Transverse instability observation network in LHC

Transverse instability observation network in LHC

1


A. Butterworth, W. Höfle, G. Kotzian, D. Valúch (BE/RF)T. Levens (BE/RF BE/BI)

T. Lefèvre, R. Steinhagen (BE/BI)J. Serrano, T. Włostowski (BE/CO)

LBOC, 11.3.2014


2

Motivation

• Review on “Functional Requirements on LHC Transverse Instability Diagnostics after LS1” chaired by W. Höfle and R.Steinhagen, 15 March 2013

• Scope and Aims of the Review:A. collect user requirementsB. present capabilities of existing systems and options for upgradesC. ensure technology and infrastructure in place to optimally use

available instruments (triggers, logging, software)D. identify missing technologies and systems that need further

developmentE. prioritize requests (feasibility, resources, impact)F. identify requests that cannot be fulfilled for the start-up after LS1G. recommend allocation of resources to management

LBOC, 11.3.2014


3

A. User requirements (from the Review)

• Simple information on mode (m=0 versus m≠0). Instantaneous bunch-by-bunch information can be crucial to take fast decisions for short-term ‘cures’• ‘Light’ changes e.g. Q, Q’, octupoles, ADT gain, bunch length, filling

schemes, intensity/brightness

• Through the whole cycle:• Instantaneous bunch-by-bunch tunes (10-4 resolution) at an update

rate of 1 Hz• Accurate knowledge of chromaticity with a precision of about 2 units at

an update rate of about 0.05 Hz.

LBOC, 11.3.2014


4

A. User requirements (from the Review)

• In an interval of time of ±3 s around the moment of the instability:• Continuous “turn-by-turn” “bunch-by-bunch” positions during the

instability with a resolution of a few mm• Intra-bunch time domain signal acquisition with an analogue

bandwidth of about 6 GHz, 20 Gs/s sampling and transverse resolution of few mm.

• Continuous bunch-by-bunch intensity at the highest rate (10 Hz)• Continuous bunch-by-bunch transverse emittance with fast scan (is a

5 Hz sequential acquisition feasible?).

• Provide a snapshot of the machine when the instability occurs. This information should be stored permanently in the logging DB

LBOC, 11.3.2014


5

B. Present capabilities of existing systems and options for upgrades• LHC transverse damper (ADT), run 1

• Two pick-ups per beam per plane at point 4• ADT saw only symmetric oscillation patterns

• Coherent oscillations are being damped

• Available information - bunch position for all bunches at full 40 MHz rate, resolution ~2 mm

• Available buffer length 256k bunch-turns (slow trigger, not too frequent data extraction ~1 buffer per 10 seconds)

• 73 turns, all bunches• 262144 turns 1 bunch, 131072 turns 2 bunches, 65536 turns 4 bunches, 32768

turns 8 bunches

• Post mortem data for the last 73 turns, beam-out trigger

LBOC, 11.3.2014


6

D. Missing technologies and systems that need further development

LBOC, 11.3.2014

being developed during LS1

W. Hofle Transverse Feedback LS1 changes LMC - 5 Feb

2014 LMC - 5 Feb

2014

7

Pickup Q7

Beam position module

Q7

Pickup Q9


Q9

Analogue output to the power amplifiers

Digital signal processing unit

DAC

Gain control (CCC)

CLEANING

Sequential observation

(Multiturn app.)

Pickup Q7


Q7

Pickup Q9


Q9 Analogue output to the power amplifiers

Digital signal processing unit

DAC

Gain control (CCC)

CLEANING

DAC

Pickup Qx


Qx

Pickup Qx


Qx

DACBeam transfer function meas.

Observation box

Tune/instability

diagnostics box

Fast b-by-b Instability

diagnostics

CCC, users, logging

Reasons for change: “DSPU”

• need to combine four pick-ups• Separate control for all features

with high resolution calls for independent output DACs

• digital links to future diagnostics HW

• adequate number of spares missing

needs separate discussionvision@end of Run1


8

D. Missing technologies being developed: ADT

• Post LS1 ADT• Four pick-ups per beam, per plane, located at point 4• Available information: bunch position for all bunches at full 40 MHz rate.

