1 of 54

Progress in the LHC Schottky Spectra Analysis

64th SY-BI Students and R&D meeting

K. Łasocha, D. Alves, T. Levens, O. Marqversen (BI-IQ)

C. Lannoy, N. Mounet (ABP-CEI)

1

2 of 54

Presentation outline

  1. Introduction and motivation
  2. Theory advances
    1. Simulation techniques
    2. Effect of octupoles
    3. Effect of impedance
    4. General Schottky spectrum
    5. Statistical properties and correlation
  3. Diagnostic advances
    • Overview of the present system
    • Typical results
    • Frequent problems
    • Most recent updates and outlook
  4. Conclusions

2

3 of 54

Synchrotron motion

3

While circulating along the machine the particles, due to the acceleration obtained in RF cavities, particles undergo longitudinal synchrotron motion.

The turn-by-turn longitudinal position of a particle can be written as:

As a consequence, particle’s momentum also oscillates.

Oscillation amplitude

Synchrotron frequency

Synchrotron phase

Revolution period

4 of 54

Betatron motion

4

As the result of focusing, mostly affected by quadrupole magnets, particles also undergo transverse betatron motion.

The turn-by-turn position of a particle in given location can be written as:

Oscillation amplitude

Betatron tune

Betatron phase

5 of 54

Tune & Chromaticity

5

To preserve beam stability, betatron tunes should avoid values such that:

nQH + mQV = k,

where n,m,k are integers.

6 of 54

Tune & Chromaticity

Courtesy: R. Steinhagen

6

To preserve beam stability, betatron tunes should avoid values such that:

nQH + mQV = k,

where n,m,k are integers.

Coll

INJ

7 of 54

Tune & Chromaticity

Courtesy: R. Jones

7

To preserve beam stability, betatron tunes should avoid values such that:

nQH + mQV = k,

where n,m,k are integers.

Coll

INJ

8 of 54

Tune & Chromaticity

8

Betatron tune is related to the particle momentum via chromaticity (another notation: Q’):

In the LHC dp/p ≈ 3×10-4, and natural, uncorrected Qξ ≈ -140, so Δq ≈ 0.04.

9 of 54

Tune & Chromaticity

9

Betatron tune is related to the particle momentum via chromaticity Qξ (another notation: Q’):

In the LHC dp/p ≈ 3×10-4, and natural, uncorrected Qξ ≈ -140, so Δq ≈ 0.04.

10 of 54

Tune & Chromaticity

10

Betatron tune is related to the particle momentum via chromaticity Qξ (another notation: Q’):

In the LHC dp/p ≈ 3×10-4, and natural, uncorrected Qξ ≈ -140, so Δq ≈ 0.04.

In addition, certain values of chromaticity have to be avoided to prevent the onrise of so called head-tail instabilities.

In the LHC, H-T instabilities arise for Q’ below or equal to 0.

11 of 54

Tune & Chromaticity

11

Betatron tune is related to the particle momentum via chromaticity Qξ (another notation: Q’):

In the LHC dp/p ≈ 3×10-4, and natural, uncorrected Qξ ≈ -140, so Δq ≈ 0.04.

Chromaticity can be corrected

using sextupole magnets.

In the LHC target value of chromaticity Q’ is in order of 5-20.

How to know that we are on target?

12 of 54

Tune & Chromaticity: measurements

The direct (and invasive) way of measuring Q and Q’ is based on kicking/modulating the beam:

  • Transversely, while observing the position: tune measurement,
  • Longitudinally, while observing the tune: chromaticity measurement.

For machine protection, this can be done only on a probe beam, before the injection of physics beam.

The non-invasive alternatives are:

  • BBQ system for tune,
  • Schottky Monitor for tune and chromaticity.

LHC design report put a requirement of controlling Q’ within 1 unit at INJ, and within 3 units at Flattop.

This can be verified only since the last year’s upgrade of Schottky system.

