Tune and Coupling Feedback Tune and Coupling Feedback Implementation and Results Implementation and Results
from RHICfrom RHIC
Peter Cameron
OutlineOutline
• A Brief History• Coupling and Coupling Feedback• RHIC Run 7
– anomalous BTF– tune scalloping
• Plans– ‘hybrid’ tune tracker– chromaticity feedback– continuous head-tail studies
Run00 LF Schottky Pickup, first tune tracker based on SRI lockin amplifierRun01 tune tracker based on RHIC BPM module
– experience gained with tune tracking, preparations for tune feedback beginRun03 tune tracker based on Pentek VME DSP system
– first successful tune feedback ramp was first try with gold (thru transition!)– performance good, but not consistent – failed ramps with no understanding of why
Run04 mechanism of failure of tune feedback in the presence of coupling is finally understood– efforts to ramp with tune feedback end– within one week the ‘eigenmode projection’ method of coupling measurement is running
Run05 no direct progress with tune/coupling feedback, however– early in the run Rhodri and Marek visit, first installation of 3D AFE in RHIC– understanding of mains harmonics is developed– last day of the run, BBQ tracks tune up the ramp
Run06 VME BBQ tune tracker - protons– AP/Controls finally ready for coupling feedback– first attempt delivers decoupled beam to full energy with ~.001 tune control– continued good success through the run, but mains harmonics prevent transition to full
operational status– last week of run - feedbacks used for first successful acceleration of polarized beam to 250GeV– the need for chromaticity control becomes ever more obvious
Run07 VME BBQ tune tracker – gold ions– chrom feedback planned for startup, but Controls Interface not completed– discovery of ‘anomalous BTF’ at injection– discovery of ‘tune scalloping’ on the ramp– mains harmonics plus tune scalloping gives a new problem
Brief HistoryBrief History
Measurement of Coupling using a Measurement of Coupling using a PLL Tune TrackerPLL Tune Tracker
Frequency
Am
plitu
de
FFT of Horizontal Acquisition Plane
Start with decoupled machine
Fully coupled machine: = |C-|
Only horizontal tune shows up in horizontal FFTGradually increase coupling Vertical mode shows up & frequencies shift
Hor
Ver
Set TunesHV
courtesy Rhodri Jones
x
y
Q1
Q2
A1,xA2,x
A1,y
A2,y
x
y
Q1
Q2
A1,xA2,x
A1,y
A2,y
x
y
Q1
Q2
1,x
1,y
2,x
2,y
Tracking the vertical mode in the horizontal plane &vice-versa allows the coupling parameters to be calculated
Measurement of Coupling using a Measurement of Coupling using a PLL Tune TrackerPLL Tune Tracker
Measurement of Coupling using a Measurement of Coupling using a PLL Tune TrackerPLL Tune Tracker
Eigenmode 2
Qx,0
Qy,0
|C-|
Eigenmode 1
Eig
en
mo
deo
r u
np
ert
urb
ed
tu
ne
va
lue
Co
up
ling
Am
plit
ud
e o
r u
np
ert
urb
ed
tu
ne
sp
lit
Eigenmode 2
Qx,0
Qy,0
|C-|
Eigenmode 1
Eig
en
mo
deo
r u
np
ert
urb
ed
tu
ne
va
lue
Co
up
ling
Am
plit
ud
e o
r u
np
ert
urb
ed
tu
ne
sp
lit
Fully coupled Tunes entirely defined by coupling – tune feedback would break here
courtesy Rhodri Jones
Tune Loop Simulink DiagramTune Loop Simulink Diagram
• PLL loop BW - a few Hz to 10Hz• Magnet loop BW ~1Hz, limited by magnet/power supply BW• Coupling loop has similar structure• Coupling loop BW ~0.1Hz• Loops are both ‘Single Input, Single Output’
OutlineOutline
• A Brief History• Coupling and Coupling Feedback
• RHIC Run 7RHIC Run 7– anomalous BTF– tune scalloping
• Plans– ‘hybrid’ tune tracker– chromaticity feedback– continuous head-tail studies
RHIC Run 7RHIC Run 70. MAINS HARMONICS REMAIN THE PRIMARY OBSTACLE1. Anomalous beam transfer function – cause is not understood (next slide)2. Tune, coupling, and chromaticity drifts due to persistent currents at
injection (also next slide)3. ‘Tune scalloping’ 4. Mains harmonics plus tune scalloping gives a very tight ‘box’5. Effect of chromaticity and tune scalloping on tune feedback6. Large noise introduced when magnet loops were closed – again, from
anomalous BTF? Kills chrom feedback at injection7. A variety of other problems
– IPM, Artus, injection damper, button monitor all turning themselves on at inappropriate times
– 5 FEC VME crate would hang frequently– ‘cache corruption’ problem would break feedback ramps
=30 deg
conventional delta-wye 6 phase 12 pulseconventional delta-wye 6 phase 12 pulse
LOAD
A
B
=30 deg
1 3 5
4 6 2
1 3 5
4 6 2
firing order:A1, A2B1, B2A3, A4B3, B4A5, A6B5, B6
Phase between A and B bridges
depends on mains voltage (booster
on/off, ???...)
