1. The SPS Transverse Damper
Acknowledgements: BE-RF-FB,
BE-RF-CS,
BE-RF-PM,
BE-OP,
BE-ABP
G. Kotzian for the SPS Damper Team
BE-RF-FB, 20. September LIU-SPS Wideband Feedback Review .
20. Sep 2016
SPS Damper - Gerd Kotzian

2. SPS Damper - Gerd Kotzian
Outline
Objective:
Have a look at the SPS Transverse Damper, illustrate recent upgrades carried out during LS1, highlight present feedback performance, and depict potential for future upgrade.
Structure:
Brief historical overview of the SPS Transverse Damper
Technical design and layout after Long Shutdown 1 (LS1)
Present performance and known limitations
20. Sep 2016
SPS Damper - Gerd Kotzian

3. Transverse Damper in General
The transverse damper is a feedback system: it measures the bunch-by-bunch oscillations and damps them by fast electrostatic kickers.
BPM
Kicker
5ns Bunches
Active closed loop feedback
BBQ
Analog
Front
End
Signal
Processing
SUM
ADC
DAC
Power Amp
transverse
position
pre-processed
sampled
position
calculated
correction data
correction
signal
drive signal
20. Sep 2016
SPS Damper - Gerd Kotzian

4. Former SPS Transverse Feedback System
Last upgrade in preparation of the SPS as LHC injector
Damping of injection errors and cure of resistive wall impedance driven coupled bunch instability (threshold ~5x1012 protons total intensity)
Operates since 2001 between lowest betatron frequency and 20 MHz (V & H plane) to damp all possible coupled bunch modes at 25 ns bunch spacing
Tetrode amplifiers with two tubes drives kicker plates in push-pull configuration installed in tunnel under kicker, W drive power per tetrode, 3 kHz to 20 MHz
Handles large injection errors of the order of several mm
Technology of electronics of system used in SPS until dates from late 1990’s
20. Sep 2016
SPS Damper - Gerd Kotzian

5. Scope and Motivation for SPS damper upgrade under LIU
Why LS1 (2013 – 2014)?
Sharing pick-ups with Orbit System (MOPOS) from Beam Instrumentation Group incompatible with MOPOS upgrade
Requirements for single bunch damping for protons used for LHC p-ion Physics (closer spaced proton bunches, 100 ns)
Ions injection damping with fixed frequency scheme, closer spaced batches
Individual bunch damping for crab cavity studies not possible with previous system (sampling was not synchronous with bunch in previous system)
LHC doublet scrubbing beam incompatible with previous system
Controls with G64 chassis and MIL-1553 (a.k.a. MMI) became obsolete  new controls for power system and LLRF, new RF function generators
20. Sep 2016
SPS Damper - Gerd Kotzian

6. Excellent availability!
SPS Damper Loops
pLHC
RF Frontend
200 MHz
BeamPos1
DSPU1
Switch M1
Damper Signal Selector
ADC1
DAC1
pLHC
Switch M2
ADC2
BeamPos2
DSPU2
DAC2
pLHC
cfv-ba2-alltrdamp[hv]1
SPS Damper Loops
pFT
Baseband Frontend
BeamPos1
DSPU1
ADC1
DAC1
pFT
DSPU2
ADC2
BeamPos2
DAC2
pFT
from BBQ
SPS Damper Loops
Ions
RF Frontend
200 MHz FSK
ADC1
BeamPos1
DSPU1
DAC1
Ions
ADC2
BeamPos2
DSPU2
DAC2
Ions
cfv-ba2-alltrdamp[hv]2
from VNA
SPS Damper Loops
Scrub
Baseband Frontend
(40 MHz)
from VNA
ADC1
BeamPos1
DSPU1
DAC1
Scrub
BeamPos2
DSPU2
ADC2
DAC2
Scrub
Stripline pickup
processing
pLHC,
Scrub
Ions
BPM
Proc.
BPM
Proc.
PLC1
PLC2
200 W
Driver
200 W
Driver
LLRF MODULES:
6 commissioned
2 Experimental
2 Ions (2016)
SURFACE BA2
TUNNEL LSS2
pFT
Damper Power System
Electrostatic PU processing
BW ~100MHz
BPM
Proc.
BPM
Proc.
Final
Stage
Final
Stage
POWER:
Excellent availability!
BEAM
PU1
PU2
KICKER1
KICKER2
20. Sep 2016
SPS Damper - Gerd Kotzian

