1 / 81

The TTC system and Jitter in LHC experiments

The TTC system and Jitter in LHC experiments. PH-ESE seminar 18 December 2012. Sophie Baron. OUTLINE. Bunch Clock Origin From the Radio Frequency to the Bunch Clock The TTC system What is a good Bunch Clock ? Vocabulary – jitter , phase noise etc..

pepin
Download Presentation

The TTC system and Jitter in LHC experiments

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. The TTC system and Jitter in LHC experiments PH-ESE seminar 18 December 2012 Sophie Baron

  2. OUTLINE • BunchClockOrigin • From the Radio Frequency to the BunchClock • The TTC system • Whatis a good BunchClock? • Vocabulary – jitter, phase noise etc.. • Whois sensitive to what? • A good BunchClock in twowords • Measuring the clockquality • BunchClockmeasurement for the detectors • BunchClockmeasurement for the sub-systems • Conclusion • Lessonslearned PH-ESE seminar - 18 Dec 2012 - Sophie Baron

  3. The Radio Frequency (RF) From Radio Frequency to Bunch Clock The Bunch Clock Clients The TTC system Bunch CLOCK ORIGIN PH-ESE seminar - 18 Dec 2012 - Sophie Baron

  4. RF cavities in LHC (4 modules@point4, Echenevex) v v v v PH-ESE seminar - 18 Dec 2012 - Sophie Baron

  5. The Radio Frequencyis not always the same protons • The Radio Frequency is not fixed: • it is a function of particle type and energy • It is ramping up at the beginning of each fill • it is modulated by beam characteristics and RF parameters • It is however extremely stable during flat top. Pb ions • 400.788860 MHz -> 400.789715 MHz (p in ring 1) • 400.784216 MHz -> 400.789639 MHz (Pb in ring 2) PH-ESE seminar - 18 Dec 2012 - Sophie Baron

  6. Beamparameter Frequency program Synchro loop rephasing LowLevel loops processor Beam Radial Position VCXO 400MHz ÷10 Beam Phase (Bunch/RF Phase and Vt/RF Phase) 40MHz ÷3564 RF-Tx 11kHz 400MHz 11kHz Cavities Controller Beam monitoring system PH-ESE seminar - 18 Dec 2012 - Sophie Baron

  7. LHC RF 400MHz Buckets 1 2 3 4 5 6 7 8 9 10 11 12 13 Bunch Beam LHC Bunch Clock 40MHz • The Bunch Clock is the frequency at which an observer sitting close to the ring could ‘see’ particles passing PH-ESE seminar - 18 Dec 2012 - Sophie Baron

  8. Simplistic case: 2 bunches, 2 beams, 1 observer PH-ESE seminar - 18 Dec 2012 - Sophie Baron

  9. Simplistic case: 2 bunches, 2 beams, 1 experiment, 1 fifo PH-ESE seminar - 18 Dec 2012 - Sophie Baron

  10. In detectors, everything is happening synchronously to the Bunch Clock: • Collisions • Signal sampling for Analogue to Digital conversion • Time measurement • Trigger transmission • Data storage • Data reduction • Data transmission => The Bunch Clock has to be delivered EVERYWHERE, ANYTIME, and with a excellent QUALITY => This is one of the mandates of the TTC, and this is on what we will focus today PH-ESE seminar - 18 Dec 2012 - Sophie Baron

  11. The Radio Frequency (RF) From the Radio Frequency to the Bunch Clock The Bunch Clock Clients The TTC system Bunch CLOCK ORIGIN PH-ESE seminar - 18 Dec 2012 - Sophie Baron

  12. P5 P4 P2 CCR P8 TTC backbone TTC off-detector TTC on-detector PH-ESE seminar - 18 Dec 2012 - Sophie Baron

  13. P5 P4 RF_Tx RF_Rx RF_Tx P2 CCR P8 TTC backbone RF_Rx RF2TTC TTCFanout TTC off-detector TTC on-detector PH-ESE seminar - 18 Dec 2012 - Sophie Baron

  14. TTCex RF_Rx RF2TTC TTC Fanout TTCrx TTCrq QPLL TTCoc CTP LTP TTCvi PH-ESE seminar - 18 Dec 2012 - Sophie Baron

  15. Vocabulary: Jitter & Co Who is sensitive to what? A good clock in 2 words WHAT IS A GOOD Bunch clock signal? PH-ESE seminar - 18 Dec 2012 - Sophie Baron

