Scintillator trigger efficiency


Alexandre Kozlov
University of Regina, CANADA
for the Hall A collaboration
(thank you to E. Brash, S. Dumalski, M. Jones, R. Roche, L. Todor)

This talk is closely related to data analysis of the E89028 experiment which consist of D(e,e'p)n and H(e,e'p) data sets taken at $Q^2$= 0.4, 1.0 and 1.6 $GeV^2$.

Two types of inefficiency discussed here:

1) absolute inefficiency can be result of scattering between (in) the S1 and S2 scintillator planes. Some events therefore can miss S2 plane causing absolute trigger inefficiency. GEANT model of HRS was used to estimate the inefficiency value as a function of momentum of detected particles (protons in our case).

2) S1 or S2 detector inefficiency due to insufficient light collection (or any other reason) which would result in loss of coincidence trigger.

A fully working example which can be used to get scintillator trigger efficiency is at: http://www.phys.uregina.ca/$\sim$alk

Comments: Due to the limited geometrical size of S2 scintillator and relatively large distance between both scintillator planes some events could simply miss S2 plane. This inefficiency is a function of the particle initial momentum and gets larger close to the detector boundaries. In order to study this problem we used MCEEP code to generate the focal plane variables for the case of D(e,e'p)n reaction. These generated events were used as an input in the GEANT code which includes all detectors corresponding to the Hadron arm spectrometer.

Comments: Absolute scintillator inefficiency as a function of initial (proton) momentum. It is clear that correction needs to be done for the proton momenta less than 1 GeV/c if one proceed with the cross-section measurements. At p=400 MeV/c this inefficiency could be more than 5 %.

Comments: Nuclear interaction turned on
Absolute scintillator inefficiency as a function of S1 x-scint variable for a number of proton momenta. Events were initially generated in MCEEP for the D(e,e'p)n reaction using 100 MeV/c steps in proton momentum and then tracked in the GEANT FPP code. Histograms show distribution of events which crossed S1 plane but haven't hit S2 plane. Spikes at the histogram edges are due to the limited number of events in these bins. At the same time real inefficiency also gets larger near the geometrical boundaries of the scintillators

Comments: Nuclear interaction turned off
It is clear that inefficiency almost entirely resulted from the N-N collisions except for a tiny area near the edge of the S1 scintillator where multiple scattering also contributed in case of the lowest proton momenta.

Comments: In order to be able to correct for trigger inefficiency one needs to weight each event with a factor 1/$\epsilon$, where $\epsilon$ is scintillator efficiency. First, in Espace histograms were filled for S1 and S2 x scint variables for all types of trigger events. Then, in PAW++counts in these histograms were scaled using prescaling factors and corrected for computer and electronic deadtime. In case of E89028, number of recorded T2 and T4 triggers was not large, so all runs were combined together to improve statistics. Dead-time values and scaling factors were weighted with the number of events detected in each run in order to get there values for each kinematic setting. Afterwards, efficiency profiles were obtained and fitted with P6 function. Resulted coefficients were written to a text file. When one needs to fill histograms from ntuple those coefficients used to calculate weighting factors according to the particle track position on S1 and S2 planes.

STEP 1
Espace analysis

E-arm cuts for each trigger type:
multiplicity	1 (in 1 plane only)
track number	1
ADC gas Cherenkov	0-6000
Scintillator TDC	0-5000
S1: X-scint (m)	-1.05 0.8
S1: Y-scint (m)	-0.18 0.18
S2: X-scint (m)	-1.3 1.0
S2: Y-scint (m)	-0.32 0.32
S1:TDC left .OR. TDC right	true
S2:TDC left .OR. TDC right	true
T1,5 events require S1 TDC .AND. S2 TDC	true
T2 events require S1 TDC .OR. S2 TDC	true
H-arm cuts for each trigger type:
multiplicity	1 (in 1 plane only)
track number	1
Scintillator TDC	0-5000
S1: X-scint (m)	-1.1 0.85
S1: Y-scint (m)	-0.18 0.18
S2: X-scint (m)	-1.25 1.05
S2: Y-scint (m)	-0.32 0.32
FPP X (m)	-0.9 0.9
FPP Y (m)	-0.2 0.2
S1:TDC left .OR. TDC right	true
S2:TDC left .OR. TDC right	true
T3,5 events require S1 TDC .AND. S2 TDC	true
T4 events require S1 TDC .OR. S2 TDC	true

Comments: In H-arm FPP front chambers were used to select good events. FPP X and Y are the x and y coordinates of the track projected to the plane z=0, which is coincident with the first plane of the vdc's

STEP 2
PAW++ analysis

 
1) use awk to search and read database (e.g. runs_info.dat) for each run 
e.g. search for run number=[nrun] and read dead-time values corresponding 
to each trigger type     

do i=1,5
t=[i]+4
xdt[i]=$SHELL('cat runs_info.dat|awk ''/'//[nrun]://'/{print $'//[t]//'}'' ')
dti[i]=1./(1.-[xdt[i]])
dt[i]=[dt[i]]+[dti[i]]*[nti[i]]
enddo

2) get averaged dead-time and prescaling values

$A_{tot}^{T_j} =\frac{\sum\limits_{i=1}^{n} A_{i}^{T_j}\cdot N_{i}^{T_j}}{\sum\limits_{i=1}^{n} N_{i}^{T_j}}$

