Quasi-Isochronous (Low-Alpha) Operation and Observation of CSR at NewSUBARU

Transcription

1 KEK Quasi-Isochronous (Low-Alpha) Operation and Observation of CSR at NewSUBARU Y. Shoji LASTI, University of Hyogo Low Alpha Y. Shoji, S. Hisao*, T. Matsubara* *students Negative Alpha A. Ando, S. Hashimoto CSR Ando, Hashimoto, Shoji T. Nakamura SPring-8 (Accelerator) T. Takahashi Kyoto University H.Kimura, T.Hirono, K.Tamasaku, M.Yabashi SPring-8 (Beam Line)

2 Where is NewSUBARU? KEK Spring-8 SR 1 GeV linac Circumference m Injection Energy 1. GeV Electron Energy GeV Type of Bending cell DBA with Inv.B Number of Bending Cell 6 RF Frequency MHz Maximum Stored Current 5 ma Natural Emittance 38 nm (1GeV) Natural Energy Spread.47% (1GeV) NewSUBARU Inteferometry(1998) Microscope(1) New Material(1998) Photo-chemical(1998) Long Undulator Short Undulator LIGA(1997) LIGA(1997) Optical Klystron EUVL(1997) LIGA(4) R & D (1997)

3 Basic Idea of Low Alpha KEK *Quasi-isochronus = small momentum compaction factor (α) L/Lo 1+α 1 δ +α δ + α 3 δ δ (Ε Ε )/Ε *Small α 1 --> short bunch f S = α 1/ 1 (ev RF -U ) 1/4 (h /πε) 1/ f REV σ T =α 1/ 1 (ev RF -U ) 1/4 (E/h) 1/ (σ E /f REV ) f S is used to estimate α 1 * High V RF is another way to compress the bunch * Linear Factor α 1 α 1 =(1/ Lo (η/ρ)ds α 1 requires negative η section --- Break achromat or negative ρ section. --- Invert bend

4 Control of α 1 ; Invert Bend KEK Change η in the invert bends --> change α 1 keeping achromatic condition 34 o -8 o +34 o =6 o dispersion function (m) Q3 Q4 α1= α1=.14 Q4 Q3 -.5 BEND INV.BEND BEND s (m) Normal Bend Invert Bend Two Q-families are used

5 Second Order Factor α KEK Sextupole & chromaticities α control is the essential part of α 1 operation α = (1/L ) η 13 K ds synch. tune α = ξ x = -(1/4π) η 1 β x K ds horizontal tune ξ X =-1.7 ξ y = (1/4π) η 1 β y K ds vertical tune ξ Y =-7.7 K = (1/B ρ )( B y / x ) fs (khz) α1=7.5 E-6 α= α3=.9 α4=-18 α5=1.e+3 α6=6.4e+6 α7= -6 E+8 This Sextupole family controls α frf (Hz) f RF dependence of f S is measured to confirm that α =

6 Result(1); Bunch Shortening KEK Reduce α 1 The σ T is reduced according to α 1 scaling low (α 1 >x1-5 ) Minimum length was 1.4ps Low current I B < ua/bunch (.4mA) No current dependent lengthening At α 1 <x1-5 Reduction of α 1 did not reduce σ T. Raise of V RF reduced σ T. bunch length (σ; ps) ua/bunch; 1kV.4 ua/bunch; 3kV 1.8 ua/bunch; 1kV 1.8 ua/bunch; 3kV VRF=1kV Resolution of the monitor was not the reason of the limitation. /α1 VRF=3kV Measured by a streak camera We will come back to this problem after some sheets

7 KEK Result(); CSR(Coherent Synchrotron Radiation) Lower α high radiation power Did the power show the same saturation? NO, maybe. bunch length (σ; ps) ua/bunch; 1kV.4 ua/bunch; 3kV 1.8 ua/bunch; 1kV 1.8 ua/bunch; 3kV VRF=1kV /α1 VRF=3kV radiation power (arbit. unit) I B =3µA/bunch V RF =1kV Measured by a bolometer, 35cm -1 filtered α 1 Integration of power up to 35cm -1