Courtesy: M. Krupa

12

13 of 54

LHC Schottky Monitor

Instrument designed to observe transverse oscillations of a selected bunch at frequencies dominated by incoherent particle motion.

13

M. Betz, O.R. Jones, T. Lefevre, M. Wendt, Nuclear Instruments and Methods in Physics Research Section A, vol. 874, 2017

14 of 54

LHC Schottky Monitor

Transverse sidebands

Longitudinal (central) band

14

Transverse beam spectrum around 427725th harmonic of the revolution frequency (≈ 4.81 GHz ± 7.5 kHz) with a sub-hertz resolution.

//

Signal perspective:

15 of 54

Schottky Spectrum: emittance

15

16 of 54

Schottky Spectrum: tune

16

17 of 54

Schottky Spectrum: chromaticity

17

18 of 54

Schottky Spectrum: chromaticity

18

19 of 54

Schottky Spectrum: chromaticity

19

20 of 54

Schottky Spectrum: synchrotron frequency

20

Fs

Synchrotron frequency determines the distance between the satellites

21 of 54

Schottky spectra analysis: status for 2022

Details in: K. Lasocha et al., HB 2023 382-388

Theory

  • Known impact of key beam parameters.
  • Developed matrix formalism, technique to efficiently calculate Schottky spectra.
  • Derived precise formula Q’.

Vague understanding on of how Schottky spectra are affected by:

  • Octupole fields
  • Beam impedance
  • Coupling and other non-linear effects

Diagnostics

  • Direct parameter calculation
    • requires perfectly clean spectra
  • Spectra fitting procedure
    • time and resource consuming
    • Need knowledge of the theory

Challenges:

  • Limited understanding on the hardware imprint on spectra
  • Proton signal featured strong coherent components
  • No online system neither for ions nor for protons

21

22 of 54

Schottky spectra analysis: status for 2023

22

What we expected

What we saw

23 of 54

Theory of Schottky spectra

23

24 of 54

Simulation techniques: matrix formalism

Details in: K. Lasocha’s PhD Thesis, K. Lasocha and Diogo Alves, Phys. Rev. Accel. Beams 23, 062803

Matrix formalism provides a fast way to calculate Schottky spectra when the theory is known. In short, it can be described by the following diagram:

Since recently matrix formalism is implemented as Acc-Py package SchMatrix:

https://gitlab.cern.ch/bi/schmatrix

Schottky matrix, determined by the machine parameters

Synchrotron amplitude distribution or longitudinal bunch profile

Schottky spectrum

24

25 of 54

Simulation techniques: macroparticle simulations

Details in: C. Lannoy et al. JINST 19 P03017, C. Lannoy’s PhD Thesis

If the theory is not yet known, or is to be benchmarked, the Schottky spectra can be calculated from one of the simulation packages, as PyHeadtail and XSuite.

C. Lannoy developed an effective procedure that calculates Schottky spectra from turn-by-turn position of macroparticles.

Such simulations are significantly slower than Matrix Formalism, but can include a myriad of effects not predicted in the Schottky theory.

The Schottky Monitor module is planned to be included in XSuite package soon.

Courtesy: C. Lannoy

25

26 of 54

Impact of octupoles on Schottky spectra

Details in: C. Lannoy et al., IPAC2024 WEPG32

Octupole magnets provide additional tune spread to stabilise LHC beam:

Transverse actions, proportional to the betatron oscillation amplitude

Detuning coefficients, dependent on beam optics and octupole currents

Expansion under assumption of uniform octupole currents

Modification of the Schottky spectrum is defined:

  • Transverse geometric emittance is given by the average action: εx = <Jx>, εy = <Jy>,
  • Particles contribute to the spectrum proportionally to Jx (or Jy).

26

27 of 54

Impact of octupoles on Schottky spectra

27

convolution

=

convolution

28 of 54

Impact of octupoles on emittance estimate

Theory:

No impact, the power of the spectrum is invariant on octupole current.

