1. STAR Near Term Upgrades Workshop Helen Caines - Yale University June 2004 The SVT Physics and Future Lifetime As impressive as the.. other.. editions are, they seem a little tame compared to the SVT. In fact, this little honey should be able to run with some pretty tall dogs with pretty fancy pedigrees......for the truly young and restless, the SVT is something special. (New Car Test Drive.com ( Ford SVT Contour))

2. Helen Caines – June 2004 2 Performance in run IV Down Feb. 7 th, 8 th, and 9 th Problems with TCD Position Missed ~2.51 M evts. 87.6% Lost MinBias were in Jan. 48 % those from 25 th -31 st

3. Helen Caines – June 2004 3 Detector Status ♦ Averaged Over the Run: ~85% of the SVT is good ♦ Three Half-Ladders (~1.5% each): L07B3, L11B3, & L12B2 ♦ L07B3: no HV above -350 V ♦ L11B3: lost ¼ during '02, rest during shutdown ♦ L12B2: exhibiting abnormally high noise ♦ RDOE7 (~4.2%) Lost due to electronics failure, March 6 th ♦ RDOW3 & RDOW4 down, rectifier failure in PS, March 29 th Recovered April 2 nd ♦ Typical fluctuation ~3%

4. Helen Caines – June 2004 4 “Bad” channels history Bad = dead or noisy 1.36 Ladders builtEnded Jan 2001 < 1% 2.Ladders mounted on end-rings & installed cone Ended March 2001 3.Run II Au-AuAug 2001 - Nov 2001 4.Run II p-pDec 2001 - Jan 2002~3.7 % 5.Shut down (leak repair)Feb 2002 - Sept 2002 6.Run III commissioning Oct 2002 - Dec 2002~10.5% 7.Run III d-Au and p-pJan 2003 - May 2003 ~12.7% 8.Start Run IVJan 2004~13.6% 9.End Run IVMay 2004~15.9% Note: % doesn’t show those used in various run but those intrinsically damaged. i.e. Read Out Boxes failed but were repaired during the run. Therefore a slightly lower operational % used than shown here. Seems that when SVT left alone number of bad channels stays stable

5. Helen Caines – June 2004 5 Detector lifetime Future difficult to predict: ♦This year a maximum of 3% bad channels added ♦ Extrapolating 5 years  70% alive BUT ♦ Most failures seem to be “grouped” ♦ one sca out of 5 fails ♦ one analog driver is bad ♦ bad/broken connector from detector to rdo ♦ bad adc in rdo ♦ Not much trend of single channel failures With minimal handling and disconnecting from RDO’s things may not be that bad

6. Helen Caines – June 2004 6

7. 7 Calibration Techniques ♦ Gain ♦Hybrid to hybrid and within hybrid. 1. Look at hits placed on tracks with given mtm and average charge should be the same. Scale “gain” to force them to be. ♦ Drift Velocity ♦ Hybrid to Hybrid and within hybrid. 1.Look at start and stop of hits – Know drift == 3cm, calc V dhybrid 2.Use laser spots to monitor temp. variation event by event.. 3.Use bench measurements to account of non-linearity of drift. 4.Use bench measurements to account for temp. profile across anodes. ♦Alignment ♦Global, Barrel, Ladder, Wafer. 1.Project TPC tracks to SVT hits, calc. residuals. 2.Refit TPC tracks with SVT hits, calc. residuals. 3.Refit matched SVT hits and primary vertex, calc. residuals. ♦Deviations from means of zero give shifts. ♦Try shifts and rotations to minimize offsets. ♦Some offsets due to TPC distortions not ONLY SVT.

8. Helen Caines – June 2004 8 Drift velocity from hits Fitting First & Last Points Charge Injectors 3 cm Mean distortion is a few 100  m Mean distortion is a few 100  m

9. Helen Caines – June 2004 9 Polynomial drift representatopm 9 th order polynomial Difference from fit Difference from fit RMS=17.9 µm Account for focusing region

10. Helen Caines – June 2004 10 Hit position reproducibility spot 1 σ=4.4μm spot 2 σ=3.0μm anode σ=5.985 μm Laboratory laser tests: anode direction: σ=6 μm 3 laser spots 2 spots are at: hybrid=1, layer=6, ladder=15, wafer=7 1 spot is at: hybrid=2, layer=6, ladder=7, wafer=1 ♦ SVT proposal Similar resolution in STAR as on bench

11. Helen Caines – June 2004 11 Time variations of laser spot positions Temperature oscillations have a period of ~2.5 min Temperature oscillation is ~1 o c peak-to-peak Position peak-to-peak change is ~70 μm slow-control temperature measurement drift distance of spot 1

12. Helen Caines – June 2004 12 water cooling  time variations of laser spot positions spot positions change in phase BUT spots behave differently after SVT is switched on and gets stabilized (~ 1 hour !) spot 1: 80 microns spot 2: stable spot 4: 300 microns Time variations of laser spot spot 1 spot 4 spot 2 Can we calibrate out the burn in?

