1 / 19

BPM Resolution Studies at PETRA III

DEELS 2019. ESRF, 3. -5. June 2019. BPM Resolution Studies at PETRA III. Gero Kube DESY (MDI). Introduction Comments on BPM Resolution „ Three BPM“ Correlation & Prinicipal Components Analysis Resolution Studies at PETRA III Conclusion. PETRA III @ DESY. PETRA history.

sherrylk
Download Presentation

BPM Resolution Studies at PETRA III

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. DEELS 2019 ESRF, 3. -5. June 2019 BPM Resolution Studies at PETRA III Gero Kube DESY (MDI) • Introduction • Comments on BPM Resolution • „Three BPM“ Correlation & Prinicipal Components Analysis • Resolution Studies at PETRA III • Conclusion

  2. PETRA III @ DESY • PETRA history • 1978 – 1986: e+e-collider(upto 23.3 GeV / beam) • 1988 – 2007: pre-acceleratorfor HERA (p @ 40 GeV, e @12 GeV) • since 2007: dedicated 3rdgeneration light source, commissionedin 2009 TDR: DESY 2004-035 • → 14 beamlines(15 experimental stations) operating in parallel • from 2014: staged extension project W. Drube et al., 2016 https://doi.org/10.1063/1.4952814 • → upto 12 additional beamlines(presently not all ofthem in operation) Extension Hall East Ada Yonath Max von Laue Hall Extension Hall North Paul P. Ewald DEELS 2019 Workshop @ ESRF, 4.6.2019 Gero Kube, DESY / MDI

  3. PETRA IV: Overview PETRA IV storage ring and pre-accelerators • useofexistingacceleratortunnel • → asymmetric ring structure • additional experimental hall • → 29 straight ID sections time line • presently • → preparation of Conceptual • Design Report DEELS 2019 Workshop @ ESRF, 4.6.2019 Gero Kube, DESY / MDI

  4. PETRA IV: Diffraction Limited Light Source DLS design BPM resolution requirements • singlebunch / single turn< 20 μm (assuming 0.5 mA in singlebunch → 2.5×1010particlesbunch ) • closedorbit< 100 nm (rms, 200 mA in 1600 bunches) at 300 Hz BW courtesy: I. Agapov (DESY) DEELS 2019 Workshop @ ESRF, 4.6.2019 Gero Kube, DESY / MDI

  5. BPM Resolution position determination in circular accelerator y position resolution (small displacements from center) x beam : monitor constant : signal-to-noise ratio • geometry (buttonsize) → signalstrength • infrastructure→ cablelength & attenuation • read-out electronics depends on goal • performancetest→ existing Libera Brilliance readout electronics @ PETRA III • pickupgeometry→ beam pipediameter • buttonsize → smallcorrection (P1+P2) – (P3+P4) (P1+P4) – (P2+P3) main focus: read-out electronics y = Ky x = Kx P1+P2+P3+P4 P1+P2+P3+P4 DEELS 2019 Workshop @ ESRF, 4.6.2019 Gero Kube, DESY / MDI

  6. BPM Resolution Measurements modern ADCs optimized for cw signals • signalfrom BPM button→ farawayfromcwsignal BPM signal measurement with beam → 2 kindsofjitter • beam jitter • → realchangeof beam angle andpositioncausedbyfluctuations in accelerator • (groundmotion, energyfluctuation, kicks, …) • → seenbyseveral / all BPMs simultaneously • (correlation via beam optics) • noiseof BPM electronics • → quantitiytobemeasured • → nocorrelationbetweenadjacent BPM readings common methods for correlation analysis • „three BPM“ correlationmethod • Principal Components Analysis(PCA) correlation analysis in order to disentangle both jitter sources BPM resolution with beam generated signals brief review DEELS 2019 Workshop @ ESRF, 4.6.2019 Gero Kube, DESY / MDI

  7. „Three BPM“ CorrelationMethod principle setup: 3 adjacent BPMs connection via transport matrices • no non-linear elementsbetween BPMs BPM1 BPM2 BPM3 beam BPM reads position information difference: measured position vs. expectation calculate variance (error propagation) all BPM readings with same error σΔ → N consecutivepositionmeasurements DEELS 2019 Workshop @ ESRF, 3.- 5.6.2019 Gero Kube, DESY / MDI