8 Result(3); CSR modes KEK relative radiation power / bunch multi-bunch; normal a1 single-bunch; normal a1 multi-bunch; low a1, high Vrf stored beam current per bunch (ma) THz radiation mode measurement with bolometer (35cm -1 ) normal α 1, high current per bunch -- burst mode CSR small α 1 -- staedy state CSR? Power I normal α 1, low current -- normal radiation Power I

9 Result(4); CSR Spectrum KEK relative radiation power / bunch CSR-- strong long λ radiation multi-bunch; normal a1 single-bunch; normal a1 multi-bunch; low a1, high Vrf stored beam current per bunch (ma) INTENSITY (arb. unit) INTENSITY (arb. unit) Burst mode had shorter λ radiation, although the bunch was long WAVE NUMBER (cm ) low α1 multi-bunch.46ma normal α1 multi-bunch ma low α1 multi-bunch.46ma normal α1 single-bunch 5.6mA WAVE NUMBER (cm -1 )

10 Result(5); Bunch compression by V RF KEK CSR spectrum Power; P=I Spectrum; short bunch -> short wave length CSR continuous spectrm? the data is not clear bunch length (σ; ps) measured by streak camera ( ) INTENSITY (photons/s/.1%b.w.) FREQUENCY (THz).1 1 1GeV: R=3.m 8mrad(H) 8mrad(V) Gaussian Bunch FWHM=5.73ps(rms=.73mm) 3.3µA/bunch incoherent I b =ma SIMULATION assuming Gaussian WAVENUMBER (cm -1 ) VRF (kv) Shorter bunch emits shorter wavelength CSR 6 4 low α 1 I B =.46mA V RF =1kV (x) V RF =7kV(x4) V RF =36kV WAVE NUMBER (cm -1 )

11 Beam Dynamics Interests KEK Two Main Subjects Low Current ---> Quasi-isochronous limit (This presentation) Current Dependence --> Instability (MWI, CSR Form factor (potential well distortion) Some theoretical predictions on the limit Limitation from the finite α --- old and practical problem Longitudinal Radiation Excitation Y.Shoji et al., PR-E.6 ps Synchro-beta coupling Linear coupling Y.Shoji, PR-STAB. ps Second order E. Forest, δ =.7 σ ΕΝ.... More practical problem --- stability RF noise Magnet ripple....

12 BESSY-II; Leading machine KEK *Achieved at BESSY II shorted bunch.7 ps rms steady state coherent radiation at low alpha *What were their problems? The phase noise of the master oscillator and of the 5MHz fast sweep voltage these noise sources add a random contribution of.4ps to the bunch length. (M. Abo-Bakr, et al., PAC3) Presently there is a limit by 3Hz noise, visible on the longitudinal beam signal,.. (J. Feikes, et al,, EPAC4).. a relative current change of the Q4-family.. of 1-4 produces a CSR power change of 5% (J. Feikes, et al, Beam Dynamics News Letter 35) A tiny change of the sextupole, the last figure of the setting panel by just a few, destroys the beam condition (G. Wuestwfeld)

13 Parameters of BESSY II and NewSUBARU BESSY II NewSUBARU Natural Energy Spread.8%.47% Natural Emittance (nm rad) 3π 3π α 1-1.4x1-6 5x1-6 α Damping time 8ms 1ms Lattice non-achromatic DB DBA+IB Time jittering at Streak C.4ps.4ps KEK BESSY II J. Feikes et al., EPAC4 bunch length (σ; ps) ua/bunch; 1kV.4 ua/bunch; 3kV 1.8 ua/bunch; 1kV 1.8 ua/bunch; 3kV VRF=1kV /α1 VRF=3kV

14 (4) Practical limitations from the magnet KEK a relative current change of the Q4-family.. of 1-4 produces a CSR power change of 5% (J. Feikes, et al, BESSY, Beam Dynamics News Letter 35) Setting resolution of Q-magnet (16 bit, stability=1-4 /8h) Q4 Setting resolution = 1 bit --> α 1 =4 X 1-6 Fine adjustment is possible using other Q magnets Setting resolution of Sext-F (1 bit, stability=1-3 /8h) Setting resolution = 1 bit --> α =.6 X 1-3 Fine adjustment is possible using other S magnets Large enough RF bucket α 3 d[l/lo]/dδ > be satisfied for all δ ---> α 1 +α δ + 3α3δ> > α 3α 3 α 1 < α =.6 X 1-3, α 3 =.9 > α 1 >.5 X 1-6 (σ τ >.5ps) Field Ripple is also important