With phase shift between A and B
bridges, this reproduces the
pattern used in the simulation.
RHIC 6 phase 12 pulseRHIC 6 phase 12 pulse
The simulationThe simulation• 12 free parameters for main dipole PS
– turn-on time of each of the 12 SCRs– also controls turn-off time of preceeding SCR
• 24 free parameters for simulation– turn-on time of each of the 12 SCRs– pulse length of each of the 12 phases
• SCR rise time ~10usec• Simulation time resolution ~14usec
– this is what was convenient– later reduced to ~7usec, no qualitative difference
familiar pattern
3D AFE
M Turn BPM
Simulation
Seen heremains harmonicsIPM firingArtus firingpower supply oscillating?
Mains Harmonics in Time DomainMains Harmonics in Time Domain
1.67msec
720Hz
Amplitude is tens of micronsBBQ sensitivity is ~10nm
Vertical Artus kicker firing
main dipole power supply turns on
main dipole power supply turns off
Chrom needs fixing here
This difficulty with acquiring lock was aggravated by persistent current decay, causing large and fast drifts at injection • tune ~.02 (tunes cross!!!)• coupling - full coupling• chromaticity ~3 unitsThese drifts are compensated for Run 8?
Anomalous beam response at injection • tunes separated and well decoupled• not power line, synchrotron freqs• similar in all 4 planes• serious obstacle to acquiring lock• serious contributor to ‘noise’• not seen with 245MHz PLL• disappeared with start of ramp• not understood – speculation on power supply regulation/phase shift at low current
BlueVert
‘Unexcited’ spectrum at betatron line
>30dB above noise floor!
The consequence of the previous slide - procedure to lock the The consequence of the previous slide - procedure to lock the tune trackers in the presence of multiple peaks and persistent tune trackers in the presence of multiple peaks and persistent current decay was ~40 steps!current decay was ~40 steps!
1. take BTFs to determine where the optimum line is for locking2. reduce proportional gains (can’t acquire initial lock with optimal gains)3. set the search windows so that H cannot lock on V, and V cannot lock on H4. turn on the H kicker, acquire a lock. This inevitably is not a lock on the line you
want, but rather a lock on the first line the BBQ comes to.5. nudge the search window to push the BBQ onto the correct line. This often
requires repeated nudges.6. tweak the phase to optimize response amplitude - because of many peaks BTF
app does not reliable give optimum phase information, has to be done manually7. turn up proportional gain8. adjust desired amplitude for kicker feedback9. open the search windows10. turn on kicker feedbacks11. repeat steps 1-10 for vertical. If we are running both beams repeat this whole thing
for the other ring12. tell shift leader BBQ is ready
‘Tune Scalloping’ with TF loops open• BBQ very precisely drives a slice up out of the tune distribution• tune shifts as amplitude increases (sextupoles dominate? Large octupole for t?) the slice moves off resonance, depopulates• BBQ falls off that slice, locks back in the center of the distribution• the process repeats
Contributing factors• large kicker excitation• large loop gains• small chromaticity
Kicker turned down here, then started locking on mains harmonics
scalloping
Makes for a tight ‘box’• need large kicker excitation because of mains harmonics• need large loop gains for reliable tracking of fast tune changes• chromaticity feedback not yet implemented – chrom control is not particularly good
A delicate balance to tune the tracking
Ramp 8245
Indication of small chromaticity
Limit of tune shift defined by dynamic aperture?