  16. Time Domain measurements • Jitter types • Cycle to cycle jitter (cy2cy) • Periodjitter • Time IntervalErrorjitter • Skewjitter • Representation • Frequency Domain representation • Spectrum • Phase noise • Time and Frequencydomainrelationships • Jitterdecomposition PH-ESE seminar - 18 Dec 2012 - Sophie Baron

  17. Time Domain measurements • Cycle-to-cyclejitter: • Short term variation in the clockperiodbetween adjacent clock cycles. • Contains the highestfrequency components of jitter. PiPi+1 Pi+2 Periodjitter: • Shortterm variation in the clockperiod over all measuredclock cycles, compared to the averageclockperiod. • Containsrelativelyhighfrequency components of jitter. • Do not mix withPeriodicjitter ΔPi=Pi-Pi+1 ΔPi+1= Pi+1-Pi+2 PH-ESE seminar - 18 Dec 2012 - Sophie Baron

  18. Time Domain measurements • Skewjitter: Phase errorbetween the referenceclock and the measuredclock over all clockperiods. TIE(n)=T(n)-nT0 TIE(n+1)=T(n+1)-(n+1)T0 TIE jitter (Time IntervalError or accumulated/phase Jitter): • Actualdeviationfrom the idealclockperiod over all clockperiods. • Includesjitterat all modulation frequencies. • Analysisof itsProbabilityDensityFunction (PDF) givessubstantial information on the jitter sources. S(n+1)=T(n+1)-T0(n+1) S(n-1)=T(n-1)-T0(n-1) S(n+2)=T(n+2)-T0(n+2) S(n)=T(n)-T0(n) Wander: very slow variations < 10Hz. PH-ESE seminar - 18 Dec 2012 - Sophie Baron

  19. σ= stddeviation = rmsjitter pkpkjitter PH-ESE seminar - 18 Dec 2012 - Sophie Baron

  20. Frequencydomain: Spectrum and Phase Noise Ideal Oscillator : videal(t) = v0 sin 2π ƒct Real Oscillator : vreal(t) = (v0 +Δv(t)) sin(2π ƒct + ϕ(t)) Spectrum=frequency spectral density Phase Noise=phase spectral density Frequency domain of the signal Frequency domain of the phase noise S(ƒ) =|F[vreal(t)]|2 = |F[ (v0 + Δv(t))sin(2πƒt + ϕ(t))]|2 Sϕ(ƒϕ) =|F[ϕ(t)]|2= |ϕ(ƒϕ ~ )|2 PH-ESE seminar - 18 Dec 2012 - Sophie Baron

  21. Time and Frequency Domain relationships • Getting RMS jitter out of phase noise plot (very close to TIE) [dBc/Hz] Phase Noise plot - discrete Offset Frequency (Hz) () () PH-ESE seminar - 18 Dec 2012 - Sophie Baron

  22. Cycle to Cycle, Period and TIE jitter in frequencydomain PH-ESE seminar - 18 Dec 2012 - Sophie Baron

  23. Wander: easy to visualize in frequencydomain PH-ESE seminar - 18 Dec 2012 - Sophie Baron

  24. JITTER Gaussian and unbounded PDF, quantityisoften « rms » Usuallyperiodic or narrowband, the PDF isbounded, quantityisoften « pkpk » Random Jitter (RJ) Deterministic Jitter (DJ) Periodic Jitter (PJ) Data Dependent Jitter (DDJ) Not discussed in this talk - Impossible to detect on a simple clock signal. (Onlyvalid for clockrecovery out of a serial data link. Onlydetected by time domainanalysis) PH-ESE seminar - 18 Dec 2012 - Sophie Baron

  25. Randomjitter sources • Caused by accumulation of a hugenumber of uncorrelatedprocessesthat have small magnitude • Random noise phenomena • Thermal noise, Shot noise, Pink noise, etc… • Occur in all semiconductors and components (PLLs, Oscillators, Tx, Rx etc…) • Typicalrepresentation of inducedjitter: Gaussian& unbounded PDF • Quantifiedby the Standard Deviation (rms) • Deterministicjitter sources • Causedby a comparativelysmallnumber of processesthatcanbecorrelatedand may have large amplitudes, • System & Data Dependentphenomena • Crosstalk, dispersion, impedancemismatch • Inter Symbol Interference (ISI), Duty-Cycle Distortion (DCD), Bit sequenceperiodicity • Typicallydetected as deviationof the PDF fromGaussianshape • Quantifiedby the pkpk value, as theyinduce a bounded phase deviation PH-ESE seminar - 18 Dec 2012 - Sophie Baron

