Note on RS scintillator thickness

Note on RS scintillator thickness


Takeshi K. Komatsubara (KEK-JPNS), since 2002.Nov.18

0. Summary

  1. E787 designed values: T 0.635 cm, L2-21 1.905 cm
  2. E787 early analysis: T 0.635 cm, L2-21 1.950 cm
  3. Rob McPherson's analysis: T 0.635 cm, L2-21 in the range 1.905 cm to 1.950 cm
  4. Mark Convery's analysis: T 0.635 cm, L2-21 1.905 cm
  5. UMC in E787 (Version 4,5,6): T 0.635 cm, L2-21 1.905 cm
  6. UMC Version 7, upto 2002.Nov.07: T 0.635 cm, L2-5 1.905 cm, L6-17 1.905 cm, L18 1.905 cm, L19 0.9525 cm
  7. E949 designed values: T 0.700 cm, L2-5 1.905 cm, L6-17(not replaced) 1.905 cm, L18 1.905 cm, L19 1.00 cm (?)
  8. UMC Version 7, updated on 2002.Nov.18:
    IRSTSC=2: T 0.700 cm, L2-5 1.905 cm, L6-17 1.905 cm, L18 1.905 cm, L19 0.9525 cm
    IRSTSC=1(default): T 0.635 cm, L2-5 1.905 cm, L6-17 1.905 cm, L18 1.905 cm, L19 0.9525 cm (as before)
  9. direct measurements: T 0.700 cm, L2-5 (1.88cm to 1.89cm), L6-17 1.905 cm, L18 1.9304 cm, L19 0.9779 cm
  10. offline estimation from rd_pmu.xxxxx


01. E787 designed values: T 0.635 cm, L2-21 1.905 cm

See M.Selen's thesis (1989) and the publication of E787 detector in 1989 [NIM A321 (1992) 129-151].


02. E787 early analysis: T 0.635 cm, L2-21 1.950 cm

In E787 TN-168 (1989) by D.Akerib, D.Marlow, and P.D.Meyers on pnn1 1988,
3.2.6.3 Corrections to the Energy, Momentum, and the Range Scales
"The energy calibration itself originates in the Monte Carlo and was done using the default range stack counter thickness of 1.905 cm. There is evidence from direct measurement and also stopping layer distributions that the actual range stack counter thickness was about 1.95cm. This would have changed the calibration. Therefore, the first correction to real data is a rescaling by 1.95/1.905. Monte Carlo data generated with thick counters is not rescaled but it is smeared with a gaussian distribution to give it the same resolution as real data. (The largest contributions to energy resolution, photoelectron statistics and counter-to-counter calibration errors, are not sumulated in UMC.)"
"The range remains uncorrected save for the fact the Monte Carlo is run with range stack counters of thickness 1.950cm, not the default value of 1.905cm."

In E787 TN-186 (1990) by D.Akerib, D.Marlow, and P.D.Meyers on pnn1 1989,
3.2.6.4 Corrections to the Energy, Momentum, and the Range Scales
"The correction scheme developed in 1988 is used again."

In A.Turcot's thesis (1994) on pnn1 1989-91 (and in E787 TN-281),
4.2.2 Energy
"The raw range stack energy was first rescaled by a factor of 1.95/1.905 to correct for a discrepancy between the actual RS layer thickness and the thickness used in the Monte Carlo simulation employed in the ADC calibration."

