BB-Diagnostics: Computer Class (Matlab/AT)

1
BB-Diagnostics Computer Class (Matlab/AT)
Using Accelerator Tracking Code AT, Middle Layer
Software, and Response Matrix Fitting

• Christoph Steier with help from James Safranek

• Monday Intro to the Middle Layer and AT
• Orbit errors
• Tuesday
• Beam measurements using SPEAR 3
• Beta function, phase advance, beam based
  alignment
• Wednesday Gradient errors, LOCO
• Thursday Detuning with amplitude, decoherence,
  resonances
• Friday TBD

2
Starting Matlab

• Start Matlab 2008a (7.6) (AT and LOCO require at
  least Matlab version 7.x) - Matlab 7.6 is the
  latest version.
• Matlab middle layer is installed on the computers
  in the first 3 rows in the computer room
• I already have preset c\bbdiag\acceleratorcontrol
  \mml on the Matlab path, this directory contains
  many routines, including the ones that set up
  memory structures and Matlab path for different
  accelerators
• setpathals
• setpathspear3
• setpathxray
• setpathvuv
• And many others
• cd c\bbdiag here we will put all the examples
  of this class you can get the example file for
  the day from http//als.lbl.gov/als_physics/csteie
  r/uspas08/computerclass.html
• Note When in doubt, setpathals (or spear3, )
  will also restore the default lattice.

3
Tuesday Gradient Errors
Add the perturbed beta to figure(1) figure(1)
subplot(2,2,1) hold on plot(MuX0, BetaX1,
'r') subplot(2,2,3) hold on plot(MuY0,
BetaY1, 'r') subplot(2,2,2) hold on
plot(MuX0, MuX1/(2pi) , 'r') subplot(2,2,4)
hold on plot(MuY0, MuY1/(2pi) , 'r') Plot
beta beat figure(2) subplot(2,2,1) plot(s,
BetaX1./BetaX0) xlabel('s m') title('Beta
Beat from the Nominal Model') subplot(2,2,2)
plot(s, BetaY1./BetaY0) xlabel('s
m') subplot(2,2,3) plot(MuX0,
BetaX1./BetaX0) xlabel('\mu_x
2\pi') subplot(2,2,4) plot(MuY0,
BetaY1./BetaY0) xlabel('\mu_y 2\pi')
Restore the lattice setsp('QF', qf, 7 1)

• Plot the beta function and phase advance for
  the
• nominal model and a model with a gradient
  error
• setpv('BPMx','Status',1,1 76 2)
• setpv('BPMy','Status',1,1 76 2)
• Get beta phase at all elements in the AT
  model (Middle Layer)
• Tune0 gettune
• BetaX0, BetaY0, s modeltwiss('Beta')
• MuX0, MuY0 modeltwiss('Phase')
• Plot beta vs. position
• figure(1) clf
• subplot(2,2,1) plot(MuX0, BetaX0, 'b')
  ylabel('Beta X')
• subplot(2,2,3) plot(MuY0, BetaY0, 'b')
  ylabel('Beta Y')
• subplot(2,2,2) plot(MuX0, MuX0/(2pi) , 'b')
  ylabel('Phase X')
• subplot(2,2,4) plot(MuY0, MuY0/(2pi) , 'b')
  ylabel('Phase Y')
• Perturb the lattice at 1 quadrupole, QF(7,1)

4
Tuesday FFT Analyze

• Simulate a phase advance measurement
• Get the orbit, beta, and phase at all elements
  in the AT model (Middle Layer)
• Tune0 gettune
• BetaX0, BetaY0, s modeltwiss('Beta',
  'BPMx')
• MuX0, MuY0 modeltwiss('Phase', 'BPMx')
• Starting condition or tracking (0.1mm)
• X0 0.0001 0 0.0001 0 0 0
• Track for 1024 turns
• global THERING
• X1 ringpass(THERING, X0, 1024)
• size(X1)
• Track coordinates for every turn along the
  ring (to all BPMs)
• BPMindex findcells(THERING, 'FamName', 'BPM')
• BPM findorbit4(THERING, 0.0, BPMindex)
• X2 linepass(THERING, X1, BPMindex)
• size(X2)
• Recover matrix structure (turns x BPM)

5
Tuesday FFT Analyze

• Calculate fractional tunes (interpolating FFT,
  sine window)
• nux, nuy, ax, ay findfreq(BPMx, BPMy)
• Calculate phase at every BPM
• (integral convolution with sine and cosine
  trajectories)
• MuX, MuY calcphase(nux, nuy, BPMx, BPMy)
• Calcphase asks for a frequency, typically just
  accepting
• the precalculated result is fine.
• Compare the 'measured' phase advance with the
  computed nominal one.
• DeltaMuX MuX()-MuX0/(2pi)
• DeltaMuY MuY()-MuY0/(2pi)
• figure(2)
• subplot(2,1,1)
• plot(MuX0, DeltaMuX-DeltaMuX(1), '.-b')
• subplot(2,1,2)
• plot(MuY0, DeltaMuY-DeltaMuY(1), '.-b')

6
Tuessday FFT Analyze

• Add noise to the BPM data and recalculation the
  phase
• BPMxNoise BPMx 5e-6randn(size(BPMx))
• BPMyNoise BPMy 5e-6randn(size(BPMy))
• Calculate fractional tunes (interpolating FFT,
  sine window)
• nux, nuy, ax, ay findfreq(BPMxNoise ,
  BPMyNoise)
• Calculate phase at every BPM
• (integral convolution with sine and cosine
  trajectories)
• MuXnoise, MuYnoise calcphase(nux, nuy,
  BPMxNoise, BPMyNoise)
• Calcphase asks for a frequency, typically just
  accepting
• the precalculated result is fine.
• Compare the 'measured' phase advance with the
  computed nominal one.
• DeltaMuX MuXnoise()-MuX0/(2pi)
• DeltaMuY MuYnoise()-MuY0/(2pi)
• figure(2)
• subplot(2,1,1) hold on

7
Tuessday FFT Analyze

• Now put a quadrupole error in the lattice and
  repeat
• steppv('QF',1,6 1)
• Track for 1024 turns
• X1 ringpass(THERING, X0, 1024)
• Track coordinates for every turn along the
  ring (to all BPMs)
• X2 linepass(THERING, X1, BPMindex)
• Recover matrix structure (turns x BPM)
• BPMx reshape(X2(1,), 1024, 122)
• BPMy reshape(X2(3,), 1024, 122)
• Add noise to the BPM data and recalculation the
  phase
• BPMxNoise BPMx 5e-6randn(size(BPMx))
• BPMyNoise BPMy 5e-6randn(size(BPMy))
• Calculate fractional tunes (interpolating FFT,
  sine window)
• nux, nuy, ax, ay findfreq(BPMxNoise ,
  BPMyNoise)
• Calculate phase at every BPM
• (integral convolution with sine and cosine
  trajectories)
• MuXnoise, MuYnoise calcphase(nux, nuy,
  BPMxNoise, BPMyNoise)
• Calcphase asks for a frequency, typically just
  accepting
• the precalculated result is fine.

8
Tuesday Quadrupole Centers

• Beam based measurement of quadrupole centers
• quadcenter
• ? QF
• ? QF(7,1)
• ? Vertical only
• Note your data got put in c\bbdiag\accelerator
  control\machine\ALS\StorageRingData\TopOff\QMS
• quadplot