Bunch-by-bunch Beam ...accelconf.web.cern.ch › AccelConf › fls2018 › talks › tua1wb03_talk.pdf · Injection Transient Observation via Bunch-by-bunch

3

Bunch-by-bunch study background

2018/3/6

Page 4: Injection Transient Observation via Bunch-by-bunch Beam ...accelconf.web.cern.ch › AccelConf › fls2018 › talks › tua1wb03_talk.pdf · Injection Transient Observation via Bunch-by-bunch

Why bunch-by-bunch study

4

For Users For BI Engineers For Physicists• FB and Top-up make users

happy usually • Unstable beam conditions

will damage data quality Beam instabilitiesThe injection transient every 5mins (top-up)

• Only static information can be monitored during daily operation

• Injection transient is a very good window to learn more about machine

• A bunch is a basic unit for physics study

• Parameters of individual bunch is better than average values for physicists

An eye on every bunches to study their behaviors

We are building a 6-dimensional bunch-by-bunchdiagnostic system

All requires

Need better knowledge of instability phenomenon for optimization

Get more information during routine injections

Need parameters of every bunches

2018/3/6

Page 5: Injection Transient Observation via Bunch-by-bunch Beam ...accelconf.web.cern.ch › AccelConf › fls2018 › talks › tua1wb03_talk.pdf · Injection Transient Observation via Bunch-by-bunch

6-dimensional bunch-by-bunch

diagnostic system

5

What we are building

0 1000 2000 3000 4000 5000 6000 7000-4000

-2000

0

2000

4000

6000

turns

horiz

ontal

posit

ion / m

um

Stored bunchinjected bunch

200 300 400 500-50

050

100150

turns

0 0.5 1 1.5 20

0.02

0.04

0.06

0.08

0.1

0.12

bunch charge / nC

meas

urem

ent d

eviat

ion / m

m

horizontal, Kx = 19.1mmvertical, Ky = 13.4mm

Beam position (x,y) Beam transverse size(horizontal , vertical)

Beam longitudinal phase (z) Beam longitudinal length

Button pickups + BBB DAQ + Δ&Σ algorithmResolution ≤ 10μm (Y. Yang, IPAC13; Y.B. Leng, IBIC13; Y.B. Leng, IPAC15; Z.C.Chen, PRST,2014)

SR light + PMT array + BBB DAQ + Gaussian fitting algorithm

(H. J. Chen, IPAC17)

0 500 1000 1500 2000 2500-320

-310

-300

-290

-280

turn index

bunc

h pha

se [p

s]

stroed bunchrefilled bunch

16 17 18 19 20 21 22 23 240

50

100

150

200

measured bunch length [ps]

coun

ts

total sample = 1300STD = 1.1ps

0 500 1000 1500 2000 250015

20

25

30

35

turn index

meas

ured

bunc

h len

gth [p

s]

refilled bunchstored bunch

injection

discussed in this talk

Button pickups + BBB DAQ + Rising edge detection Resolution ≤ 2ps (Y.M. Zhou, IPAC17; Poster TUPWC02 in IBIC17)

Button pickups + Two-frequency system(L.W.Duan,NST, 2017, H. J. Chen, IPAC17)

2018/3/6

Page 6: Injection Transient Observation via Bunch-by-bunch Beam ...accelconf.web.cern.ch › AccelConf › fls2018 › talks › tua1wb03_talk.pdf · Injection Transient Observation via Bunch-by-bunch

62018/3/6

Fast beam size measure in future light source

• For second-phase upgrade project at SSRF

• For injection process optimization

• For diffraction-limited storage ring in the future

More insertion devices transverse instabilities raise up Bunch-

by-bunch beam size measuresystem

Study and monitor the instabilities

No available method to detect injection process

A guild for injection optimization

Provide a effective tool for problem research, like impendence mismatch, large interference during injection

Page 7: Injection Transient Observation via Bunch-by-bunch Beam ...accelconf.web.cern.ch › AccelConf › fls2018 › talks › tua1wb03_talk.pdf · Injection Transient Observation via Bunch-by-bunch

72018/3/6

Fast beam size system at CESR

Horizontal size measurement system at CESR

Interferometer (Hor.) π- polarization (Ver.) PMT array Turn-by-turn signal readout @ CESR

