Docstoc

Capture losses caused by intensity effects in the CERN SPS

Document Sample
Capture losses caused by intensity effects in the CERN SPS Powered By Docstoc
					Capture losses caused by intensity
    effects in the CERN SPS
           T. Linnecar
          CERN, Geneva
      The work reported here was carried out by the
                    following team:

                           T. Bohl
                     E. Shaposhnikova
                       J. Tuckmantel



                    Acknowledgements:

                          G. Arduini
                       P. Baudrenghien

T. Linnecar, CERN   ICFA HB-2004 Bensheim Oct. 2004   2
                        Outline of the talk
       • LHC Beam parameters in the SPS
       • Preparation of the SPS for LHC before beam, 2001
             – Impedance reduction results
       • Attaining nominal beam, 2001 - 2003
       • Nominal beam and losses, 2003 - 2004
       • Losses at capture
             – Energy loss
       • Losses along the injection plateau
             – Noise
             – Transverse
             – Instability
       • Summary and references
T. Linnecar, CERN            ICFA HB-2004 Bensheim Oct. 2004   3
                 LHC beam parameters in the SPS
                                          Unit         Injection, 26 GeV   Extraction, 450 GeV
Bunch area (2)                           eVs                0.35                  0.7
Bunch length (4)                          ns                 4.2                  1.8
Energy spread (4)                        10-3                3.6                 1.16
Intensity per bunch (Nominal)             1011p               1.3                 1.15
Number of bunches/batch                                       72                   72
Number of batches                                           3 or 4               3 or 4
Intensity/beam                            1013p            2.8 or 3.7           2.5 or 3.3
Bunch spacing                              ns                 25                   25
Frequency (accelerating system)           MHz              200.2644             200.3944
Voltage (accelerating system)              kV                2000                 7000
Frequency (Landau system)                 MHz              801.0576             801.5776
Voltage (Landau system)                    kV               0 (200)                700
Bucket area (2)                          eVs                0.68                   3
Bucket height, E/E                       10-3               4.1                  1.1
Synchrotron frequency                      Hz                 257                  238
Peak current (@ 200 MHz) in batch          A                 1.08                 1.42
pk-pk phase modulation along batch         ps             not defined             250

T. Linnecar, CERN                    ICFA HB-2004 Bensheim Oct. 2004                         4
    Preparation of the SPS before beam, 2001


• Impedance reduction              • RF equipment
   – Shielding vacuum                    –   RF feedback (upgraded)
     ports                               –   RF feedforward (new)
   – Shielding kickers                   –   Beam control (low–noise)
   – Removal lepton                      –   1 MW couplers (new)
     cavities
   – Removal non-essential
     equipment



T. Linnecar, CERN   ICFA HB-2004 Bensheim Oct. 2004               5
                         Impedance reduction results




Bunch length increase as a     Quadrupole oscillation             Coherent vertical tune shift
function of bunch intensity    frequency as a function of         with intensity before and
600ms after injection before   intensity before and after         after the impedance reduction.
and after impedance            impedance reduction.               Measurements: H. Burkhardt,
reduction                      (Note effect of new kickers)       see talk, this workshop.



     T. Linnecar, CERN          ICFA HB-2004 Bensheim Oct. 2004                          6
          Attaining nominal beam, 2001-2003
• Studies with high intensity beams showed that extra steps
  were required to reach nominal beams:
    – Beam unstable at intensity 4 times below nominal
    – Addition of low-mode number longitudinal feedback, using the
      main RF cavities, to control slow growing instabilities at 26 GeV
    – Landau damping at high energy with the 800 MHz system
    – Controlled emittance blow-up on the ramp to control high energy
      coupled bunch instabilities




    Bunch length evolution, 800 MHz on,              Improvement in
    with controlled blow-up on and off               threshold with 800 MHz
    – Electron cloud instabilities – solved by scrubbing run
T. Linnecar, CERN         ICFA HB-2004 Bensheim Oct. 2004                     7
      Nominal beam and losses, 2003-2004 (1)
• Nominal beam achieved at 450 GeV/c
    – Intensity / bunch 1.1 x 1011
    – L = (1.51 – 1.58) ns
    – L = (0.52-0.56) eVs
• Losses to 30 GeV (capture plus injection plateau)




                                           • Losses to 30 GeV




     As a function of batch intensity,            As a function of capture voltage
     VRF = 2MV. Note increase in 2003             1 batch, Nave= 7.8 x 1012
 T. Linnecar, CERN          ICFA HB-2004 Bensheim Oct. 2004                          8
                    Nominal beam and losses (2)

•   Losses approach 40% for matched voltage.
•   Are still at > 10% for best conditions.
•   Causes radiation.
•   Puts a strain on the pre-injectors. To achieve nominal in SPS
    the pre-injectors have to work near to their foreseen ultimate
    intensity. Ultimate intensity in the SPS, and LHC is
    compromised - 1.7 x 1011 / bunch, a 50% increase.