dots are Artus kicked tune measurements
Three successive wiggles ramps during beam commissioningThree successive wiggles ramps during beam commissioning
second ramp – turn down kickers, BBQ captured by mains
first ramp – we see scalloping
Ramp 8253
Ramp 8254
Ramp 8255
third ramp – turn kickers back up, again see scalloping
Third ramp, and effect of scalloping on chrom measurementThird ramp, and effect of scalloping on chrom measurement
This is NOT problem with chrom algorithm phase
tunes
During Run 7 BBQ worked very well, despite many issuesDuring Run 7 BBQ worked very well, despite many issues
0. MAINS HARMONICS REMAIN THE PRIMARY OBSTACLE1. Anomalous beam transfer function – cause is not understood (next slide)2. Tune, coupling, and chromaticity drifts due to persistent currents at
injection (also next slide)3. ‘Tune scalloping’ 4. Mains harmonics plus tune scalloping gives a very tight ‘box’5. Effect of chrom and scalloping on tune/chrom
measurement/feedback6. Large noise introduced when magnet loops were closed – again, from
anomalous BTF? Kills chrom feedback at injection7. A variety of other problems
– IPM, Artus, injection damper, button monitor all turning themselves on at inappropriate times
– 5 FEC VME crate would hang frequently– ‘cache corruption’ problem would break feedback ramps
Saturday – ramp 8211 (the first wiggles ramp)Chromaticities and tunes
Chrom measurement affected by tune scalloping here
dots are Artus kicked tune measurements
The first wiggles ramp came 10 ramps after the first
successful TF ramp. Chrom was not corrected until
the following Wednesday evening
Effect of chrom on TF
Loops open ‘ungracefully’, BBQ continues to track
Effect of chrom on Tune and Coupling Feedback – ramp 8217
Scalloping with TF loops closed
Chrom small here
Vert amplitudeand phase
horiz amplitudeand phase
vert tune scalloping with TF on
Loop gains and kicker excitation turned down here, back up here, down again here
H tune
V tune
Ramp 8667IBS lattice
Chromaticity via radial modulation with BBQ in RHIC Run 7 two successive ramps for correcting and measuring result
Chromaticity corrected here and here
With ongoing tuning we eventually had reasonably good chrom measurements on
the ramp
first ramp
second ramp
Simultaneous tune and coupling feedback in both rings , ~97% transmission
Loops closed thru transition
Ramp 8498
Yellow beam current
H tune
V tune
Blue beam current
H tune
V tune
Slight tune swing down to keep between mains during ramping
With ongoing tuning we eventually had good feedback ramps
OutlineOutline
• A Brief History• Coupling and Coupling Feedback• RHIC Run 7
– anomalous BTF– tune scalloping
• Plans– ‘hybrid’ tune tracker– chromaticity feedback– continuous head-tail studies
• Dynamic range at transition was the showstopper for the 245MHz PLL• 3D AFE and BBQ appeared to be the solution to this problem
– Large dynamic range of the diodes – first components in the signal paths– ~160dB suppression of the revolution line plus ~10nm sensitivity
• However, BBQ operates in the coherent spectrum– We discover mains harmonics, which are ~70dB above BBQ noise floor, ~50dB
more severe in RHIC than Tevatron or SPS– We discover the ‘anomalous BTF’ at injection with ions– No obvious solution on the horizon for either of these problems in the coherent
spectrum, they both are solid obstacles to making tune tracking feedback operational
• lock acquisition at injection cannot be automated• needed kick (due to mains harmonics) blows up emittance
• So how about a ‘hybrid’ system? Direct diode detection of 245MHz pickup• Some data from last run (following slides), many questions
– Noise floor? At high frequency?– Coherence at high frequency?– Performance of 3D AFE peak detector with resonant pickup?– Phase compensation? Mix in tunnel?