Experiment:

Power of spectrum over the theoretical predictions for low octupole current

28

MD 9263

29 of 54

Impact of octupoles on tune estimate

Theory:

Center of sidebands shifted, and coincides with the average tune, weighted by Jx. For Gaussian bunches:

Qx, Schottky = Qx,0 + 2 αx εx + αxy εy

Experiment:

Effect of octupoles clearly visible and agree with the theoretical predictions.

29

MD 9263

30 of 54

Impact of octupoles on chromaticity estimate

Theory:

The width of sidebands increases, changing the relation between the width and chromaticity. For Gaussian bunches:

σ± = σ±,0 + 2 αx2 εx2 + αxy2 εy2

This correction is however negligible for the LHC.

Experiment:

Changes in octupole current does not affect the chromaticity estimate.

Exception: Ioct = 0.

30

MD 9263

31 of 54

Longitudinal impedance

Details in: C. Lannoy et al. JINST 19 P03017, C. Lannoy’s PhD Thesis

Surrounded by vacuum pipe, beam interacts with the components creating EM field, that may affect trailing particles.

Integrated over the whole beam spectrum, the effect is called beam coupling impedance. It can significantly affect the beam dynamics, and deform the Schottky spectrum.

The effect of the impedance can be seen in Schottky Spectra by scanning bunches of different intensities.

Courtesy: C. Lannoy

31

32 of 54

Longitudinal impedance

C. Lannoy developed the theory for Schottky spectra that includes the longitudinal impedance.

Dependent on the impedance model, the synchrotron frequency is changed as a function of the synchrotron amplitude.

The theory was implemented in the matrix formalism and benchmarked wrt to the macroparticle simulations, assuming the broadband resonator impedance model.

Courtesy: C. Lannoy

32

Nominal frequency shift

33 of 54

Longitudinal impedance

Courtesy: C. Lannoy

Comparing the theory and simulations to the real measurements had two goals:

  • Verification of the theory
  • Benchmarking the LHC impedance model

The longitudinal model is assumed to follow the broadband resonator model with parameters:

  • Im(Z||)/n = 70 mΩ
  • fr = 5 GHz
  • Q = 1

The data collected during dedicated MD data matched the model, but with Im(Z||)/n = 135 mΩ.

Investigations ongoing on the sources of the discrepancy.

33

34 of 54

General RF system

Longitudinal motion has a dominant impact on the shape of the Schottky spectrum.

We have good models which address the dependence of:

  • Longitudinal bunch profile,
  • Synchrotron frequency,
  • Longitudinal impedance

Nevertheless:

  • Models come with their limitations and assumptions
  • Require from the user good knowledge of the beam conditions

What can we say about Schottky spectrum without taking any assumptions on RF system?

34

H. Damerau, RF manipulations, RF CAS 2023

PSB/SPS like RF bucket

LHC like RF bucket

35 of 54

Higher order chromaticity

The linear dependence of tune on the momentum is only an approximation.

In the LHC non-linear chromaticities were measured up to the fifth order. It is important to know if these high-order terms affect our spectra, and how.

One could develop theory for Schottky spectra with higher order terms included (e.g. K. Lasocha et al. IPAC’24 WEPG30), but

what can we say about Schottky spectrum without taking any assumptions on highest included chromaticity term?

Courtesy: M. Le Garrec,

IPAC'23 MOPL027

35

Beam 1 Horizontal

36 of 54

Generalised theory

Details in: K. Lasocha et al. Phys. Rev. Accel. Beams 27, 112801

Main conclusions:

  • Irrelevant of the RF waveform and chromaticity, the Schottky spectrum consist of series of satellites, distributed in intervals of Fsynch.
  • Transverse emittance is invariantly given by the the power of the spectrum.
  • The centre of the spectrum coincides with the average tune of the bunch, taking into the account chromatic tune shifts.
  • Baseline method for 1st order chromaticity estimate remains valid for arbitrary RF system. Presence of higher order terms requires corrections, derived in a closed form.