 Study of losses – give actual (incomplete) understanding


T. Linnecar, CERN         ICFA HB-2004 Bensheim Oct. 2004       9
                    Losses at capture
Can be caused by:

• Long injected bunches – 4.2 ± 0.4 ns (0.35 eVs)
  minimum bunch length due to:
   – present RF systems in PS are 40 MHz + 80 MHz.
   – there is emittance blow-up on the rise for stability
• Injection errors, energy, 10-4 dE/E, phase ± 0.2 ns.




T. Linnecar, CERN     ICFA HB-2004 Bensheim Oct. 2004       10
                    Losses at capture (2)
• Position of receiving buckets (RF waveform with feedback
  etc.) ± 0.13 ns due to beam loading in the 200 MHz
  impedance


                                              Bunch position
                                              along batch




Worst case bunch length 4.2 ± 0.73 ns. Stationary bucket 5 ns!
• But there is also bucket distortion and size reduction due to
  the SPS impedance
      particles lose energy
T. Linnecar, CERN      ICFA HB-2004 Bensheim Oct. 2004         11
    Losses at capture – energy loss effects (1)
• Movement of lost particles – always to low energy side
  – Machines are well matched.




                                                                  1ns
Height is a
function of
voltage

Continuous
flux of
particles       200 MHz component. 2 seconds       Density plot. Lost particles move
                between pictures.                  between accelerating buckets.

T. Linnecar, CERN           ICFA HB-2004 Bensheim Oct. 2004                     12
   Losses at capture – energy loss effects (2)
• Even though dB/B = 0, there is an accelerating bucket due
  to energy loss, U, in resistive impedance. Particles outside
  the bucket move through the hole between the buckets and
  are decelerated.
• The non-zero s = U/eVRF also decreases the available
  bucket area (length) – Increase of VRF helps!
• Azimuthal gap size  = 2(sin s)1/2. For U = 60 keV and
  VRF = 2 MV, s = 1.8° @ 200 MHz and  = 0.5 ns

                                   Bucket shape and particle
                                   trajectories in phase space for
                                   s = 1.8°



T. Linnecar, CERN    ICFA HB-2004 Bensheim Oct. 2004                 13
    Losses at capture – energy loss effects (3)

                                        Contributions to energy loss
                                        in the SPS are mainly from the
                                        kickers and accelerating cavities
                                        without active damping (FB and FF).



                                        The solid line is the calculated
                                        energy loss from the two sources as
                                        a function of intensity.
                                        The dots give the energy loss
                                        measured directly from the stable
                                        phase of a single bunch with 1010p
                                        – good agreement


T. Linnecar, CERN   ICFA HB-2004 Bensheim Oct. 2004                     14
                    Losses at capture (summary)
   • Bunch length plus errors plus accelerating bucket
      capture losses
   • Lost particles move to low energy and are either lost
     from the machine at acceleration or are recaptured to
     form satellite bunches (bad for LHC physics!).
   • Solutions
      – Smaller injected bunches (instability problems in the
        PS and SPS)
      – Further reduction of SPS impedance (kickers etc.).
   • But ...

T. Linnecar, CERN         ICFA HB-2004 Bensheim Oct. 2004   15
                 Loss along the injection plateau
What we have seen:
• Continuous flux of particles out of the batch
• For 75 ns spacing (used to reduce electron cloud effects) the losses
  are twice less and  to total time batch spends on the plateau
• Injecting in 2 MV then raising to 3 MV is better than injecting
  directly in 3 MV – shortest bunches
            Diffusion like process on the plateau
   – IBS (several hours, not significant)
   – RF noise (intensity dependent?)
   – Transverse losses (high  required to cure e-cloud instability)
   – Instabilities causing blow-up