7. 6x Beam Position Monitors installed in LS1
2 BPCR (H/V) for LHC type beams
(couplers maximum 200 MHz)
2 BPH electrostatic PU (pFT)
2 BPV electrostatic PU (pFT)
RF Faraday cage
SPS Transverse FB
BPCR.221
all BPMs re-cabled with 7/8-inch smooth wall coaxial cables during LS1; [length ~ m]
BPH/BPV: electronics and amplifiers in the tunnel tested and verified; "hot-spares" readily checked and kept in the tunnel
BDH / BDV kickers unchanged
BPH/BPV electronics (in pits right below the BPMS) are extremely robust.
Very reliable also during operation 2015/16.
20. Sep 2016
SPS Damper - Gerd Kotzian

8. Power System Consolidated during LS1
new process control based on PLC in BA2
 FESA remotely accessible
EM field simulation of kicker
Tetrode amplifier in SPS LSS2
20. Sep 2016
SPS Damper - Gerd Kotzian

9. SPS Damper - Gerd Kotzian
SPS Damper Racks BA2
Local consoles in BA2, as of 2015 most of the controls done remotely
20. Sep 2016
SPS Damper - Gerd Kotzian

10. SPS Damper Loops Module
Features:
8 modules installed, separate boards for H- and V-plane and beam type (pFT, pLHC, Ions, Scrub)
Delay adjustment:
Splitting ADC and DAC clocks, four different clock domains for delay adjustment 8 ADCs (16 bit, 120 MSPS)
4 DACs (14 bit, 120 MSPS)
Signal processing:
40 MSPS bunch synchronous
Two analog frontend types:
RF frontend, used for pLHC and Ions, LO = 200 MHz, I/Q components
Baseband frontend (BB), used for pFT and Scrub, only I-component
Input gain control: RF frontend
External: Input termination / attenuation / amplifier
RF: switching attenuators + amp, delay matched Large dynamic range: -39 dB to + 20 dB = 59 dB
Input gain control: BB frontend
BB: variable gain amplifier (VGA)
Dynamic range: dB to + 26 dB = 53.4 dB
This image:
EDA EDA Mezzanine .. SPS Transverse Damper Loops
Not shown:
EDA Damper RF Frontend
EDA Damper Baseband Frontend
EDA Damper Signal Selector
EDA RF Function Generator Module
AED SPS Damper 120 MHz Clock Generator
20. Sep 2016
SPS Damper - Gerd Kotzian

11. Signal processing (simplified): pLHC 25ns beam
Analogue front-end
(here for LHC beam)
Digital Board
ExtCtrlA
H1 (V1)
H2 (V2)
ExtCtrlB
Output gain control (12 bit)
variable delays for clocks ADC/DAC (120 MHz)
RfGainCtrl
(BbGain)
Courtesy: D. Valuch
20. Sep 2016
SPS Damper - Gerd Kotzian

12. Digital Signal Processing
2 Beam Position Modules per Damper Loop
Two inputs per pickup: Delta and Sigma
Calculates normalized transverse position: x n =Δ/Σ
Digital Signal processing Unit (DSPU)
Notch filter
PU vector sum / Phase shifter (up to 16 taps)
Bunch masking
Phase equalizer FIR (up to 64 taps), linearize phase of power system, operated at 40 MSPS
Gain equalizer (up to 64 taps), digital low pass shaping output signal, operated at 120 MSPS
 Mostly set up once during commissioning.
Closed loop delay compensation
Multiple clock domain crossing
Fixed delay, i.e. cycle dependent delays per beam type, cycle independent delays, and variable delay
ADC sampling delay (adjust during setting-up at flat bottom)
DAC fine delay (adjust with closed loop measurements)
Delay compensation for momentum change, i.e. function(fREV), including stable phase program
 Automated in LSA using make rules
Phase compensation for sidebands (Q20 vertical)
20. Sep 2016
SPS Damper - Gerd Kotzian

13. Transverse Feedback – Key Parameters
Closed Loop Feedback Response:
𝐶𝐿 𝑠 = 𝑋 𝑠 𝑌 𝑠 = 𝑂𝐿 𝑠 1+𝑂𝐿 𝑠
𝑂𝐿 𝑠 =𝐺 𝐻 𝑆 𝐻 𝐶 𝐻 𝐴 … open loop transfer function
Stability Criterion:
𝐺 𝐻 𝑆 𝐻 𝐶 𝐻 𝐴 <(−1)
with:
𝐻 𝑆 𝐻 𝐶 𝐻 𝐴 =𝐾⋅ 𝑒 𝑗 (𝜔𝑇+𝜙 𝑃𝐾 )
Stability Conditions
abs(𝐺)⋅𝐾<1
arg 𝐺 +𝜔𝑇+ 𝜙 𝑃𝐾 =− 180 ∘ ±tol.
Key Parameters:
𝐾……. Feedback loop gain
𝜙 𝑃𝐾 … Feedback phase
𝑇……..Total loop delay
𝐷𝑒𝑙𝑎𝑦 𝑻
𝑃ℎ𝑎𝑠𝑒 𝝓 𝑷𝑲
𝐺𝑎𝑖𝑛 𝑲
Denominator ≠𝟎 !
Closed loop stability requires
these parameters to be properly adjusted.
𝑁𝐵: 𝜙 𝑃𝐾 depends on frequency, i.e. USB = 180 ∘ LSB
20. Sep 2016
SPS Damper - Gerd Kotzian