Resolution (equal or better than) 2 mm

• Internal buffer length increased 16x to 4M bunch-turns i.e.• 1 bunch for 4 194 304 turns• or all bunches for 1100 turns

• Fast on-the-fly data analysis, can detect intra-bunch, symmetric oscillation patterns, bunch-by-bunch, within few turns and generate a trigger

• Alternative data processing algorithm can also give hints on anti-symmetric oscillation patterns (see G. Kotzian et al: Sensitivity of LHC ADT to intra-bunch motion, BE-ABP-HSC meeting, 22.1. 2014) ...and generate a trigger

LBOC, 11.3.2014


9

Post LS1 ADT

LBOC, 11.3.2014

Normalized transverseposition

Transverse oscillation

pattern

Movement of centre-of-charge

Longitudinal profile symmetricasymmetric

even symmetricodd symmetric

Stri

plin

e pi

ckup

Q9

I

Q

ADC

Nor

mal

ized

pos

ition

ca

lcul

atio

n

Beam Position module

COMBFILTER

A

B

180°HYBRID

beamADC

I

Q

ADCCOMBFILTER

ADCR

aw e

lect

rode

si

gnal

s

Pos

ition

and

in

tens

ity s

igna

ls

Ban

dwid

th

limita

tion

Gai

n co

ntro

l

Dow

n co

nver

sion

Dig

itiza

tion

Nor

mal

ized

bun

ch

by b

unch

pos

ition

ca

lcul

atio

n


10

Post LS1 ADT

LBOC, 11.3.2014

I

Q

I’

Q’

I’Q’

0 0.5 1 1.5 2 2.5 3-0.2

0

0.2

0.4

0.6

0.8

1

1.2

Frequency / GHz

Det

ecte

d A

mpl

itude

[Nor

mal

ized

]/ m

m

xN(fx)

xbar(fx)

Movement of centre-of-charges

Normalized position as currently implemented

in the BeamPos HW

• Damper sensitivity to symmetric intra-bunch motion is a function of the longitudinal beam spectra

• The current normalization scheme sees only symmetric (even) oscillation patterns (needed for closed loop feedback)

• For the anti-symmetric case no oscillation amplitude is detected, odd modes not visible to the damper


11

Post LS1 ADT

LBOC, 11.3.2014

• This algorithm can detect odd-mode oscillations• None of the algorithms can resolve the original oscillation frequency, and

absolute oscillation amplitude (except m=0)

• But they can detect activity and distinguish between symmetric (even) and asymmetric (odd) modes of every bunch trigger!

• Logging of a bunch-by-bunch snapshot for 1000 turns at the moment of trigger, with information which bunches were unstable

Alternate processing scheme to detect and indicate anti-symmetric oscillations:

0 1 2 3 4 5-0.1

0

0.1

0.2

0.3

0.4

0.5

0.6

Frequency / GHz

Det

ecte

d A

mpl

itude

[Nor

mal

ized

]/ m

m

xN R (fx)

xbar(fx)


12

D. Missing technologies being developed: MIM

• Upgrade of the “head-tail” instrument• Classic “brute force” time-domain acquisition: next generation digitizer

upgrade, 6 GHz analogue bandwidth, 20 GS/s sampling, 32 GB/channel sampling buffer (>1 s of beam data). Massive on the fly data analysis.

• Multiband-Instability-Monitor (MIM), three different acquisition options:• Balanced Schottky Diode Detector. This could provide nm-level resolution and

could be used as an early instability trigger, but would not allow to distinguish on which bunch the instability occurs planned to be deployed after LS1

• Bunch-by-Bunch Balanced Schottky Diode Detector. This would provide bunch-by-bunch magnitude (no-phase information for bands > 400 MHz) data with sub-μm resolution. Needs ADC-DAQ and SW integration

• Direct-Down-Conversion Receiver providing full amplitude & phase information (identical info to time-domain digitizer). Needs ADC-DAQ and SW integration

LBOC, 11.3.2014

in R&D phase → possible deployment LS2 (?)