   T. Linnecar, CERN     ICFA HB-2004 Bensheim Oct. 2004        16
 Loss along the injection plateau – RF noise (1)
• The bunch lifetime, given by the peak detected signal along
  injection plateau of 10.86s, is - 0.2% / s at VRF = 3 MV.
• Noise due to magnetic field checked unimportant.
• By adding noise at various points in the RF system until a
  significant reduction in lifetime is obtained in coast mode we
  can conclude:
   – The RF feedback and feedforward are not dominant sources
      of noise. Also power amplifier noise appears OK
   – The beam control « white » noise measured at the phase
      discriminator of –88 dBV/(Hz)1/2 implies an RF noise
      lifetime of 0.03% / s.
   – Hence other loss mechanisms than white noise giving bad
      lifetime of 0.17% / s
  T. Linnecar, CERN    ICFA HB-2004 Bensheim Oct. 2004       17
 Losses along the injection plateau – noise (2)
• Bunches at the end of the batch decay faster than those at
  the front.

Peak detected signal for head
of batch (top) and tail of batch
(bottom) during the cycle.                                     blow-up
Head of batch decay ~ 2.25%.                                   on ramp



                                         offset zero



                             Batch after ~30 mins. coast




T. Linnecar, CERN            ICFA HB-2004 Bensheim Oct. 2004   18
 Losses along the injection plateau – noise (3)
• There is some evidence that the bunch length always tends
  towards the same value (3.4 ns).
• This can be due to a combination of:
   – a sharply rising diffusion constant at the bucket edges
   – white noise modified by the phase loop (reduction
     around fs)


                                 Bunch length evolution along
                                 the injection plateau.
                                 Top 2 MV, Bottom 3 MV
                                 Intensity 1.25 x 1011 / bunch




T. Linnecar, CERN    ICFA HB-2004 Bensheim Oct. 2004             19
 Losses along the injection plateau – noise (4)

Peak detected longitudinal Schottky.
• The peak detected signal from the wideband longitudinal
  monitor is analysed during the store using an averaging
  FFT spectrum analyser:
• The tail of the batch behaves differently from the head and
  the detailed behaviour varies with RF voltage (fs different)
• Coherent excitation lines are important, can create holes or
  depopulate areas of bunch (when in conjunction with white
  noise ?). One line from the feedback electronics (LO) has
  been identified and removed – transmission improved by ~
  0.5%

T. Linnecar, CERN    ICFA HB-2004 Bensheim Oct. 2004        20
 Losses along the injection plateau – noise (5)
 Peak detected Schottky                  395 Hz hole caused by coherent line at 790 Hz.
 after 17 mins. store.                           The other?




                                                                        Head of batch




                                                                        Tail of batch




     VRF = 2MV, fs = 257 Hz                     Depopulated area – cause?

T. Linnecar, CERN             ICFA HB-2004 Bensheim Oct. 2004                    21
 Losses along the injection plateau – noise (6)
 Peak detected Schottky       Holes in spectrum are less            527 Hz = 790/3 x 2!
 after 15 mins. store         for this voltage.




                                                                         Head of batch




                                                                         Tail of batch




  VRF = 3 MV, fs = 315 Hz                           Shorter bunch, concave distribution

T. Linnecar, CERN           ICFA HB-2004 Bensheim Oct. 2004                     22
 Losses along the injection plateau – RF noise
                  (summary)
• White noise alone is not the dominant cause of decay.
• Coherent lines can drill holes in the bunch distribution. In
  conjunction with white noise they can perhaps do more
  damage.
• The sharply rising diffusion coefficient at the bucket edge
  increases losses from the bucket.
• There are other mechanisms producing particle loss along
  the flat bottom
• At the moment 10.86s is a long time to hold the beam!
  In the LHC it is 20 mins.! (with no radiation damping,
  which helps a lot at top energy 7 TeV)
T. Linnecar, CERN    ICFA HB-2004 Bensheim Oct. 2004         23
  Losses along the injection plateau – transverse (1)
   Effect of change in working point – 1 batch; 2% decrease in losses
                       Old working point                        New working point

batch
head


batch
tail

offset
zero
                   qh=0.19, qv=0.13 (Integer 26)             qh=0.13, qv=0.19 (Integer 26)

                   Decay (loss) at 26 GeV:
                   head of batch: 2.25%                       head of batch: 1.5%
                   tail of batch: 3.65%                       tail of batch: 2%
         T. Linnecar, CERN            ICFA HB-2004 Bensheim Oct. 2004                        24
Losses along the injection plateau – transverse (2)

• Transverse information (G. Arduini):
   – h= 0.2, v= 0.42 to control electron cloud instabilities

