I estimated the efficiencies of the Cerenkovs using ARS and I compared
with the results found by Goulven and Maud. The method presented here
is for an "ideal" case ( not likewise the Scalers ) since I used
the Fastbus data and I didnt take the size the gates for the Scalers
into account.
Signal integration and
corrections :
So, first the ARS signals were integrated and calibrated in terms of
number of pe's (see previous updates on the ARS here ) as shown
on Figure 1.
Figure 1a : Integrated ARS for a french octrnat, LH2 run. The functions
used for the fits are 3 gaussians and the one in the middle is adjusted
manually (the default value) since the gains can change.
Figure 1b : same as above but calibrated
In order to integrate them accuratly, I ajusted my gates of integration
with respect to the distributions of arrival time of the pulses in the
ARS. The following figure is an example of arrival time distribution
for a 32ns run.
Figures 2.1 - Distribution of the Arrival Time of the first pulse seen
in the ARS for octant 6
The ARS is not synchronized to the fastbus data accuratly so I had to
add some corrections on the cer_tdc's . On the figure 3, one can see
that some signals are not recorded in the ARS (ie the ARS record only
noise) but do have a tdc.
Figure 3 : integrated ARS for pmt1 octant 6 as a function of the
cerenkov tdc cer_tdc01, the red part in the middle are events that are
not correctly recorded by the ARS. Note that it is a production run
there.
So I tried to separate those "bad" events from my sample, that may add
a systematic effect on the results but not as big I think (the
percentage of bad events is about 5 or 10% and that s all I can do for
now).
Efficiencies as function of
fpd's :
In order to estimate the efficiencies with the ARS, I define the
Cerenkov multiplicity 2 trigger using the integrated pulses (figure 1b)
for each tube and I put a threshold of about 0.5pe on them. The cut
would be as follow : (arspmt1>0.5 and arspmt2>0.5) or
(arspmt2>0.5 and arspmt3>0.5) or (arspmt3>0.5 and
arspmt4>0.5) ...and so on. The
efficiences can then be found if I plot the arssum signal as a function
of the fpd meantimers as shown on the figure 4.
Figure 4 - The pions and electrons appear well on those plots so
defining a pure electron sample and a pure pion sample is pretty easy.
Once the pions and electrons samples are found, I simply apply the
mutliplicity 2 cut on them. This will define the electron efficiency,
the pion efficiency, the contamination of the electrons in the pion
space and the contamination of the pions in the electron space. For LH2
there are not enough pions (compared to the statistics I have) to
define a pion efficiency so only the electron efficiency is defined.
The results as function of pfd's for all octants are shown on figures 5.
Figure 5 various efficiencies as a function of fpd's.
Now I made fits to find a global value for each octant and I compared
what I got with Goulven and Maud's results as seen on figure 6.
Figure 6 - ARS results in blue, Goulven and Maud's results in red (for
25ns validation gate, green for 32ns validation gate.