14. Parameter Extraction: BTF Measurements[1]
Classical method with VNA (Vector Network Analyzer)
VNA excites transverse resonances (sinusoidal excitation)
Narrowband response recorded per frequency  Eigenvalues {Magnitude(freq), Phase(freq)}  represented as POLAR plot
Feedback phase and loop delay adjustment
Repeated over several frequency bands
dB(RX/TX)
frequency 𝑞
(1−𝑞)
KNOB 1
KNOB 2
Target: aligned
NB: arg 𝐺 +𝜔𝑇+ 𝜙 𝑃𝐾 =− 180 ∘
Effect of PHASE 𝝓 𝑷𝑲
Effect of DELAY 𝑻
+ℑ𝒎
+ℑ𝒎
+ℑ𝒎 𝜔𝑇
[1] J. Borer, G. Guignard, A. Hoffmann, E. Peschardt, F. Sacherer and B. Zotter, “INFORMATION FROM BEAM RESPONSE TO LONGITUDINAL AND TRANSVERSE EXCITATION”, in IEEE Transactions on Nuclear Science, Vol. NS-26, No. 3, June 1979.
+Δ𝜑 𝕽𝒆 𝕽𝒆 𝕽𝒆
−Δ𝜑
−ℑ𝒎
−ℑ𝒎
−ℑ𝒎
20. Sep 2016
SPS Damper - Gerd Kotzian

15. 2015/2016: Semi-Auto Parameter Extraction
Open loop response (polar):
Basis is the classical method with VNA(s)
automation with scripts
automatic data collection
Measure OL 1 MHz to 20 MHz
(here: Meas39, BPCR.221.V  V2)
20 sweeps x SC period … ~20-30 min
Post-processing allows parameters to be extracted, e.g.
Phase advance PU  Kicker
Feedback COARSE + FINE delay
Tune
Overall loop phase response
 Now fully integrated in FESA, available for regular operation from CCC
(Meas /11/04)
20. Sep 2016
SPS Damper - Gerd Kotzian

16. Feedback Parameter Extraction – An Alternative Approach
Feedback Phase:
Kick by ADT: all bunches excited coherently during one turn:
Loop Delay: two consecutive measurements,
COSINE, N periods per revolution, 1 turn excitation + record oscillation
SINE, N periods per revolution, 1 turn excitation + record oscillation
Initial condition
𝜙 𝑃𝐾 … constant phase advance after kick, for the first turn and all subsequent
𝑧 𝑛 =𝐴 𝑛 exp(𝑗𝛼 𝑛 ) where 𝛼 𝑛 =2𝜋𝑞⋅𝑛+ 𝜙 𝑃𝐾
Batches of
24 bunches
COSINE, or in-phase term
N = 2 periods per revolution
ADTmDSPUVerM2B1
Acquisition date: 26-Jun :48:
SINE, or quadrature term
Acquisition date: 26-Jun :50:
[1] G. Kotzian and D. Valuch, “SPS AND LHC TRANSVERSE FEEDBACK PARAMETER EXTRACTION: AN ALTERNATIVE APPROACH”, in LHC Beam Operations Committee, 08 December 2015, .
Shown here: LHC ADT
20. Sep 2016
SPS Damper - Gerd Kotzian

17. SPS Damper - Gerd Kotzian
Feedback Loop Delay (2)
Measurement and analysis:
COSINE, N periods per revolution, 1 turn excitation + record oscillation
SINE, N periods per revolution, 1 turn excitation + record oscillation
Polar plot of data in the same turn  Loop Delay
ADTmDSPUVerM2B1
Acquisition date:
26-Jun :48: Jun :50:
DELAY
Wind-up bunch positions by Δ𝑇 𝑛 =𝑛⋅𝑁/3564⋅2𝜋
Shown here: LHC ADT
20. Sep 2016
SPS Damper - Gerd Kotzian