15 (5) The α 3 -- Another essential part KEK The α 3 makes stable RF bucket for large δ The α 1 and α 3 should have same signs. --> enough RF bucket When α 1 and α 3 have opposite sign --> short life time Large α 3 enlarges tolerance of α Large α 3 with the energy spread effectively enlarge α 1 d[l/lo]/d =α 1 + α 1 δ + 3α 3 δ α 3 =.9 σ ΕΝ =.47% (natural spread) <3α 3 δ > =6α 3 σ ΕΝ α 1 =1.X1-6

16 (6) Coherent Longitudinal Oscillation KEK Longitudinal oscillation [ τ]/[ δ] α 1 1/ At low alpha, the energy deviation by the coherent oscillation becomes large. Two kinds of oscillation on-resonance oscillation induced by a broad-band noise out-of resonance oscillation at harmonic components of the primary line (6Hz) Low frequency; dx signal fs (khz) Hz frf (Hz)

17 Effect of Slow Oscillation ; Simulation KEK Beam parameters E 1GeV τ ε 11.4 ms α 1 5x1-6 α 3.9 V RF 3kV f S 547Hz σ E /E.48x1-3 σ T.69 ps external perturbation f EXT 16Hz amp +1Hz (4ps) number of particles 1 bunch length ( σ; ps) time displacement (ps) 3 6 α3= α3= α3=.9 - α3= - -4 fext=16hz external perturbation phase (deg.) α3=.9 α α3= 3 =.9 α3= α3=.9 α 3 = fext=16hz energy displacement (1^-3) energy spread ( σ; 1^-3)

18 Circumference Feed-Back KEK A dipole error produces a longitudinal oscillation Dipole error at dispersive sections changes circumference L = θ ERR η ERR The response of COD to θ ERR is x = [ β/sinπν] β ERR θ ERR cos[ ψ-ψ ERR -πν] -(1/α 1 ) ( L /L ) η In small α 1 ring, the second term is dominant. dx has high sensitivity -- good diagnostic of a longitudinal oscillation τ = (α + jω ) ω E C S ω S ω + jωα E P e jωt ε = ( ω S α ) Active feed back to the dipole magnet reduces longitudinal oscillation-- good for a slow feed-back ω S C + jω P e jωt ω + jωα E

19 Circumference Feed-Back; test KEK A dipole steering is used to reduce longitudinal oscillation Dipole error at dispersive sections changes circumference dx signal BPM signal (X) (w. LPF; x) Feed-back OFF ON

20 Resonant Oscillation KEK Reduce Oscillation by Low level phase feed-back In operation at SPring-8 (T. Ohshima, PAC1) The similar effect also at NewSUBARU (Y. Kawashima, Y. Shoji) T.Ohshima, PAC 1

21 Unknown Process KEK Synchrotron Oscillation is enhanced at δ > The bunch length was not always shortest at where the f S was the smallest. It strongly depended on δ (or f RF ) at small α 1 (α 1 = ) khz khz khz 4 relative strenghth time (ps) bunch shape. FFT spectrum of the beam signal

22 Measured bunch length and CSR KEK Bunch length vs fs 4/6/13 α 1 =5x1-6 ; I B =4µA V RF =1kV CSR power vs fs 4/1/14 α 1 =1x1-5 ; I B =5µA; V RF =1kV bunch length (σ; ps) frf (Hz) fs (Hz) radiation power (arbit. unit) frf (Hz) fs (khz)

23 Beam Dynamics Interests Again APAC Coupling of Longitudinal Oscillation & Transversal Oscillation energy displacement δ > horizontal displacement dx = η δ dispersion at RF cavity > synchro-beta resonance shift of circumference < c.o.d. timing spread at H is not < betatron oscillation spread of circumference < chromatic tune spread Requires High Stability& Low Noise synchrotron frequency damping time, harmonic noise of primary line Non-linearity of RF bucket the linear part is extremely small, the second order term is almost zero 3-8 Hz or less 1Hz 6Hz - 7Hz