‘‘Hybrid’ tune trackerHybrid’ tune tracker
Ramp 8687 yellow BBQRamp 8687 yellow BBQ
Ramp 8687 yellow LF SchottkyRamp 8687 yellow LF Schottky
Ramp 8687 (early May)Ramp 8687 (early May)
Yellow LF Schottky 245MHz signal as seen thru 3D AFE
Yellow 1m long stripline signal
as seen thru 3D AFE
What is this???• ‘clock’ jitter? Blue is master• higher order transverse modes?
• we consistently see stronger beam response below t
• else?
Ramp 8687 around transitionRamp 8687 around transition
Yellow LF Schottky 245MHz signal as seen thru 3D AFE
Yellow 1m long stripline signal
as seen thru 3D AFE
Ramp 8687 time domain around transitionRamp 8687 time domain around transition
-13 sec 0 +12
IPM firingYellow LF Schottky 245MHz signal as seen thru 3D AFE
Yellow 1m stripline signal as seen thru 3D AFE
245MHz pickup
existing 3D AFE
245MHz kicker
Possible ‘Hybrid’ Tune Tracker Block DiagramPossible ‘Hybrid’ Tune Tracker Block Diagram
existing BBQprocessing
AGC
tunnel
1002a
beam
motion control
pickup sum mode
30.625 MHzbandpass
diode doubler
61.25 MHzbandpass
diode doubler
122.5 MHzbandpass
diode doubler
245 MHzcavity filter
245 MHzcavity filter
Advantages of this approach to ‘hybrid’:• minimal phase compensation needed• excite the same spectrum you are measuring• ‘bleedthru’ less problematic• minimal modification to existing BBQ
• Transition Chromaticity Monitor• ‘Hybrid’ tune tracker?• Chromaticity measurement and feedback• Other Stuff
– Beam transfer functions– Streaming audio– Coupling echoes?
OutlineOutline
• Scenario 1 – no anomalous BTF at injection with ions– Early on (before any ramps in blue?)
• turn on tune and coupling feedback, turn on radial wiggles, assess quality of chromaticity measurement (all hardware and software is in place)
• If good, ramp in this configuration (it will improve chrom measurement on the ramp, therefore will speed ramp development)
– When controls infrastructure is sorted out, do APEX ramps with chrom feedback
– During proton ramp development – use chrom feedback (250GeV!!!)
• Scenario 2 – anomalous BTF at injection with ions, no anomalous BTF at injection with protons– During APEX at store with ions
• turn on tune and coupling feedback, turn on radial wiggles, assess quality of chromaticity measurement
• If good, close chromaticity feedback loop and explore
– During proton ramp development – use chrom feedback
• Scenario 3 – anomalous BTF at injection with ions AND protons– Hybrid tune tracker
Run 8 Plan for Chromaticity Feedback Run 8 Plan for Chromaticity Feedback (assuming present BBQ tune tracker)(assuming present BBQ tune tracker)
p
we excite ‘on-momentum’
longitudinal beam profile
p
sum signal
(with positive chromaticity)In betatron phase• particles from the head arrive at the tail early• particles from the tail arrive at the head late
(scaled) difference signalis time shifted
sum signal
(scaled) difference signalis time shifted
the (scaled) difference of the sum and difference signals looks like this
the (scaled) difference of the sum and difference signals
rectified difference of the sum and difference signals
sum signal
FFT showing relative amplitudes of the 3 signals for a 0.1 radian phase shift and 1ns rms bunch length
the (scaled) difference of the sum and difference signals
rectified difference of the sum and difference signals
sum signal
zoom on the previous FFT
BPM off-set
hi-pass
BBQ AFE
a possible processing chain
Other Stuff• Beam transfer functions – improvements?
– White noise? Chirped?– Beam-beam transfer functions?
• Coupling BBTF for LHC orbit feedback - Steinhagen
• Streaming audio of BBQ – Internet radio• Coupling echoes?
Backup material
Top Related