18. SPS Damper - Gerd Kotzian
COARSE Loop Delay
Measurement and analysis:
COSINE, N periods per revolution, 1 turn excitation + record oscillation
SINE, N periods per revolution, 1 turn excitation + record oscillation
Wind-up one turn: Δ𝑇 𝑛 =𝑛⋅𝑁/3564⋅2𝜋
Polar plot of data in the same turn  Loop Delay
ADTmDSPUVerM2B1
Acquisition date:
26-Jun :48: Jun :50:
Coarse Delay OK
For this measurement N=2.
By increasing N, i.e. the number of excitation periods per revolution the time resolution can be further refined.
FINE DELAY
Shown here: LHC ADT
20. Sep 2016
SPS Damper - Gerd Kotzian

19. FINE Delay Measurement (1)
If we excite with N = 891 then every other bunch should be kicked, i.e. 10 MHz
(Note that the excitation frequency is locked to the RF):
SINE, or quadrature term
N = 891 periods per revolution:
 exciting ODD bunch numbers
COSINE, or in-phase term
N = 891 periods per revolution:
 Exciting EVEN bunch numbers
[1] G. Kotzian and D. Valuch, “SPS AND LHC TRANSVERSE FEEDBACK PARAMETER EXTRACTION: AN ALTERNATIVE APPROACH”, in LHC Beam Operations Committee, 08 December 2015, .
 The combination of the two results in a 4-QAM, a.k.a. QPSK
NB: increasing N results in a reduced excitation due to limited BW of the ADT power amplifier + kicker.
20. Sep 2016
SPS Damper - Gerd Kotzian

20. FINE Delay Measurement (2)
ADTmDSPUVerM1B1
Acquisition date:
26-Jun :10: Jun :50:
 Make use of information content in subsequent decay turns
Subsequent turns after excitation with N=891 (10 MHz)
Each circle = average over 72 bunches
Results of two Pickups (blue Q7, red Q9)
Recommended
Delay adjustment:
𝑇 𝐹𝐼𝑁𝐸 =−0.45 ns
Remember: result is generated by a system running at 40 MSPS processing, or 𝑇 𝑆𝐴𝑀𝑃𝐿𝐼𝑁𝐺 =25 ns
 Methods to be implemented within the newly installed ObsBoxes in 2016 …
Shown here: LHC ADT
20. Sep 2016
SPS Damper - Gerd Kotzian

21. Performance Example: LHC25NS
GOAL:
Feedback system well adjusted to (visibly) suppress tune lines.
HORIZONTAL
VERTICAL
NB: tunes are at
Qh = 0.13
Qv = 0.18
20. Sep 2016
SPS Damper - Gerd Kotzian

22. Performance Example: SFTPRO2
Fixed Target beam:
5ns bunch spacing
Half a machine turn injected
Damping times of 20 turns achieved
Analysed data provides valuable information about machine parameters
Diagnostics published for every injection, implemented within the newly installed ObsBoxes in 2016
20. Sep 2016
SPS Damper - Gerd Kotzian

23. Present performance and known limitations
Damper works very reliably: no more than 2 interventions per year
Running since September 2014
Designed to damp 5 mm injection error (𝛽=100m) at 26 GeV/c in 20 turns
Operates between lowest betatron frequency and 20 MHz (V & H plane) to damp all possible coupled bunch modes at 25 ns bunch spacing
Known limitations
None.
New state-of-the-art digital low level RF hardware was designed to overcome all known limitations of the previous system
No limitations, few potential ideas for improvements:
Reduce complexity in view of machine operation (daily use)
Towards a next generation RF frontend (common effort with LHC ADT)
Unify the power system with LHC ADT (same tetrodes RS2048)
20. Sep 2016
SPS Damper - Gerd Kotzian

24. Summary: Damper Status 2014 - 2016
System entirely redesigned during LS1
New signal processing, aka. DamperLoops:
6 commissioned, 2 experimental, 2 Ions modules (in preparation for Ions 2016)
Damper Power System consolidated
Excellent availability during 2014, 2015, and 2016
Special thanks to colleagues from RF-PM!
Observation Box, allows data processing for
Feedback parameters (gain, phase, delay)
Beam parameters (tune, damping time/decoherence time)
Damping performance and diagnostics published for every injection
Damper fully integrated in CERN Control System
All parameters remotely controllable (FESA)
Per-cycle configuration stored in LSA data base
Alternative methods for feedback parameter extraction
Semi-automated parameter extraction using VNAs
Coherent transverse excitation, generated by ADT
Tests can be carried out as routine health checks
Novel ideas for alternative feedback controllers on the way …
This all was possible thanks to a great team!
20. Sep 2016
SPS Damper - Gerd Kotzian

25. SPS Damper - Gerd Kotzian
THANK YOU
Questions?
20. Sep 2016
SPS Damper - Gerd Kotzian