  8. „Three BPM“ CorrelationMethod (2) procedure • N consecutivepositionmeasurementswith 3 adjacent BPMs • determinationoftransfermatrixelementsX21, X23 • → straightforward:calculationaccordingto beam optics • → modelindependent:Moore-Penrose pseudo inverse • (least-squareestimateforX) • BPM resolution • → evaluateformula constant offset ≈ 0 • successiveapplicationto all BPMs • → groupingthreeadjacent BPMs • example: KEK-BM. Arinaga et al., NIM A499 (2003) 100 restrictions • no non-linear elements→ difficult at DLS • same errorof BPM readings→ not possible at PETRA III • sometimesweakcorrelationswithneighbour BPMs • → large uncertainty in BPM resolution (especiallywith Moore-Penrose) phase advance ≠ n/2 π n = ±1, 3, 5, … DEELS 2019 Workshop @ ESRF, 4.6.2019 Gero Kube, DESY / MDI

  9. Principal Components Analysis (PCA) method of multivariate statistics • conversion of set of correlated variables into set of linearly uncorrelated ones • → principalcomponents (PC) • cleansingofcorrelations in datasets • → structuringof large datasets, compression, … prinicipal axis determination • orientation of firstprincipalaxis • → axisrotation such thatoveralldatavarianceismaximized (requirescenteringofdata) → alternative: minimizeprojections (hint: χ2) • orientation of second (third, …) principalaxis • → remove contribution of 1st PC from data • → repeat rotation and variance maximization • condition: uncorrelated with (i.e., perpendicular to) first principal component mathematics behind • eigen vectors → principalcomponents • eigen values (Λ)→ amountofvariancefor PC • form covariance matrix C→ real & symmetricmatrix • diagonalization of C→ C = V Λ VT • V :formedbyorthonormal eigen vectors • sort eigenvectors according to eigenvalues DEELS 2019 Workshop @ ESRF, 4.6.2019 Gero Kube, DESY / MDI

  10. PrincipalComponents Analysis (2) alternative numerical method → Singular Value Decomposition (SVD) orbit (spacecoordinate) • instead of diagonalization of covariance matrix • → SVDofdatamatrixM • relationbetweensingularandeigen values: • → turns (timecoordinate) (M real: V* → VT) courtesy: Wikipedia • advantage • → SVD numericallymorestable (formation of MMTcan cause loss of precision → Läuchli matrix) • → benefit: SVD provides additional information (acceleratorphysics) application to BPM data • construction of BPM matrixM • → BPM datacentered • → normalization: C.X. Wang, SLAC-R-547 (2003) • exploration of SVD matrix properties • → U:columnvectorscontaininformationabouttemporalpattern (tune, …) • → V:columnvectorscontaininformationaboutspatialpattern (orbit / βfunction, …) • comment: U/Vas temporal/spatialvectors → depends on orientationofmatrixM DEELS 2019 Workshop @ ESRF, 4.6.2019 Gero Kube, DESY / MDI

  11. PCA Example test measurement @ PETRA III SVD analysis • fillpattern: 960 bunches @ 5.6 mA • singlevertical kick withexcitationkicker • 2048 turnsrecorded • dominant modes (singularvalues) → mode 1 & 2 temporal modes • informationabout tune • → FFT ofu1, u2 spatial modes • informationabout beam optics→ BPM resolution • singularvaluesσ1, σ2, …= 0 → calculatecleanedorbitdata DEELS 2019 Workshop @ ESRF, 4.6.2019 Gero Kube, DESY / MDI

  12. PETRA III BPM System 11 different pickuptypes: 3 different cable types • Max von Laue hall RFA ½̋, ⅜̋ , ⅞̋ – 50 Ω cable lengths: 10m … 200m 246 individual settings • standardoctants: normalizetoKx,y = 10 mm RMSx,y = f(signal power) I. Krouptchenkov et al., Proc. DIPAC 2009, Basel, TUPD03, p. 291 DEELS 2019 Workshop @ ESRF, 4.6.2019 Gero Kube, DESY / MDI

  13. BPM Resolution Studies resolution studies @ PETRA III • singlebunchwithQbvarying • singlevertical kick withexcitationkicker • 2048 turns Kx,y normalization PCA for beam jitter removement combinationofmeasurementsforvariousQb plotasfunction ofsignal power DEELS 2019 Workshop @ ESRF, 4.6.2019 Gero Kube, DESY / MDI

