Slide N 1

University of Iowa

HF sourcing – Histograming mode

Alexi Mestvirishvili

ECLIPSE 06Antalya, Turkey

Slide N 2

HF minus

Ten 10o sectors – 28 – 36 and 1 analyzed so far

Signal+Pedeatal

QIE counts vs position

Expected signal – 0.01

Main calibrationHTR running in histograming mode, QIEs running in normal scale. Reading out of one spigot at the time (24 channels, 12EM, 12HAD)During sourcing source was stopped in three different location First and last – for safety reasons. Second stop for 3 minutes – to accumulate data. Position of second stop point was chosen in the way to have both long and short fibers illuminated.

x10

Slide N 3

Signal + pedestal and pedestal separation and signal versus tower

EM fibers

HA fibers

Small peak – pure pedestalHigh peak – Signal+pedestal

Gaussian fit applied

For these distributionsare 0.0018.

5 wedges 24 tower each

Slide N 4

Problems

Gaussian fit sigma for S+P vs Tower

EM channels

HAD channels

Run 1566 & 1567Tower 12, phi 36Run 1490

Tower 23, phi 33

Slide N 5

Problems (run 1566)

Red signal + pedestalGreen pedestal

Four different capacitors

Slide N 6

Calibration Precision

Calibration precision is limited by several factor

1. Statistical precision of signal (relative error on signal)

2. Pedestal stability

3. Internal nonlinearities of absorber and fibers (believed to be at the level 5%, not studied for this presentation)

4. etc.

Slide N 7

Statistical PrecisionPhi=29, tower=16, QIE ch 2,8

a) b)

a) Signal + pedestal Nentries = 14290Mean = 5.05Sigma = 0.0018Err on Mean /sqrt(Nentries) =1.5x10-5

b) PedestalNentries = 3408Mean = 5.039Sigma = 0.0018Err on mean = 3x10-5

= 5.05 – 5.039 = 0.011 = 3.3x10-5

X axis units – linearized QIE counts

/=0.3%

Slide N 8

Relative error on signal

Small towers has smaller signals hence big error

Slide N 9

Pedestal stability

Pedestal stability can be studied from same data. In one 10o

24 channels (12EM, 12HAD) are read out during sourcing ofone tube. Either 16 or 15 tubes are in one phi sector. In eachrun only two channels (1EM, 1HAD) caries signal. All otherchannels gives pure pedestal, provided adjacent towers of one which is sourced are excluded from data samples

One can easily select samples for the same channel and wellseparated in time. App 5 minute is necessary to source onetube. For example, two samples for tower 5, when source is in tube 1 (tower 1) and in tube 16 (tower 12) are separated by more than one 1 hour.

Slide N 10

Pedestal stability

1,1,,1,1, kkkjjji

– tower for which pedestal is determined

– towers which are sourced, they are

separated in time as far as possible

ikj,

Condition which has to be fulfilled to avoid contributionFrom energy leakage from tower to tower

Measure of pedestal stabilityis width of pedestal distribution ij, ik

Slide N 11

of pedestal distributionstower 1 to 12

Tow 12,=36

Smallest tower Is noisy as well

Slide N 12

of pedestal distribution

towers 13 to 24

Tow 12,=36

Small tower noisy as well

Slide N 13

Geometrical corrections

Geometrical corrections has to be applied to signals from each tower to exclude the differences in towershape.

Coefficients were taken from Monte Carlo. They arecalculated with respect to the response of infinitesize tower taken into account actual tower geometry

Coefficients are available in HF monitoring databaseand they were calculated in CMS IN 2004/002

i

iicorr GC

SigSig ,

Slide N 14

Geometrical correction how they work (problems ?)

for HAD towers

for EM towers

High Medium Low

PMT gains

Gains in each groupare more or less equal

Uncorrected, pedestalsubtracted signal forphi = 28

For towers having two source tubes geometrical correction should bringtwo points close to each other

Slide N 15

Geometrical corrections how they work (problems)

pedestal subtracted,Corrected,

signal for phi = 28

Same as on previous slide

High Medium Low

PMT gains

Points in red ovalsmove in oppositedirection

Points in last groupbehaves as it should

Slide N 16

SUMMARY

1. In general, data quality is good, signal can be separated well even for small towers2. Statistical precision of HF calibration will be very high, less than 1%. 3. Stable pedestal4. Some problems were identified – bad capacitors, noisy channels5. Geometrical scale factors has to be recalculated with correct tower geometry (!!!!!!!!!)

Slide N 17

Backup slides

Slide N 18

Stability of pedestal Mean

RMS = 0.00046 RMS = 0.00034 RMS = 0.00027 RMS = 0.00034

),min( ijik MM

MM ikij – tower for which pedestal is determined

– towers which are sourced, they are

separated in time as far as possible

ikj,