UMC for RS ADC calibration: 1.905cm and UMC for analysis: 1.950cm; 1.95/1.905=1.0236


03. Rob McPherson's analysis: T 0.635 cm, L2-21 in the range 1.905 cm to 1.950 cm

In R.McPherson's thesis (1995) on pnn1 1989-91,
4.6 The Kmu2 branching ratio and kaon stopping fraction, f_S
"The efficiency for a Kmu2 decay at rest satisfying the km21 trigger and the DIPANG cut is measured with UMC, and summarized in Rable 4.32. This efficiency depends on the average range stack scintillator thickness used in the simulation. The RS scintillator thickness was nominally specified as 3/4 inch (1.905 cm) per layer. In previous analyses, using the value 1.95cm in UMC has been somewhat more popular, but it is unreasonable to assume that the "correct" value is known to better than somewhere near the thickness. (A brief study was undertaken varying the UMC RS scintillator thickness parameter, and comparing the Kmu2 and Kpi2 stopping layer distributions between data and KOFIA reconstructed UMC. The Kmu2's favoured about 1.97 cm, while the Kpi2's favoured about 1.91 cm.) For this analysis, both 1.905cm and 1.95cm are tried for the acceptance calculation; the average is used for the f_S measurement, and the difference is used for the (approximately 4%) error. It is worth emphasizing that, by design, the K+ -> pi+nunu and K+ -> pi+X acceptances are essentially independent of small variations of this assumption; unfortunately, since our sensitivity is normalized to the Kmu2 branching ratio, it has become deendent on a parameter that we do not know very well."

In A.Adler's thesis (1995) on kpgg 1991 (and in E787 TN-303),
3.10.1 Range Stack Counter Thickness
"The range stack counter thickness was designed to be 3/4" thick. This translates to 1.905 cm. ... an uncertainty in the range stack counter thicknes of 2.3%(1.905cm -> 1.950cm) causes a variation in the acceptance of 14%. There is evidence from studies comparing the stopping distribution of pions and muons from UMC Kpi2 and Kmu2 decays with those of data, that a range stack counter thickness of 1.950 cm may be favored. Therefore a systematic error of 14% will be assigned to the online range cut."


04. Mark Convery's analysis: T 0.635 cm, L2-21 1.905 cm

In M.Comvery's thesis (1996) on kmu2g 1994 (and in E787 TN-317),
5.5.1 Charged Track, Allowed Range of Counter Thickness
"In previous E787 analyses, the range counter thickness has been assumed to be in the range 1.905 cm to 1.950 cm. These analyses have looked at the stopping layer distribution of Kpi2 and Kmu2 events and tried to match these up with the distributions from UMC data generated at different range counter thicknesses. This effort, however, has been somewhat hampered by the trigger bias inherent in most E787 data. ..."
"From the Kmu2gamma trigger, however, we get a source of Kmu2 events in which the stopping layer is allowed to be anywhere in the range from from 16 to 20, thereby giving a much better measurement of the stopping layer distribution. These Kmu2 events can be selected with good purity from the Kmu2gamma trigger by simply applying the NOTPROMPT cut along with the fiducial-type cuts. These cuts are shown in Table 5.21. We then generate UMC Kmu2 events at each range counter thickness requiring only a TdotA as a trigger. We apply the "fiducial" cuts shown in Table 5.21, including the STLAY cut which imposes the Kmu2gamma trigger requirement of stopping layer 16-20. The stopping layer distribution for the data sample is then compared with that of the UMC samples at several different range counter thicknesses. This is shown in Figure 5.14. From these plots it is apparent that a range counter thickness around 1.900 cm is favored."
"Assigning an allowed range is not a completely well defined procedure, but a conservavive range of +-0.02cm is unlikely to be an underestimate."


05. UMC in E787 (Version 4,5,6): T 0.635 cm, L2-21 1.905 cm


06. UMC Version 7, upto 2002.Nov.07: T 0.635 cm, L2-5 1.905 cm, L6-17 1.905 cm, L18 1.905 cm, L19 0.9525 cm


07. E949 designed values: T 0.700 cm, L2-5 1.905 cm, L6-17(not replaced) 1.905 cm, L18 1.905 cm, L19 1.00 cm(?)

In E949 TN-K013 (2000) "T-counters with WLS fiber readout",
2 Fabrication of T-counters
"The counter was cut of Bicron BC404 scintillator plate of 7 mm thickness"
"The thickness of all assembled counters was measured to be 7.75 +- 0.25 mm, where 0.75 mm comes from wrapping and the protrusive fibers."