Horizontal size result vs CCD Vertical size result vs CCD

• 3ns gate time• Single-shot

mode• Interferometer

optical front-end

Fast gated camera method

Single shot gated camera result

Visible light part

X-ray part

• Single Slit or Coded aperture front-end

• 32 channels linear InGaAsdetector

• 4ns time resolution

Page 8: Injection Transient Observation via Bunch-by-bunch Beam ...accelconf.web.cern.ch › AccelConf › fls2018 › talks › tua1wb03_talk.pdf · Injection Transient Observation via Bunch-by-bunch

82018/3/6

Fast beam size system at BINP

Vertical size measurement based on PMT array at VEPP-4M

Turn-by-turn result of injection Spectra result near tune

PMT detector

PMT array detector problems• Has a low value of an average anode current• can’t be used for multi-bunched beam

(A.V. Bogomyagkov et al, Rev. Sci. Instrum.78, 2007)

APDA detector(V. Dorokhov et al., IBIC2016)

Avalanche photodiode array detector Comparison of PMT and APDA Result of bunches in one turn

5ns time resolution

Page 9: Injection Transient Observation via Bunch-by-bunch Beam ...accelconf.web.cern.ch › AccelConf › fls2018 › talks › tua1wb03_talk.pdf · Injection Transient Observation via Bunch-by-bunch

92018/3/6

Fast beam size system at HLS

Two acquisition devices:• Signal digitizer• High-speed oscilloscope

Direct imaging system

Turn-by-turn result without exaction

Position spectra with vertical betatron oscillation

(C.C.Cheng, et al., Chinese Physics C, vol 40, 2016)

Page 10: Injection Transient Observation via Bunch-by-bunch Beam ...accelconf.web.cern.ch › AccelConf › fls2018 › talks › tua1wb03_talk.pdf · Injection Transient Observation via Bunch-by-bunch

10

Bunch-by-bunch beam size measurement system

2018/3/6

Page 11: Injection Transient Observation via Bunch-by-bunch Beam ...accelconf.web.cern.ch › AccelConf › fls2018 › talks › tua1wb03_talk.pdf · Injection Transient Observation via Bunch-by-bunch

11

System Setup

Optical direct imaging system

Photomultiplier array detector

Signal pickupSignal

OptimizationSignal

Acquisition

Data resampling

Data Optimization

Size calculation

Optical part

Electronic part

Analysis part

2018/3/6

Page 12: Injection Transient Observation via Bunch-by-bunch Beam ...accelconf.web.cern.ch › AccelConf › fls2018 › talks › tua1wb03_talk.pdf · Injection Transient Observation via Bunch-by-bunch

12

Optical imaging

SSRF Storage Ring

Diaphragm Lens 1(f=2m)

Lens 2(f=24mm)

CCD CameraHamamatsu 16-channels PMT

Tunnel

1677 cm

247.4 cm

Horizontal Interferometer

241 cm

194 cm

30.4 cm

• Average and precision size measure

• About 53 μm• For system PSF

calibration

• Choose one part of input light spot

• Defend the effect of Be mirror deformation

Detect virtual image after focus point for proper size at PMT

2018/3/6

Page 13: Injection Transient Observation via Bunch-by-bunch Beam ...accelconf.web.cern.ch › AccelConf › fls2018 › talks › tua1wb03_talk.pdf · Injection Transient Observation via Bunch-by-bunch

13

Signal conditioning & acquisition

Signal Front-end Pickup

board

SignalAmplifiers

High-speedsignal

acquisition

PMT array detector

• Pick up signal from plug pins

• Export signal with SMA connector

• Supply -800 V to PMT array

• Hamamatsu H10515B• 16 channels• 0.6 ns rise time• 1mm channel pitch• Spectral range:

300-800nm

• Hamamatsu C5594 amplifier module

• High bandwidth50kHz – 1.5GHz

• High gain: 36dB• Signal level at about

100-200mV

• Agilent DSO9064A oscilloscope

• Analog bandwidth600MHz

• High sampling rate:5GS/s

2018/3/6

Page 14: Injection Transient Observation via Bunch-by-bunch Beam ...accelconf.web.cern.ch › AccelConf › fls2018 › talks › tua1wb03_talk.pdf · Injection Transient Observation via Bunch-by-bunch