13

Multiband Instability Monitor

• MIM principle – instead of “brute force” sampling, look at discrete frequency bands (harmonics of 400MHz)

LBOC, 11.3.2014


14


LBOC, 11.3.2014

Instability trigger


15


• Minimum deliverable system (i.e. based on what has been prototyped/ demonstrated in 2012/13):• 6 analog-front-ends (ΔH, ΔV,ΣH/V x B1/2; 16 frequency bands spread between 0.4 - 6

GHz• Will equip only 4 channels/front-end with digital-front-ends (tbc.), initially same DAB

as for Diode Orbit/Beta-Beat System• Specific channels can be changed if necessary (e.g. during technical stop)• Simple triggering scheme, for details see BI Seminar 23rd Aug. 2013 &

CERN-STUDENTS-Note-2013-103

• Observables• Provide an early/high-sensitivity trigger fed into the White-Rabbit Trigger Network• Distinguish whether it was rigid or intra-bunch related motion. No bunch information.• Aim at getting data logged:

• continuously (e.g. eigenmode/band amplitude vs. time)• Snapshot of the raw data of all bands during the triggered instability

• Help needed (OP & ABP): GUI integration, perhaps online display of amplitudes vs. time?

LBOC, 11.3.2014

https://espace.cern.ch/be-dep/BI/BI%20Seminar%20Presentations/BI%20seminars/Trigger%20Development%20for%20the%20Multi-band%20Instability%20Monitor%20-%2023Aug13%20-%20%20J.Ellis%20%5BUniv.%20of%20Melbourne%5D.pdf

https://espace.cern.ch/be-dep/BI/BI%20Seminar%20Presentations/BI%20seminars/Trigger%20Development%20for%20the%20Multi-band%20Instability%20Monitor%20-%2023Aug13%20-%20%20J.Ellis%20%5BUniv.%20of%20Melbourne%5D.pdf

http://cds.cern.ch/record/1595013/files/Report.pdf


16

D. Missing technologies being developed

• ADT “observation box”• Bunch-by-bunch data, with all precious beam information, are

available within the ADT signal processing blocks• Difficult to extract, or store due to the very high data rates• Difficult to do sophisticated data analysis directly in the ADT FPGAs

• Ambitious project proposed by the ADT team: stream the full rate data to an external computer(s):• Allows massive number crunching e.g. for tune measurement• Continuous, online analysis of the transverse (and longitudinal)

motion, all kinds of fixed displays…• Storage of a full 40 MHz, bunch-by-bunch data for an entire fill for

offline analysis• ……..and generate/receive a trigger

LBOC, 11.3.2014


17

ADT “observation box”

• First results with data transmission and reception T. Levens• Data processing by graphic chips F. Dubouchet (see.

Betatron tune measurement with the LHC damper using a GPU, CERN-THESIS-2013-035)

• Full implementation by BE/RF/CS section

LBOC, 11.3.2014

ADT Beam position module

Observation box receiver (PC)

Bunch by bunch dataread out via PCIe

Processed data (e.g. tune extraction)

20ns !

bunch #0+frevSPEC board

GPU

http://cds.cern.ch/record/1545785/files/CERN-THESIS-2013-035.pdf


18

C. Technology and infrastructure to optimally use available instruments (triggers, logging, software)• Some instruments can (or will be able to) analyze the signals and

generate a trigger

• Most instruments in the machine have buffers with very valuable data• Very different data rates, very different record lengths• …just if we were able to freeze and use them!

• We needA. Fast, deterministic, configurable, machine-wide trigger distribution

network (multiple inputs, multiple outputs)B. To collect, store and analyze all those data.

LBOC, 11.3.2014


19

Instability trigger network functionality

• A set of observation instruments can be located anywhere in the LHC or CCC

LBOC, 11.3.2014

MIMH.B1

MIMV.B1

MIMH.B2

MIMV.B2

ADTH.B1

ADTV.B1

ADTH.B2

ADTV.B2

OBS BOXB1

OBS BOXB2

APWB1

APWB2

ACSB1

ACSB2

BLMs

BPMs

???

???

???