  26. Vocabulary: Jitter & Co Who is sensitive to what? A good clock in 2 words WHAT IS A GOOD Bunch clock signal? PH-ESE seminar - 18 Dec 2012 - Sophie Baron

  27. Digital Systems: • Very sensitive to setup and hold time • basicallyrelated to pkpk cy2cy and periodjitter. • PLLs: • Track the slow variations of the clocks, and filter out the highfrequency components. • Can not deal withsudden jumps whichmayunlockthem. • pkpk cy2cy jitter • Wandercanalsobe a problemwhenitmeansfrequencydrifting out of the locking range. • ADCs: • Very sensitive to timing errors as jitterisdirectlyconvertedinto amplitude samplingerrors, and SNR. • Unregularlysamplingedgescandistort of the shape of digitized pulses. • This isthus more about pkpk cy2cy and periodjitterthan about TIE. PH-ESE seminar - 18 Dec 2012 - Sophie Baron

  28. Serial Data Links: • Need to combine low Bit Error Rate (BER) and good ClockRecovery • BER isrelated to the quality of the clock • Transmitterisvery sensitive to anyclockjitter(because of clock multiplication). • On the channel, data jitteriscorrelated to Duty Cycle Distortionof the clock (DCD) • Receiver and CDR are highly sensitive to highfrequencyjitter • Quality of the ClockRecoveryis a trade off betweenlow BER (requireshighbandwidth PLL) and noise rejection (requiresnarrowbandwidth PLL) • Serial Data Links understandingrequiresTIEdecomposition and oftenfrequencydomainanalysis PH-ESE seminar - 18 Dec 2012 - Sophie Baron

  29. Detectors - Event reconstruction over a huge system • 1000s of BunchClock destinations spread all over the detectors • LowSKEW JITTER betweenall clocksignals(fromevery branches of the distribution tree)to guarantychannel-to-channelconsistency • Stable phase betweenBunchClock and Beam • LowSKEW JITTER betweenBunchClock and Bunches over a fill • limitedWANDERduringbroadcast, in particular on long-haul transmission (between point4 and experiments) • DeterministicStatic Phase betweenbranches and top of the clocktreefromfill to fill and between power cycles PH-ESE seminar - 18 Dec 2012 - Sophie Baron

  30. Vocabulary: Jitter & Co Who is sensitive to what? A good clock in 2 words WHAT IS A GOOD Bunch clock signal? PH-ESE seminar - 18 Dec 2012 - Sophie Baron

  31. Detector cares about Stable Phase (slow variations or skew jitter versus reference) • Sub-systems care about Low Jitter (of Bunch Clock as such) PH-ESE seminar - 18 Dec 2012 - Sophie Baron

  32. P5 P4 • Stable Phase • Beam Jitter: • Beam vs RF • Beam at experiments versus BC • Temperature drift P2 CCR P8 TTC backbone TTC off-detector TTC on-detector PH-ESE seminar - 18 Dec 2012 - Sophie Baron

  33. Stable Phase • Beam Jitter: • Beam vs RF • Beam at experiments versus BC • Temperature drift • Determinism • BC at Front End Boards versus BC at Central Trigger • Channel skew jitter • BC at Front End Board X versus BC at Front End Board Y (channel to channel skew jitter) TTCex RF_Rx RF2TTC TTcrx CTP LTP TTCvi PH-ESE seminar - 18 Dec 2012 - Sophie Baron

  34. Low Jitter • Cycle to cycle jitter • Period jitter • Overall jitter • Time Interval Error jitter • Phase Noise jitter PH-ESE seminar - 18 Dec 2012 - Sophie Baron

  35. Phase Beam Jitter Temperature Drift Determinism Channel Skew Jitter Jitter Frequency Domain Analysis Time Domain Analysis Using Jitter Information Measuring the clock quality PH-ESE seminar - 18 Dec 2012 - Sophie Baron