24 KEK StreakCamera ; Bunch Length Measurement Hamamatsu C686 is set at BL6 Time resolution.5ps Scanning freq. 83.3MHz ( f RF /6) 1 measurement takes 1/3~1sec bending magnet acceptance horizontal + 1mrad vertical + 1.9mrad radiation shield concrete tunnel wall Insert mirror (Si wafer) σ =1.45 ps time (ps) o 6 o 51 to BL6 end station M;cylindrical mirror Rs=13 vacuum window cobar glass 4.5mm optical filter M1;toroidal mirror Rs=113 (concave) Rm=3 (convex) streak camera Hamamatsu C686 Al flat mirror crystal lens

25 Observation of CSR ---Bolometer KEK A bolometer is set at the line for a beam diagnostics Si Bolometer chopping frequency 1Hz Inside filter 35cm -1 Spectrum -- interferometer Spectrometer Chopper Filters from the ring

26 (1) Longitudinal Radiation Excitation KEK A stochastic fluctuation of where the photo-emission takes place produces a fluctuation of RF phase and enlarges the equilibrium bunch length. Because of this radiation excitation the bunch length cannot be larger than σ TI = T σ ΕΝ I α at any locations of the ring. This is the intrinsic limit of a storage ring determined only by T (revolution period), σ ΕΝ (natural energy spread), and I α (a variance of partial momentum compaction factor: α*). I α = <[α*(s) -< α*(s) >],α*(s) = (1/L) sl (η/ρ) ds In NewSUBARU, at 1GeV σ TI =.6 ps --- small

27 () Linear Coupling with Betatron Motion KEK An electron in a storage ring passing through bending magnets at the outer side or inner side of a central orbit according to its betatron oscillation amplitude and phase. It makes a deviation in the path length and produces a bunch lengthening. σ B1 = εh H = γη +αηη +βη H ( m ) In NewSUBARU s (m) at the light source point for the streak camera. σ B1 =. ps --- samll H NB IB NB NB ψ H IB ψ H (degrees)

28 (3) Higher Order Coupling KEK A contribution of betatron oscillation to the circumference L π ε X ξ X (E. Forest) This shift of L/L is cancelled out by an energy shift δ ( L/L) = -(α 1 δ +α 3 δ 3 ). NewSUBARU ε X =3 1-8 m (natural emittance), ξ X =-1.7, α 1 =4 X 1-6 δ =.33%=.7 σ ΕΝ --> ξ X should be under control (accuracy is not required) A shift of the synchronous phase φ S φ S =tanφ S [k β RF ε CSI /4 -(+D)δ] --- negligibly small

29 Forced Longitudinal Oscillation KEK Forced Coherent Synchrotron Oscillation dτ dt = αε + Ce jωt dε dt = ω S α (τ + Pe jωt ) α E ε Phase Noise P ; on resonance τ = (α + jω ) ω E C S ω S ω + jωα E ε = ( ω S α ) ω S P e jωt C + jω P e jωt ω + jωα E τ MAX =(ω S /α E ) P ε MAX =[ ω S /α)/(α E )] P small for small ω S no dependence on ω S if V RF is constant Phase Noise P ; ω<<ω S τ P no dependence on ω S neither on α 1 ε ω/α) P large at low α 1 (small ω S )

30 Harmonic Oscillation KEK Presently there is a limit by 3Hz noise, visible on the longitudinal beam signal,.. (BESSY, EPAC4) Noise Source -- Phase noise of RF power? dx at BPM8 (dbm) Hz (dbm) 6Hz (dbm) 1Hz (dbm) 18Hz (dbm) 4Hz (dbm) 36Hz (dbm) fs (khz) dx at BPM8 (dbm) Hz (dbm) 6Hz (dbm) 1Hz (dbm) 18Hz (dbm) 4Hz (dbm) 36Hz (dbm) fs (khz) dx signal depends on a 1 -- longitudinal oscillation

31 Resonant Oscillation KEK relative side-band peak height (db) Noise Source -- Phase noise of RF power Relative height of the f S side-band of RF frequency. Width ; constant =8Hz Vrf~1kV Vrf~kV Vrf~3kV f S dependence; 1dB/Oct 6dB/Oct from the basic eq. 4dB/Oct f dependence of the noise V RF dependence of phase noise A noise of the phase detector has an input power dependence. Is it correct that the main noise source is the phase detector? fs (khz)