Using a realistic longitudinal motion (X-Suite simulation) and high-order chromaticities measured in the LHC, the developed theory estimates the required chromaticity corrections below 0.1.

36

Example Schottky spectrum for 2 harmonic RF system

37 of 54

Statistical properties of spectra

Schottky spectra are inherently random, and require averaging if their shape is to be determined.

The power of each bin follows exponential distribution in time:

  • Observed experimentally (C. Lannoy’s PhD)
  • Heuristic argument (Reyleigh, Nature 72, 318)

This fact can be used to filter out the noise from the signal in the spectrum, as for the bins containing Schottky signal, we expect:

37

Number of samples

38 of 54

Coupling between the planes

If the motion is coupled between the planes, we see a copy of vertical signal in the horizontal signal, and vice versa.

The coupled components can be detected by calculating Pearson correlation coefficients of the time signal between the respective bins from both planes.

This can serve as an additional filter, or be used to correct the spectra and determine the coupling coefficients.

38

During MD 14343 we tried to measure previously introduced coupling. The results were not conclusive.

39 of 54

LHC Schottky Diagnostic system

39

40 of 54

Online Schottky analysis pipeline (2024)

40

NXCALS

44.9 kHz

1 Hz

0.1 Hz

  • FFT calculation
  • Hardware control
  • Initial data processing:
    • Data buffering,
    • Baseline correction,
    • Mean, std and Pearson coeff.

  • Data filtering,
  • Coupled components correction (from Oct)
  • Curve fitting of the resulting signal

Accpit

Expert GUI

41 of 54

Curve fitting

Previous attempts to measure the Q’ from the Schottky spectra were based on fitting the spectra.

The chromaticity is however given by:

so fitting the spectrum puts too big focus on centres, and too little on tails.

The adopted procedure assumes that the original spectrum is bell-shaped (q-Gaussian or super-Gaussian) and fits the variance with 5 parameters:

  • Left and right sideband width
  • Spectrum magnitude
  • Additional curve parameter (q or order)
  • Betatron tune

(Frequency - tune)2 ⋅ Spectrum

Spectrum

41

42 of 54

Typical performance (physics fill)

42

Fill 10074

43 of 54

Q’ measurements overview (proton Flattop)

43

Median, IQR and histograms for 90 “good” fills that made it to the STABLE between 26.07 and 10.10

44 of 54

B2 hardware swap

44

Based on 53 fills

Based on 37 fills

45 of 54

What influences Schottky estimate

45

  • Geometric emittance & intensity
    • Total power ∝ emittance x intensity —> witness bunches perfect, probes failing
  • Rapid changes
    • Spectra are averaged for ~120 seconds, during the change the spectra are mixture of two
  • Presence of 50 Hz harmonics
    • Mostly in B1V and B2H, since late RAMP to early STABLE
  • Coupling
    • Particularly prominent effect at SQUEEZE & ADJUST
  • Particle charge
    • Total power ∝ particle charge squared → O and Ne this year will be very interesting.

46 of 54

Hardware imprints

46

  • Glitches during the bunch change
    • Affects the estimate for 2 mins (buffer length)
    • Easy to filter out

47 of 54

Hardware imprints

47

  • Glitches during the bunch change
    • Affects the estimate for 2 mins (buffer length)
    • Easy to filter out
  • Q’ jumps during longitudinal blow-up
    • Saturation, extra gating added in the last YETS

48 of 54

Hardware imprints

48

  • Glitches during the bunch change
    • Affects the estimate for 2 mins (buffer length)
    • Easy to filter out
  • Q’ jumps during longitudinal blow-up
    • Saturation, extra gating added in the last YETS
  • Simultaneous glitches in H&V plane
    • Most likely LO problem