  36. P5 P4 P2 CCR P8 TTC backbone TTC off-detector TTC on-detector PH-ESE seminar - 18 Dec 2012 - Sophie Baron

  37. Beam jitter versus RF • phase variations of the beam with respect to its reference (400MHz RF) • Maintained and monitored by the RF low level loops • 1374 bunches • Bunch position pkpk<5ps, rms<1.3ps Beamloadingeffectreflecting the bunch structure PH-ESE seminar - 18 Dec 2012 - Sophie Baron

  38. Beam jitter versus experiment Bunch Clock • BPTX systems Use four screwed Electrostatic Button Electrodes to obtain Horizontal and Vertical position <50ps resolution by averaging First phase adjustmentjustbefore first collisions, ATLAS, nov 2009: Phase between 2 BPTX at ATLAS after fine phase with 2 beams just before collisions (90ps) Phase between 2 BPTX at ATLAS after cogging with 2 beams (900ps) Courtesy: Thilo Pauly Courtesy: Thilo Pauly PH-ESE seminar - 18 Dec 2012 - Sophie Baron

  39. Analysis of BPTX data (example of CMS) 14-12-2009 Courtesy: Jan Troska PH-ESE seminar - 18 Dec 2012 - Sophie Baron

  40. P5 P4 P2 CCR P8 TTC backbone TTC off-detector TTC on-detector PH-ESE seminar - 18 Dec 2012 - Sophie Baron

  41. Up to 14km of burriedfiberfrom SR4 to ALICE, ATLAS, LHCb (1m deep) • The fibre length changes withtemperatureby 7ppm/degC • Measurements on a sparefiberfrom CCC to ATLAS and back (9km) • 8ns seasonal drift for 14km • VERY slow variation of the phase betweenbeam and clock (wander) • Need for regular calibration (usuallybeforeeachrun) with the help of BPTX and of the CORDE module for example • Keep in mindthat : the diurnal variation canbe150 ps= expect drift duringfills D. Gigi PH-ESE seminar - 18 Dec 2012 - Sophie Baron

  42. TTCex RF_Rx RF2TTC TTcrx CTP LTP TTCvi PH-ESE seminar - 18 Dec 2012 - Sophie Baron

  43. Determinism • Typical issue with commercial PLLs 1:M CLK IN CLK OUT VCO LPF Example: N=M=3 1:N CLK IN VCO out CLK OUT 3 possible phases at startup CLK OUT’ CLK OUT’’

  44. Commercial PLL usual design GHz! VCO LPF 1:R 1xP1 1:M1 CLK IN CLK OUT1 1xP2 1:M2 CLK OUT2 1:N 1xPX 1:MX CLK OUTX

  45. Solving phase determinism issues X X X Solution 1: • Used in • TTCex (discrete PLL with 160MHz VCO) • QPLL (custom ASIC made at CERN) • Extremely RARE in commercial PLL (no commercial interest) 1:M 1:M 1:M CLK IN CLK IN CLK IN CLK OUT CLK OUT CLK OUT VCO VCO VCO LPF LPF LPF CLK OUT 1:N 1:N 1:N Solution 3: • Found in some Commercial PLLs • Beware of METASTABILITY INTERNALLY SYNCHRONIZED RESET EXTERNAL RESET Solution 2: • Needs external phase monitoring • GBT implements this option • Can be provided by FPGA in counting rooms

  46. Solution 3: Metastability issues 1:M CLK IN CLK OUT VCO LPF Sync Example of the CDCE62005 40MHz in VCXO out 1:M 1:N 1:M INTERNAL SYNC RESET

  47. Solution 3: Metastabilityissues • Tests in climatic chamber

  48. TTCex RF_Rx RF2TTC TTcrx CTP LTP TTCvi PH-ESE seminar - 18 Dec 2012 - Sophie Baron

  49. Channel-to-channel skew • For a good event reconstruction • Time Of Flight (TOF) detectors • need less than 10ps rms between channels for particle mass identification • Reducing it: Trade off for the PLLs in the clocktree • Narrowbandwidth to clean the clock as much as possible • Widebandwidth to be sure PLLs do not drift toomuchfromeachother PH-ESE seminar - 18 Dec 2012 - Sophie Baron

More Related