32 Correction of Phase Jittering KEK The phase noise of the master oscillator and of the 5MHz fast sweep voltage these noise sources add a random contribution of.4ps to the bunch length. (BESSY; M. Abo-Bakr, et al., PAC3) Effect of phase jitter Compare the measurements in 1s and in 1/3s. (σ 1S -σ 1/3S ), was.3 Correction of phase jitter measurements with.1 ms gate. (~ photons/measurements) Shift them by their peak positions and make a sum of profiles ( ).4 ps. Number of photons Correction of peak drift not corrected σ =1.53 ps corrected σ =1.45 ps time (ps)

33 Testing New Phase Detector KEK Present System 3 NIM modules Noise Level of output signal 1ms/Div 1dB/div Hand-maid module by Y. Kawashima (SPring-8)

34 Negative α 1 APAC Potential well distortion --> deformation of bunch α 1 >; Gauss --> parabolic α 1 <; Gauss --> rectangular (better form factor) We could not see the threshold of CSR (~5mA) α1 =+.14 α 1= mA.1mA 4mA 6mA 8mA 1.1mA 1.mA 14.1mA 16.1mA 18.1mA.3mA ma 4mA 6 4.mA ma 4mA 6mA 8mA 1mA 1.1mA 14mA 16.mA 18.1mA.1mA ma 4.1mA time (ps) time (ps)

35 Bunch length vs. current APAC Bunch Lengthening for α p =.13 K. Oide's Q=1, K=.5 Z / n =.7 Ω --> R s = k Ω 3.5 Bunch Lengthening for α p < ( R s < ) K. Oide's Q=1, K=.5 Z / n =.7 Ω --> R s = k Ω σ τ / σ τ 3 1 Meas- σ τ /σ τ Fit-Meas. Oide- σ z /σ z ZAP(.33 Ω) σ τ / σ τ Meas- σ τ σ τ Fit-meas. Oide- σ z /σ z I b (ma) I b (ma) A. Ando, 5 ann. meeting of Japanese Society for Synchrotron Radiation Research

36 power spectrum of longitudinal oscillation APAC LLC phase klystron noise feedback Power of Phase Ripple Signal (dbv) Hz Beam Phase (pick up electrode) 4Hz -11 5dB/Oct kly-pll ON f (Hz) 7Hz kly-pll OFF power (db) Energy (BPM dx) Hz 1Hz 18Hz 4Hz 36Hz 54Hz -8Hz -4Hz Hz 4Hz 8Hz frf; 76dB dfrf (Hz) 7Hz ghost

37 How to identify the noise source KEK Forced Coherent Synchrotron Oscillation τ = (α + jω ) ω E C S ω S ω + jωα E P e jωt ε = ( ω S α ) ω S C + jω P e jωt ω + jωα E The ratio, τ/ε is different for a different noise source Phase noise; P τ/ε = jα 1 /ω RF control Circumference noise; C τ/ε =(α E +jω)(α 1 /ω S ) Magnet ripple = (jα 1 /ω)(ω/ω S ) + α E α 1 /ω S By the measurement of τ and ε at the same time, we can identify the noise source.

38 Non-linear bucket with phase noise KEK time displacement (ps) bunch length ( σ; ps) α3=.9 fext=51hz external perturbation phase (deg.) α3= natural spread natural spread at de/e= fext=51hz energy displacement (1^-3) time displacement (ps) energy spread ( σ; 1^-3) bunch length ( σ; ps) α3=.9 α3= α3= α3=.9 fext=16hz external perturbation phase (deg.) α3=.9 α3= α3= α3=.9 fext=16hz energy displacement (1^-3) time displacement (ps) energy spread ( σ; 1^-3) bunch length ( σ; ps) α3=.9 fext=316hz external perturbation phase (deg.) α3=.9 natural spread at de/e= natural spread fext=316hz energy displacement (1^-3) energy spread ( σ; 1^-3)

39 SUMMARY APAC Present Status at NewSUBARU *The ring reaches to the bunch shortening limit at α 1 x1-5 (1.4ps) *Burst mode and steady state CSR is observed. Progressing R&D at NewSUBARU *Mechanism which limited bunch shortening at NewSUBARU *Negative α 1 (understand form factor, MWI) *A beam line for the observation of CSR More R&Ds *Maximum current for users, Thresholds of instabilities *Refinement of parameter (test of IB-in-gap sextupole) *Make the machine more stable *Low and High Energy Operation....