13. Helen Caines – June 2004 13 Drift velocity calibration Time difference of spot 1 and 2 Time variations of drift velocity d 1 – d 2 = d 12 = v(t 1 -t 2 ) ♦ Use 2 spots on one wafer ♦ Distance is in fact the same so apparent change is due to  v ♦ Using and can calc.  v event-by-event

14. Helen Caines – June 2004 14 Corrected laser spot positions black before, red after the drift velocity calibration before: σ=(22.1±0.2)μm after: σ=(10.0±0.1)μm ♦ Get delta function for spots used in calc – phew! Laser spot on other wafer improves by factor 2 time direction: σ= (5 – 23) μm (depending on the position) (NIM A439, 2000; SVT proposal)

15. Helen Caines – June 2004 15 Anode temperature profile Wafer width ● 40 ns/TB = ~270 µm/TB ● ~150  m max shift ♦ Temperature gradient across wafers must be taken into account ♦ Due to resistor chains at edges Have bench measurement for each hybrid needs to be used

16. Helen Caines – June 2004 16 Alignment We seek for 6 parameters that must be adjusted in order to have the SVT aligned to the TPC:  x shift  y shift  z shift  xy rotation  xz rotation  yz rotation Have to calculate for each wafer – 216 in total ♦ How to disentangle and extract them without ambiguity from the data? ♦ Many approaches are possible. We are using two of them... The Question

17. Helen Caines – June 2004 17 The two approaches First approach:  Calculate the “residuals” between the projections of TPC tracks and the closest SVT hit in a particular wafer.  Advantage: l can be done immediately TPC calibration is OK (not final), even without B=0 data.  Disadvantage: l highly dependent on TPC calibration. l the width of these “residuals” distributions and therefore the precision of the procedure is determined by the projection resolution. Second approach:  Use only SVT hits in order to perform a self-alignment of the detector.  Advantage: l a better precision can be achieved. l does not depend on TPC calibration.  Disadvantage: l it is harder to disentangle the various degrees of freedom of the detector (need to use primary vertex as an external reference). l depends on B=0 data (can take longer to get started).

18. Helen Caines – June 2004 18 First approach: TPC track projection Try to disentangle the 6 correction parameters in 2 classes:  x shift, y shift and xy rotation.  z shift, xz rotation and yz rotation. Look at “residuals” from the SVT anode direction (global z direction); Choose tracks with dip angle close to zero ( tracks parallel to the xy plane); Study as function of z: deviations from a flat distribution centered at zero show mis-alignment They are not completely disentangled, but it works as a first approximation.. Make the alignment in steps: 1.global alignment, i.e., one set of parameters for the whole detector; 2.ladder by ladder alignment, i.e., a set of parameters per ladder; 3.wafer by wafer alignment, i.e., a set of parameters per wafer.

19. Helen Caines – June 2004 19  x = -0.25 mm  y = 0.10 mm  = -0.0018 rad  x,  y,  corrections  x = -1.9 mm  y = 0.36 mm  = -0.017 rad Matches well the survey data Looks pretty good after 2 nd iteration

20. Helen Caines – June 2004 20 First look z shifts, no corrections ♦ Only ladders at xz plane ♦ Only ladders at yz plane No sizable correction needed

21. Helen Caines – June 2004 21 Next step ladder by ladder ♦ Look at “residuals” from the SVT drift direction (global x-y plane). ♦ Study them as a function of drift distance (x local ) for each wafer. ♦ Now influence of mis-calibration (t 0 and drift velocity) cannot be neglected. v` is the correct drift velocity and t 0 ` is the correct time zero. 0, if t 0 is Ok These two equations can be used to fit the “residuals” distribution fixing the same geometrical parameters for all wafers.

22. Helen Caines – June 2004 22 One ladder as example  x = -0.81 mm  y = 0.56 mm  x = -0.19 mm  y = 0.024 mm

23. Helen Caines – June 2004 23 Technique works!  Need to go ladder by ladder (36 total) checking the correction numbers and the effect of them on the “residuals”.  Still need to consider the rotation degree of freedom. Next step is to fit each wafer separately.

24. Helen Caines – June 2004 24 Near future perspectives Finalize first approach:  calculate  x,  y, and  ladder by ladder. (Just finished for inner barrel)  extend it to wafer by wafer making small corrections if necessary;  calculate z shift, xz rotation and yz rotation (global, ladder by ladder and wafer by wafer - they should be small);  use B=0 data. ( Work being done now) ♦ It is a lot of work, but it depends mostly on man power. ♦ Software is ready ♦ The whole procedure does not depend on many iterations of the reconstruction chain. Can be applied and tested without full reconstruction

25. Helen Caines – June 2004 25 Track matching efficiency SVT is most efficient between PV +- 10 cm!!! HIJING 200 GeVAuAu Prod62 GeV Global Tracks 80%76% Primary Tracks 85%83% Note: Get similar tracking results with Sti

26. Helen Caines – June 2004 26 Track matching efficiency II Number of EST tracks does not increase with multiplicity, an indication that there is no increase of ghost tracks.