14

Data processing

Raw Data from scope Load to Matlab Preview to raw

data

Calculate resampling period T

Resampling based on T and raw data

Data resampling

Data reshape PCA optimization

Organize 4 point number from 4-channels of one

bunch one turn

Gaussian Fitting for size, position

Save data & analysis

Gaussian Fitting

Data Optimization

2018/3/6

Page 15: Injection Transient Observation via Bunch-by-bunch Beam ...accelconf.web.cern.ch › AccelConf › fls2018 › talks › tua1wb03_talk.pdf · Injection Transient Observation via Bunch-by-bunch

15

PCA optimization

Original Signal Signal after PCA optimization

Eigenvalue index of Channel B original signal Temporal wave of first four orders

1st order 2nd order

3rd order 4th order

2018/3/6

Page 16: Injection Transient Observation via Bunch-by-bunch Beam ...accelconf.web.cern.ch › AccelConf › fls2018 › talks › tua1wb03_talk.pdf · Injection Transient Observation via Bunch-by-bunch

16

Injection transient study

2018/3/6

Page 17: Injection Transient Observation via Bunch-by-bunch Beam ...accelconf.web.cern.ch › AccelConf › fls2018 › talks › tua1wb03_talk.pdf · Injection Transient Observation via Bunch-by-bunch

17

Injection capture

Filling pattern of top-up operating mode at SSRF

• Beam current: about 260mA• Bunch charge: about 0.78nC• Four long beam chains,

each one about 120 filled buckets and 60 blank buckets• Transverse feedback system OFF

Screen snapshot of captured injection event

Normal signal level

Large jump in injection • Injection occurs about every 5 minutes• Set a larger trigger level than normal

value • Set ’single’ trigger mode in the

oscilloscope

Detailed plot of injection event

2018/3/6

Page 18: Injection Transient Observation via Bunch-by-bunch Beam ...accelconf.web.cern.ch › AccelConf › fls2018 › talks › tua1wb03_talk.pdf · Injection Transient Observation via Bunch-by-bunch

18

Turn-by-turn oscillation

Initial position oscillation amplitude of every buckets

Betatron oscillation introduced by • mismatch of injection kicker field• wakefield effects of previous bunchesDifferent bunch showed different oscillation amplitude• Largest amplitude bunch #1• Smallest amplitude bunch #231

as two bunches for further study

Bunch #1 contributed by kicker mismatch• typical dumping oscillation

Bunch #231 contributed by wakefield• enlarge oscillation after transmission

Similar results of position oscillation from BPM system, published in Zhizhu Chen et al. PRST2014

2018/3/6

Page 19: Injection Transient Observation via Bunch-by-bunch Beam ...accelconf.web.cern.ch › AccelConf › fls2018 › talks › tua1wb03_talk.pdf · Injection Transient Observation via Bunch-by-bunch

19

Spectra analysis

Size Spectra

Position Spectra

Horizontal tune frequency

second harmonic frequencyTwo synchrotron sidebands on bunch#1

• Clearly peak on tune and its second harmonic frequency as theoretical combining process of stored bunch and refilled bunch

• Two synchrotron sidebands near tune appeared on bunch #1 as its main wide-band contribution of kicker-mismatch

• Similar results as injection spectra from BPM system

A format near DC on bunch#1

• Same tune and its second harmonic • For bunch #1, main peak on second

harmonic frequency of tune with two synchrotron sidebands

• For bunch #231, similar amplitude of tune and second harmonic without synchrotron sideband

• A format near DC appeared on bunch #1, no idea about it.