  14. TbT Single Bunch Resolution machine studies at PETRA III with existing BPM system • LiberaBrilliance (Instrumentation Technologies) min(RMSx,y) ≈ 30 μm LiberaBrillianceresolution not sufficientfor PETRA IV PETRA IV CDR: rely on successor model → Libera Brilliance+ DEELS 2019 Workshop @ ESRF, 4.6.2019 Gero Kube, DESY / MDI

  15. Test of Read-out Electronics beam test @ PETRA III: december 2018 • LiberaBrilliance • → BPM_SOL_24, in use at PETRA III • LiberaBrilliance+ • → CDR: PETRA IV system • → onesystemboughtfor DORIS / Olympus • → onesystemrecentlybought • Libera Spark • → newplatform, nolong-term stabilization • → borrowedfrom I-Tech (thanksto Peter Leban) Combiner / Splitter Combiner / Splitter Combiner / Splitter Combiner / Splitter Signal Combiner/Splitter MACOM DS-409-4 LiberaBrilliance+ (old) LiberaBrilliance LiberaBrilliance+ Libera Spark Signal Combiner/Splitter MACOM DS-409-4 ~ 1 dB Attenuation BPM TbT resolution determination • orbitdatacontaincontributions due to • → correlated beam jitter • → noiseof BPM electronics 10 dB Attenuator disentangle contributions BPM_SOL_24 • correlationanalysis→ does not workforsingle BPM • eliminatecorrelatedjitter→ sum & splitorbitdata DEELS 2019 Workshop @ ESRF, 4.6.2019 Gero Kube, DESY / MDI

  16. Resolution Comparison Libera Brilliance+ • digital down conversion (DDC) • σ < 10 μm neverachieved • σ = 19.8 μm @ IB = 1.549 mA (-29 dB) • → correctforattenuation: IB= 0.38 mA (≈ -42 dB) • time domainprocessing (TDP) • σ = 10.3 μm @ IB = 1.323 mA (-30 dB) • → correctforattenuation: IB= 0.38 mA wouldfulfilrequirements → but Libera Spark isbetter… DDC mode < 20 μm (rms) → IB ≈ 0.4 mA TDP mode≈ 10 μm (rms) → IB ≈ 0.4 mA DEELS 2019 Workshop @ ESRF, 4.6.2019 Gero Kube, DESY / MDI

  17. Closed Orbit & Stability long term stability • useroperation: 480 bunches in 100 mA, top-up • all Liberas in closedorbit (SA) mode→ driftcompensation (digital signalconditioning, DSC) on Spark: no DSC available Brilliance+ without DSC good climatization (hutches, racks) mandatory DEELS 2019 Workshop @ ESRF, 4.6.2019 Gero Kube, DESY / MDI

  18. Closed Orbit Resolution specification: < 100nm (rms) in brightnessmode(200mA in 1600 bunches) @ BW 300 Hz • horizontal orbit first 12 hours (before beam dump): σrms = 20.76 nmin SA mode (BW 4Hz, Kx,y = 10mm; BW 4Hz → seeBrilliance+ User Manual, p.34) → scalingwith band width: × sqrt(300/4) = 8.66 → σrms = 180 nm (@ BW 300Hz) DEELS 2019 Workshop @ ESRF, 4.6.2019 Gero Kube, DESY / MDI

  19. Conclusion BPM resolution studies • require correlation analysis in order to disentangle noise and beam generated jitter • → „three BPM“ correlationmethod & PCA correlation analysis • „three BPM“ correlationmethod→ not suitablefor PETRA III BPM system • PCA → powerful tool, not onlyforresolutionstudies (e.g. BPM performanceevaluation @ SSRF, …) • → modelindependent, but limited bymodemixing Z.-C. Chen et al., Chinese Physics 38 (2014) 077004 Nucl. Sci and Tech. 25 (2014) 020102 • closedorbit→ specsnot fulfilled (< 100 nm @ 300 Hz bandwidth) σy,rms ≈ 180 nm @ 300 Hz bandwidth andKx,y= 10 mm Libera Brilliance • singlebunchresolution→ specs not fulfilledfor PETRA IV Libera Brilliance+ • singlebunchresolution→ specsfulfilled @ bunchcurent IB ≈ 0.4 mA andmonitorconstantKx,y = 10 mm DDC mode< 20 μm (rms), TDP mode≈ 10 μm (rms) Libera Spark • singlebunchresolutionmuchbetter→ closedorbit: needsstabilization next step: Independent Component Analysis (ICA) DEELS 2019 Workshop @ ESRF, 4.6.2019 Gero Kube, DESY / MDI

More Related