49 of 54

Hardware imprints

49

  • Glitches during the bunch change
    • Affects the estimate for 2 mins (buffer length)
    • Easy to filter out
  • Q’ jumps during longitudinal blow-up
    • Saturation, extra gating added in the last YETS
  • Simultaneous glitches in H&V plane
    • Most likely LO problem
  • Gain variations
    • Slow ones: due to amplifier gain drifts
    • Fast ones: under investigation, might be LOs
      • On 10 Oct B1 gain dropped by over 90%

Fill 10226

50 of 54

Hardware imprints

50

  • Glitches during the bunch change
    • Affects the estimate for 2 mins (buffer length)
    • Easy to filter out
  • Q’ jumps during longitudinal blow-up
    • Saturation, extra gating added in the last YETS
  • Simultaneous glitches in H&V plane
    • Most likely LO problem
  • Gain variations
    • Slow ones: due to amplifier gain drifts
    • Fast ones: under investigation, might be LOs
      • On 10 Oct B1 gain dropped by over 90%

Fill 10226

LOs power is monitored continuously since last YETS, with a dedicated UCAP converter to report fast variations

51 of 54

Relative transverse emittance measurements

The emittance is mostly determined by the noisy spectrum centers.

As the result, we cannot rely on the fit of the filtered spectrum.

For clean spectra (Pb82+), plausible emittance estimates can be obtained even from instantaneous 1-second spectra.

During the MD 14343 the Schottky-based emittances were compared to WS. Despite good correlation, the absolute values deviated:

  • WS assumption of Gaussian profile?
  • Schottky power non-linearities?

Study requires more work, but is interesting, as Schottky can measure the ion emittance at injection.

51

MD 14343

52 of 54

ML Schottky analysis

SchMatrix package makes it possible to efficiently generate a big number of artificial Schottky spectra (in this moment we have 2.7 TB of artificial Schottky spectra).

These can serve as a dataset for machine-learning studies, aiming at:

  • Directly derive the beam and machine parameters from Schottky spectra,
  • Denoise the spectra and prepare them for “traditional analysis”,
  • Detect abnormalities and signal them to the user/operator

The pioneering work was done by M. Bradicic. During his Summer Student project he trained an autoencoder model obtaining good results in predicting chromaticity and betatron tune.

CERN-STUDENTS-Note-2024-224

52

LHC Fill 9996, STABLE

53 of 54

2025 analysis code upgrade

Original analysis code was a bricolage of various corrections and improvements.

This year’s upgrade had a few goals:

  • Improve the readability and maintainability of the code,
  • Migrate all the analysis from FESA to UCAP,
  • Facilitate analysis testing and benchmarking,
  • Introduce fast prediction functionality.

The analysis is now split into 8 distinct UCAP converters. Each converter comes with tests, including

  • Unit tests,
  • SchMatrix-based tests,
  • E2E nxcals data tests (e.g. Analyze fill XXX)

53

Schottky Analysis

Stats

Schottky Preprocess

Schottky Analysis Correlation

Schottky Analysis

Tune Fast

Schottky Analysis Emittance Fast

Schottky Analysis

Tune

Schottky Analysis Emittance

Schottky Analysis

Chroma

FESA

54 of 54

Conclusion

Last three years witnessed a significant progress in the Schottky signal analysis.

Theory:

  • Development and upgrade of spectra simulation techniques,
  • Understanding the effects of
    • octupole magnets,
    • longitudinal impedance,
    • higher-order chromaticity and non-sinusoidal RF systems,
  • We learned how to profit from signal statistical properties and plane-plane correlations.

Diagnostics:

  • We implemented an on-line system estimating Q’ for proton and ion beams, at INJ and Flattop,
    • After 16 years since the start of LHC we have a non-invasive Q’ diagnostics!
    • As a sanity-check provides also tune and dp/p.

Promising studies have been started on:

  • Schottky-based measurements of longitudinal impedance, emittance and coupling constant,
  • ML-based analysis of Schottky spectra.

54