The Absolute Calibration of the HiRes

Detectors

J.N. Matthews, S.B. Thomas, N. Manago,

L. Perera, G. Burt, and R. SnowFor the HiRes Collaboration

RXF Calibration

• The width/mean calibration

• Calibration of the Standard Candle

• Comparison of the two techniques

• Conclusion

Energy Spectrum - HiRes & AGASA

AGASA energy scaled by 0.79

Systematic Uncertainties• PMT calibration: 10%

• Fluorescence yield: 10%

• Unobserved energy: 5%

• Atmospheric absorption: most sensitive to vertical aerosol optical depth (VAOD)

– Mean VAOD = 0.04

– VAOD RMS = 0.02

– VAOD systematic is smaller.

– Modify MC and analysis programs to use VAOD = 0.02 and 0.06, reanalyze.

– J(E) changes by 15%

• Total systematic uncertainty = 21%

PMT Clusters/Cameras

16 x 16 PMT Array

•2 Sites – 12 km

•34 Buildings

•64 Cameras

•16,384 PMTs

PMT Calibration

• Roving Xenon Flasher (RXF) as a Standard Candle

• RXF sits at the center of the mirror and illuminates the entire cluster

• Can take it to each PMT cluster at both sites

• Very stable (< 2% over a night)

Current Statistical Method

• The measured QDC distribution is used to calculate the number of photo-electrons for each PMT:

Npe = α (μ/σ)2

• Applying the quantum efficiency then gives us the gain of the PMT or the number of ADC counts per

Applied Calibration Result

SigmaSigma

Problems with the Current Method

• The excess noise factor, α, is not well known. The value was determined by a small number of single photo-electrons measurements and atmospheric measurements (molecular edge). It is consistent with values reported in the literature and discussions with Photonis.

• The QE() is not well understood.

Calibration of the Standard Candle

• A two step approach using NIST calibrated silicon photo-diodes and hybrid photo-diodes (HPD)

• Use the NIST Si photo-diodes to determine the response of the HPD by measuring the single photo-electron peak

• Then use the HPD to measure the luminosity of the RXF (# /mm2 delivered at the mirror-cluster separation distance)

SPE Distribution

P(0)P(0)

Mean: 907.9 ; Events : 509991Mean: 907.9 ; Events : 509991 P(1) 1015.6; Events: 534870P(1) 1015.6; Events: 534870

P(1)P(1)

P(0)P(0)

A Gaussian A Gaussian fit is applied fit is applied to P(1) to to P(1) to obtain the obtain the mean.mean.

RXF Measurement with HPDMean: 907.5 Events: 2275 Mean: 907.5 Events: 2275 Mean: 8412 Events: 3203Mean: 8412 Events: 3203

CalculationsCalculations

(RXF signal – RXF background)(RXF signal – RXF background)

0.183* 0.68 *(SPE Peak – background)*Area0.183* 0.68 *(SPE Peak – background)*Area

HPD Calibration ConstantsHPD Calibration Constants

• 0.183 is the HPD efficiency0.183 is the HPD efficiency• 0.68 is a distribution correction factor for the HPD. It is 0.68 is a distribution correction factor for the HPD. It is the ratio of the SPE mean over the SPE peak. the ratio of the SPE mean over the SPE peak.

γγmmmm22 ==

HPD Systematics• Si diode current: I = 0.192*10-9 A ± 1% ± 2.6%• Si diode responsivity: β = 0.11325 A.s/J ± 5% ± 0%• Si diode area: A = 100mm2 ± 1% ± 0%• HPD count rate: μ = 16.22*103 counts/sec ± 5% ± 0%• Counting Efficiency: ε = 1.054 PE/count ± 3.5% ± 0%• Neutral density filter attenuation: η = 1.36*104 ± 1.5% ± 0%• Geometrical and other effects: ±5% ± 0%

• HPD Efficiency = 11.2 ± 9.6%Syst ± 2.6%Stat (# /pe/mm2)

A Test:Measure and compare the light output of the RXF via:

The HiRes detector width/mean methodThe HPD traceable to NIST

The Plan:• Measure RXF with the HPD in the University of

Utah lab• Take RXF to HiRes to calibrate the detector.

Used 20 cameras at HiRes-I for this test. Ten minutes (about 900 shots) of data are recorded at each detector.

• Return to the U and remeasure RXF in the lab.

Result:

0

2

4

6

8

10

12

Statistical Method:

8.97 /mm2 ± 10% (Syst)

HPD Method:

9.55 /mm2 ± 9.6% (Syst) ± 2.6% (Stat)

Comparison percentage is 6.5%.

Well within expected values.

Conclusions

• The NIST Si-PD and HPD calibration will allow us to absolutely calibrate the RXF and thus the HiRes PMT Clusters.

• Good agreement with the statistical method.

• We need to work on the uncertainties to get them down. Should be able to get to the ~5% level.

The High Resolution Fly’s Eye (HiRes) Experiment

• University of Utah• Rutgers University• Columbia University• University of Montana• University of New

Mexico• University of Adelaide• Los Alamos National

Laboratory (LANL)• University of Tokyo