27. Helen Caines – June 2004 27 Identifying primary tracks ♦Hijing ♦ Data Significant improvement in impact parameter for primary tracks Significant improvement in impact parameter for primary tracks

28. Helen Caines – June 2004 28 Track residuals: simulated data ♦ Dependency on the attributed hit error – simulation had 20  m smearing p-p

29. Helen Caines – June 2004 29 Track Residuals AnodeDrift Only TPC hits 320  m350  m PV + 2 SVT hits + TPC hits 120  m PV + 2 SVT hits 110  m120  m PV + 3 SVT hits 72  m 70  m HIJING AuAu simulated data, with SVT hits position smeared by 80  m. Resolution of ~SQRT(110*72) = 90  m Au-Au 62 GeV AnodeDriftSolution Average over all Barrel 2 180  m 300  m Barrel Alignment

30. Helen Caines – June 2004 30 Track Residuals: Au+Au 62 GeV AnodeDriftSolution Average over all Barrel 2 180  m300  m Ladder 03 84  m140  m Alignment between ladders. Ladder 10 280  m180  m

31. Helen Caines – June 2004 31 Track Residuals: Au+Au 62 GeV AnodeDriftSolution Average over all Barrel 2 180  m300  m Ladder 03 84  m140  m Ladder Alignment L03/wafer 48 60  m140  m Wafer Alignment L03/wafer 48/hybrid-02 60  m Alignment & drift velocity Z vs. Drift Distance Y vs. Drift Distance

32. Helen Caines – June 2004 32

33. Helen Caines – June 2004 33 Physics interests ♦ Measurement of low p T (60-200 MeV/c) particles ♦ Improved reconstruction of strange baryons ♦ Reconstruction of heavy quark mesons (not going to discuss) ♦ Improvement of high p T primaries reconstruction/resolution (already shown dca improvement)

34. Helen Caines – June 2004 34 Low-p T yields compared to models Event generators unable to consistently describe low p T yields. HIJING overpredicts yields at low p T. Ratio of measured to HIJING yields averaged over low p T range: ( Strangeness Production in PHOBOS”, C. Henderson, MIT, RHIC/AGS Users’ Meeting, May 2004, BNL)

35. Helen Caines – June 2004 35 Low p T physics – particle yields p-p dN/dy  : Exponential: 0.00268 +/- 0.0005 Power Law: 0.00181 +/- 0.0004  : Exponential: 0.00270 +/- 0.0005 Power Law: 0.00178 +/- 0.0004 Distinguishing between fit models is critical to determine yields!

36. Helen Caines – June 2004 36 Sti - efficiency and purity ♦ Clean up via normal tracking first ♦ Start with SVT and move outwards ♦ ♦ Efficiency is low but that would be OK as long as pure ♦ ♦ By removing obviously bad hits purity good Work just starting but looks promising!

37. Helen Caines – June 2004 37 K 0 s in Au-Au 62 GeV K 0 s raw yield vs P T SVT/TPC yield P T < 400 Mev/c 3% improvement for K 0 s 120 % improvement in lowest p T bin similar result seen in p-p

38. Helen Caines – June 2004 38  in Au-Au 62 GeV P T < 800 Mev/c 15% improvement for  40% improvement in lowest p T bin 100% seen in p-p for p T < 0.5 GeV

39. Helen Caines – June 2004 39 MSB from Au+Au-Simulation  enhanced Hijing events SVT+TPC Tracks: 3210 Ω’s TPC only Tracks: 2020 Ω’s Increase of signal by 60% 

40. Helen Caines – June 2004 40 peak ADC BLACK-real AuAu 200GeV GREEN- Hijing simulation ADC sum time bin T0T0 Slow simulator close to tuned

41. Helen Caines – June 2004 41 simulated hits time real data to embed into anode time embedded data anode time simulated raw data anode time anode Embedding Works

42. Helen Caines – June 2004 42 Preliminary test on 200 GeV dAu data Black – TPC only Red – TPC + SVT requiring 2 SVT hits on EST track Correcting data seems OK

43. Helen Caines – June 2004 43 SVT lifetime expectations ♦ We do not see evidence for aging or continuous channel loss therefore the limitation of the usefulness of the detector is due to the RDO. ♦ The detector can run up to a maximum DAQ rate of around 200 Hz without serious performance limitations ♦ Between 200 to 300 Hz the required SCA settling time will lead to increased noise levels and the detector requires hardware changes to the RDO. ♦ The detector can not operate above 300 Hz. ♦ The detector electronics can not be upgraded as was done in the case of the TPC.

44. Helen Caines – June 2004 44 SVT future plans ♦ We do not anticipate to remove the detector for repair or electronics upgrades. ♦ We would like to run the detector until STAR decides to remove it due to readout speed limitations. ♦ We like to be in STAR at least until 2009 beam time. ♦ The detector is designed such that the inner layer can be removed to generate space for the APS upgrade.