08. UMC Version 7, updated on 2002.Nov.18:


09. direct measurements: T 0.700 cm, L2-5 (1.88cm to 1.89cm), L6-17 1.905 cm, L18 1.9304 cm, L19 0.9779 cm

Layer 2-5: in Benji's message on Fri, 8 Nov 2002,
"I averaged the measurements of RS scintillator layers 2-5 of the old and new. These measurements were taken by a sonic depth meter in the fiducial region (3 points at distances of 27", 31.5", 36", 40.5", 45" from the long edge (15 total)). The layer 6 measurement is the average of 23 sectors with one measurement taken near the center of the counter for each sector.

                 FYI: the t-counter is 20.5" in length.

                 UMC value = 1.905cm
                
                 Lay   old    new
                 ---   ---    ---
                  1  0.6440   ???
                  2  1.8718  1.8863
                  3  1.8756  1.8843
                  4  1.8758  1.8907
                  5  1.8751  1.8922
                  6  1.8450   ---
     
The new Layers 2-5 are 0.73% thicker than the old Layers 2-5. And the differences to UMC (for Layers 2-5) are 0.87% thinner for the new and 1.60% thinner for the old."

Layer 18: In $UMC_SOURCE/geodat.F, David's comments (with the information from Jim?),

                ------
                C  MOD=7
                C  Scintillator(#18) = 0.760" = 1.9304, Total = 0.816"
                C               actual TSCNRS = 1.905
                C  Scintillator(#19) = 0.375" = 0.9525, Total = 0.393"
                ------
                ( Total == scintillator plus wrappings and foils )
    

Layer 19: In Dima's message to me on Tue, 12 Nov 2002,

                ------
                If we really care about these tenths of milimeter 
                I can add that according to Kelvin although by design 
                the thickness of layer 19 is 0.375''=0.9525cm 
                the real avarage thickness is bigger 0.385''=0.978cm.
                (see his message below).

                >> 1. Is it correct that: T-layer thickness is 6.35mm, 
                >> layer 19 thickness - 10.0mm ?
                >>
                >  The new T-counter thickness by design is 7mm 
                > with 1mm dia readout fibers.  Due to this small increase 
                > in thickness, the radial positions of layer 2-5 will be 
                > also slightly increased from the original data.
                > However, from layer 6 on they remain unchanged.
                > The new layer 19 thickness by design is 0.375".  
                > The real thickness is however on the average thicker, 
                > like 0.385".
                ------
    


10. offline estimation from rd_pmu.xxxxx

In E949 TN-K007 (1999) "Post Mortem Examination of Range Stack Sector 1, Layers 18-21 and Comparison to some new E949 Counters",
Physical inspection
"The table below summarizes some dimensions found. The thickness was determined from several measurements made along one edge of the scintillator. The calculated thicknesscomes from the rd_pmu calibration table, where the assumed saturated energy deposit of 3.04 MeV was scaled to the assumed thickness of 1.905 cm. The calculated thickness may be systematically higher than the measurement, but the variation between layer 18 and layer 19 isn't too far off."

      ------------------------------------------------------------------------------------
                 |            Dimensions (cm)
      ------------------------------------------------------------------------------------
                 |  Minimum	Maximum 	Calculated	Minumum		Maximum
                 |  thickness	thickness	thickness	width		width
		 |				(from rd_pmu)
	Layer 18 |  1.946	1.978		2.041+-0.005	20.97		21.03
	Layer 19 |  1.831	1.854		1.869+-0.021	
	Layer 20 |  1.816	1.862		 -		
	Layer 21 |  1.913	1.948		 -		21.49		21.54
      ------------------------------------------------------------------------------------
   



Last modified by T.K.Komatsubara(KEK), Tue Nov 19 09:54:47 JST 2002