2018/3/6

Page 20: Injection Transient Observation via Bunch-by-bunch Beam ...accelconf.web.cern.ch › AccelConf › fls2018 › talks › tua1wb03_talk.pdf · Injection Transient Observation via Bunch-by-bunch

20

Size spectra waterfall

Horizontal tune

Second harmonic frequency of tune

From #574 to #579

2018/3/6

Page 21: Injection Transient Observation via Bunch-by-bunch Beam ...accelconf.web.cern.ch › AccelConf › fls2018 › talks › tua1wb03_talk.pdf · Injection Transient Observation via Bunch-by-bunch

21

Next works

2018/3/6

Page 22: Injection Transient Observation via Bunch-by-bunch Beam ...accelconf.web.cern.ch › AccelConf › fls2018 › talks › tua1wb03_talk.pdf · Injection Transient Observation via Bunch-by-bunch

8-16 channels upgrade and synchronized sampling

22

System error in data calculation

Gaussian fitting

Data resampling

More channels: 8-16

Synchronized sampling

New acquisition device e – ADQ14 digitizer from SP device

Parameter ValueChannels 4

Sampling rate/channel 1GS/s

ADC bit 14bitData memory 2GByte

Coupling ACInput range 1.9Vpp

Bandwidth -3dB 1200MHzExternal clock

reference input frequency

1GHz

External trigger frequency >1MHz

External trigger signallevel -0.5~3.3V

Application MATLAB C/C++

.NET, PythonLABVIEW

2018/3/6

Page 23: Injection Transient Observation via Bunch-by-bunch Beam ...accelconf.web.cern.ch › AccelConf › fls2018 › talks › tua1wb03_talk.pdf · Injection Transient Observation via Bunch-by-bunch

New calculation algorithm & FPGA development

23

1 2 3 40

10

20

30

pixel index

signa

l am

plitu

de

measurementfitted curve The challenges of Gaussian fitting online processing

• Fitting calculation takes too much time in Matlab• Fitting algorithm is hard to achieve in FPGA • Gaussian model is not precision in injection process

New calculation algorithm

A B C D

Centroid = (A+B) – (C+D)Size = (B+C) – (A+D)

A good upgrade direction & need verification next step

Data workflow of Xilinx FPGA in ADQ14 digitizer

Design FPGA algorithm in User logic 2 module

2018/3/6

Page 24: Injection Transient Observation via Bunch-by-bunch Beam ...accelconf.web.cern.ch › AccelConf › fls2018 › talks › tua1wb03_talk.pdf · Injection Transient Observation via Bunch-by-bunch

6D Bunch-by-Bunch System integration

24

16 channels data acquisition module:

14bit,1Gs/ssampling rate,

1.2GHz bandwidth,Synchronized

trigger

SR light

6-chs bunch

size one dimensi

on

BufferCompute Unit

Storage

Intelligent trigger systemBased on FPGA

Abnormal trigger (transverse and longitudinal instabilities; beam loss; etc. )

Acquisition and calculation triggerButton BPMs

4-chs position

2-chs bunch length

4-chs phase

RF front-end module

Data

6-dimentional bunch-by-bunch diagnostic system architecture

Abnormal status capturing and analysis (Beam instabilities events, beam loss, etc.)

2018/3/6

Page 25: Injection Transient Observation via Bunch-by-bunch Beam ...accelconf.web.cern.ch › AccelConf › fls2018 › talks › tua1wb03_talk.pdf · Injection Transient Observation via Bunch-by-bunch

Summary

25

• Four channels bunch-by-bunch beam size measurement system were designed, built and implemented at SSRF

• Injection transient had been successfully captured with a high sampling-rate oscilloscope.

• The betatron oscillation characteristics introduced by kicker field mismatching and wakefield effects were analyzed from position and size results

• 8-16 channels setup, synchronized sampling, new calculation algorithm, online FPGA process and 6D bunch-by-bunch will be upgraded next step for high system resolution, instability process capturing and global parameters analysis.

2018/3/6

Page 26: Injection Transient Observation via Bunch-by-bunch Beam ...accelconf.web.cern.ch › AccelConf › fls2018 › talks › tua1wb03_talk.pdf · Injection Transient Observation via Bunch-by-bunch

Acknowledge

26

• Appreciated for the support from National Natural Science Foundation of China (No. 11375255)

• Appreciated for the help from beam physics group of SSRF in design and analysis of beam experiment

• Appreciated for the help from beam operation group of SSRF in beam experiment

2018/3/6

Page 27: Injection Transient Observation via Bunch-by-bunch Beam ...accelconf.web.cern.ch › AccelConf › fls2018 › talks › tua1wb03_talk.pdf · Injection Transient Observation via Bunch-by-bunch

Thanks for your attention

Contact: [email protected]

272018/3/6