$N_{i}^{T_j}$ is number of events corresponding to trigger type ${T_j}$,
where ($j=1-5$) for the run i from the total number of runs n
$A_{i}^{T_j}$ is either CDT, EDT or prescaling factors
$A_{tot}^{T_j}$ is an averaged value for either CDT, EDT or prescaling factors

3) multiply histograms with the corresponding factors so that efficiency $I$ is:

$I=\frac{N_{T_{1,3}}*ps_{T_{1,3}}*edt_{T_{1,3}}*cdt_{T_{1,3}}+N_{T_5}*ps_{T_5}*e... ...{T_5}*edt_{T_5}*cdt_{T_5}+N_{T_{2,4}}*ps_{T_{2,4}}*edt_{T_{2,4}}*cdt_{T_{2,4}}}$

where edt (cdt) are correction factors due to the electronic (computer) deadtime

4) fit efficiency profile with a function (P6 in this case)

STEP 3
HOW TO CORRECT SCINTILLATOR INEFFICIENCY EVENT-BY-EVENT

* Include this peace of code into the PAW++ KUMAC file used to fill
* histograms from NTUPLE
* Cut to limit scintillator acceptance within the physical boundaries
ntu/cut $60 -1.<e_s1_x<0.75
ntu/cut $61 -1.1<e_s2_x<0.9
ntu/cut $62 -1.<h_s1_x<0.8
ntu/cut $63 -1.1<h_s2_x<0.9
* Read P6 polinomial coefficients from the file
file=[main]efficiency/eff_[kinem].txt 
v/cr scint(7,4) R 28*0.
v/read scint [file] ' ' 
do j=1,4
 do i=1,7
   k=$EVAL([i]-1)
   a[k][j]=$EVAL(scint([i],[j]))
   mess a[k][j] = [a[k][j]]
 enddo
enddo
* Calculate efficiency for each scintillator plane
ntu/cut $64 $60*1/([a01]+[a11]*e_s1_x +[a21]*(e_s1_x*$60)**2 + ...)
ntu/cut $65 $61*1/([a02]+[a12]*e_s2_x +[a22]*(e_s2_x*$61)**2 + ...)
ntu/cut $66 $62*1/([a03]+[a13]*h_s1_x +[a23]*(h_s1_x*$62)**2 + ...)
ntu/cut $67 $63*1/([a04]+[a14]*h_s2_x +[a24]*(h_s2_x*$63)**2 + ...)
* Weight each event with corresponding correction factor 
* Cuts 52-55 are not related to this procedure
ntu/cut $99 ($52.AND.$53.AND.$54.AND.$55)*[dtime_cor]*[vdc]*$64*$65*$66*$67

--------------------------------------------------------------------------------
INPUT FILE P6 coeff. for each of the four scint. planes 

0.99662 0.75063E-04 -.72454E-02 0.16073E-02 0.46816E-01 -.68234E-02 -.91917E-01
0.99576 0.19941E-03 -.31770E-02 0.49745E-03 0.56260E-02 -.44618E-02 -.11486E-01
0.97318 0.18918E-01 -.45118E-01 -.13530     0.26501     0.20152     -.36370
0.98820 0.34333E-02 -.89079E-02 -.30557E-01 0.19490E-01 0.47556E-01 0.72603E-02

Comments: Number of T1,T5,T2, and T3,T5,T4 triggers (from left to right) corrected for deadtime and scaling factors for the case of $Q^2$=1.6$GeV^2$ kinematics.

Comments: Efficiency profile, dashed red curve is result of the fit

Comments: Efficiency profile, dashed red curve is result of the fit

Comments: Efficiency profile, dashed red curve is result of the fit

Comments: Efficiency profile, dashed red curve is result of the fit. Efficiency fall sharply near the edge of the acceptance, significant inefficiency related to a single scintillator paddle in the S1 plane can be seen

Comments: Efficiency profile, dashed red curve is result of the fit. Efficiency fall sharply near the edge of the acceptance, significant inefficiency related to a single scintillator paddle in the S1 plane can be seen

Comments: Efficiency for both S1 and S2 are higher for these low proton momenta since at this energies protons deposit more energy and generate more light in the scintillator material. The fact that efficiency has such dependence indicate that we are dealing with ``real'' inefficiency due to insufficient light collection.

Comments: Energy loss for ionization from the Bethe-Bloch formula. Arrows show particle momenta corresponding to the E89028 experiment. It can be seen that amount of light generated in scintillators by electrons and protons was approximately the same for all kinematics except for the case of low proton momentum setting of 700 MeV/c.

Conclusion

• Simulation studies indicate that under certain conditions (hadron momentum lower than 800-1000 MeV/c and also close to the detector boundaries) we observed significant trigger inefficiency (in S2 plane) due to the elastic and inelastic N-N collisions in the detector materials. These type of absolute inefficiency will not be accounted completely if we include a 3rd detector condition in order to do selection of ``good'' events in the ``standard'' correction procedure. More work needs to be done in order to get better understanding of how to avoid this problem. At the same time this effect is significant only for the low energy hadrons and should not effect the electron arm spectrometer.
• Standard procedure (with some modifications) used to correct for loss of the good events due to insufficient light collection or other reason was briefly described here. Further modifications might be necessary to make it applicable for the cases of other reactions or detector sets.