Trigger network& routing


20



• Many of them can generate a trigger

LBOC, 11.3.2014

MIMH.B1

MIMV.B1

MIMH.B2

MIMV.B2

ADTH.B1

ADTV.B1

ADTH.B2

ADTV.B2

OBS BOXB1

OBS BOXB2

APWB1

APWB2

ACSB1

ACSB2

BLMs

BPMs

???

???

???



21



• Many of them can generate a trigger

• Many of them can receive trigger and freeze a data buffer

LBOC, 11.3.2014

MIMH.B1

MIMV.B1

MIMH.B2

MIMV.B2

ADTH.B1

ADTV.B1

ADTH.B2

ADTV.B2

OBS BOXB1

OBS BOXB2

APWB1

APWB2

ACSB1

ACSB2

BLMs

BPMs

???

???

???



22


• Trigger propagation path has to be defined

LBOC, 11.3.2014

MIMH.B1

MIMV.B1

MIMH.B2

MIMV.B2

ADTH.B1

ADTV.B1

ADTH.B2

ADTV.B2

OBS BOXB1

OBS BOXB2

APWB1

APWB2

ACSB1

ACSB2

BLMs

BPMs

???

???

???



23



• E.g. ADT, horizontal, B2 detected an activity

LBOC, 11.3.2014

MIMH.B1

MIMV.B1

MIMH.B2

MIMV.B2

ADTH.B1

ADTV.B1

ADTH.B2

ADTV.B2

OBS BOXB1

OBS BOXB2

APWB1

APWB2

ACSB1

ACSB2

BLMs

BPMs

???

???

???



24



• E.g. ADT, horizontal, B2 detected an activity

• System should freeze only relevant devices

LBOC, 11.3.2014

MIMH.B1

MIMV.B1

MIMH.B2

MIMV.B2

ADTH.B1

ADTV.B1

ADTH.B2

ADTV.B2

OBS BOXB1

OBS BOXB2

APWB1

APWB2

ACSB1

ACSB2

BLMs

BPMs

???

???

???



25


• Sequence of triggers has to be time-stamped and each trigger originator identified

• Indicate to CCC that the data needs to be collected

LBOC, 11.3.2014

MIMH.B1

MIMV.B1

MIMH.B2

MIMV.B2

ADTH.B1

ADTV.B1

ADTH.B2

ADTV.B2

OBS BOXB1

OBS BOXB2

APWB1

APWB2

ACSB1

ACSB2

BLMs

BPMs

???

???

???



26


• Sequence of triggers has to be time-stamped and each trigger originator identified

• Indicate to CCC that the data needs to be collected

• Some instruments need the revolution frequency and beam synchronous 40 MHz signals

LBOC, 11.3.2014

MIMH.B1

MIMV.B1

MIMH.B2

MIMV.B2

ADTH.B1

ADTV.B1

ADTH.B2

ADTV.B2

OBS BOXB1

OBS BOXB2

APWB1

APWB2

ACSB1

ACSB2

BLMs

BPMs

???

???

???



27

C. Technology and infrastructure to optimally use available instruments (triggers, logging, software)

LBOC, 11.3.2014

• White rabbit network• Solution proposed and technology provided by

the BE/CO group

• During LS1 BE/RF and BE/BI invested into the White rabbit infrastructure connecting the ADT, Head Tail, future MIM, and other future instruments (e.g. diamond BLMs)

• Functional specification being completed (March 2014)

• Equipment installation (Summer 2014)


28

C. Technology and infrastructure to optimally use available instruments (triggers, logging, software)

LBOC, 11.3.2014

• Software development to provide starting functionality by BE/CO critical (will be covered by T. Włostowski)• Configuration of the inputs/outputs, delays• Trigger propagation routing, re-triggering etc.

• Collection of data from all instruments after a trigger (??/??)

• Data storage (??/??)• measurement database, logging, special database?

• CCC fixed displays (??/??)

• Online/Offline data analysis (??/??)


29

Thank you for your attention.

LBOC, 11.3.2014

Questions?

Transverse instability observation network in LHC

Documents

bunch information

bunch length

instantaneous bunch

bunch intensity

hzcontinuous bunch

bunch tunes

bunch positions

bunch transverse emittance