   –  spread, qh= 0.026,  qv= 0.052 for dp/p (2) = 5 x 10-3 (3 MV)
   – direct space charge shift qh= - 0.041 and  qv= - 0.054 (nominal)
   – impedance shift single bunch qh= 0.012 and  qv= 0.023 (nominal)

   – electron cloud causes shift along batch qh= 0.025 and  qv= 0.02
   – resistive wall, q= 0.003 both planes

• Transverse normalised emittance 3.0 m @ 26 GeV, 3.5 m @ 450 GeV


 T. Linnecar, CERN         ICFA HB-2004 Bensheim Oct. 2004               25
Losses along the injection plateau – transverse (3)
  • Tune diagram (G. Arduini)




                         Old WP                                                                 New WP
     Thick lines are systematic resonances, thin lines non-systematic resonances.
     Color code:Yellow = first order, Green = second order, Blue = third order, Purple = fourth order, Red = fifth order
     Qmin = Qmeas + q(space charge) + q(imp.) - q(chrom.); Qmax = Qmeas + q(imp.) + q(chrom.)


  • Understanding incomplete – knowledge of resonance
    strengths to be measured and optimum placing of pilot,
    single nominal bunch and full beam to be found.
 T. Linnecar, CERN                        ICFA HB-2004 Bensheim Oct. 2004                                                  26
Losses along the injection plateau – instability


• Injecting lower emittance bunches to improve the capture
  loss produced beams that were more unstable on the
  injection plateau. This could be controlled by using the 800
  MHz system in bunch shortening mode (used during
  acceleration).
• Analysis of the resulting bunch trains showed that bunches
  towards the end of the batch were still slightly unstable.
  The bunch pattern oscillated with a frequency of ~ 1.5
  MHz ~ 1/filling time of TW cavities.



T. Linnecar, CERN    ICFA HB-2004 Bensheim Oct. 2004       27
  Losses along the injection plateau – instability (2)
standard deviation
of bunch position.
high value = more unstable


  bunch lifetime.
  high value = long
  lifetime



Strong correlation
in this case between
lifetime and lack of
stability
                             Standard deviation of bunch position and bunch lifetime vs
                             bunch position in batch.
     T. Linnecar, CERN              ICFA HB-2004 Bensheim Oct. 2004               28
                             Summary
• Nominal beam for LHC has been achieved in the SPS injector
• Transmission loss in the SPS is important for future intensity increase
• Capture losses are due to a combination of injected bunch length,
  injection errors and bucket size reduction due to energy loss
• 26 GeV plateau losses are not fully understood but various
  improvements have reduced the losses to the 7% level
• Areas for study, improvement and better understanding are:
   – RF noise: Coherent excitation lines and white noise
   – Effect of transverse working point
   – Instabilities
   – Other ...


T. Linnecar, CERN        ICFA HB-2004 Bensheim Oct. 2004                29
                                     References
•     E. Shaposhnikova, T. Bohl, T. Linnecar, J. Tuckmantel, Capture loss of the LHC beam in
      the CERN SPS, EPAC 2004
•     P. Baudrenghien, T. Bohl, T. Linnecar, E. Shaposhnikova, J. Tuckmantel, Nominal
      longitudinal parameters for the LHC beam in the CERN SPS., CERN LHC Project Report
      652
•     T. Bohl, T. Linnecar, E. Shaposhnikova, Impedance reduction in the CERN SPS as seen
      from longitudinal beam measurements, CERN-SL-2002-023
•     H. Burkhardt, G. Rumolo, F. Zimmermann, Coherent Beam Oscillations and Transverse
      Impedance in the SPS, CERN-SL-2002-030
•     E. Shaposhnikova, LHC beams in the SPS: Longitudinal Plane, LHC Performance
      Workshop, Chamonix X11, 2003, CERN-AB-2003-008 ADM
•     E. Shaposhnikova, Impedance effects in the SPS, Chamonix X111, 2004, CERN-AB-
      2004-014 ADM
•     T. Bohl, A. Hofmann, T. Linnecar, E. Shaposhnikova, J. Tuckmantel, Energy Loss of a
      single bunch in the SPS, AB-Note-2004-017-RF, also EPAC 2004
•     J. Tuckmantel, T. Bohl, T. Linnecar, E. Shaposhnikova, Experimental studies of
      controlled longitudinal blow-up in the SPS as LHC injector and LHC test bed, EPAC 2004


    T. Linnecar, CERN             ICFA HB-2004 Bensheim Oct. 2004                        30

				
DOCUMENT INFO
Shared By:
Categories:
Stats:
views:4
posted:3/27/2011
language:English
pages:30