Here I present my work on the Cerenkov efficiency using the ARS
signals. For all this study I put cuts in the elastic locus in order to
have fairly reliable results.
For the ARS integration, I used the arrival time distribution made by
Goulven (dont pay attention to the peak on the left of the figure 1.1,
it is an artefact of his program ) :
Figure 1.1. exemple of ARS 1pe arrival time for octant 1.
Figure 1.2. exemple of ARS 1pe arrival time for octant XX.
From this distribution, I took a gate for the integration between 40
and 85 channels roughly and I performed a pedestal subtraction by
averaging values in the background (channels 10 to 20 in the ARS pulses
vs time) and substract the result while integrating.
So I did loops of integration over all events and finally got some new
distributions that looks much better than the regular cer_adc's :
Figure 1.3
I took the results of the fits (using 3 gaussians ) to try to calibrate
the signal in terms of number of Pe's :
Figure 1.4
As one may notice, the pedestal is still very high (up to several
thousand events depending on the run) and the results at
higher beam current and for the 2ns beam structure, the distributions
are not clean at all and some quite disturbing negative values does
appear. We can see this better when ploting the Sum over the 4 PMT's :
Figure 1.5.
If we look back at goulven's plots, we notice that a bunch of events
comes too early, which hurt my pedestal subraction !!
So I decided to just drop those events by comparing the average value
of my noise that I take between channel 10 and 20 with some value in my
gate of integration. By doing this I can just drop the event if what I
consider "noise" in this case is to high. The following plots shows a
much lower and cleaner pedestal :
Figure 1.6: Integrated ARS distribution for a french octant :
Figure 1.6 : Integrated ARS distribution for a NA octant :
Figure 1.7.
At this point I noticed that the NA octant had a very disturbing
problem : why is there a huge number of event that peaks at 0 ?????
So I found a way to get ride of
it by modifying my program (the issue is still there right now).
Right now the results starts to look decent. I made also some plots of
the ARS sum signal (sum of all 4 pmt's after calibration) :
Figure 1.8.
Now, let's try to apply multiplicity 2 (0.6pe on at least 2 PMT's):
Figure 1.9. Multiplicity 2 cut - The blue curve corresponds
to what I assume to be mostly electrons. The red curve is the
left over of my Multi 2 cut.
32ns ARS analysis (last update :Aug 29):
I looked at some 32ns runs with ARS and I tried to estimate the
Cerenkov efficiency by looking at my integrated ARS signal vs f_mt12
(FPD12 mean timer ) :
Figure 1.10 : Integrated ARS versus fpd mean timers for octant 7 , run
30186. The blue dots correspond to a cut on multiplicity 2.
As you may notice there is a big amount of electrons recognized
as pion in the ARS :/ . Note : this issue does not appear
on the run 29304 for octant 7 (spring 32ns run).
Those miss-recognized electrons by the ARS can divise my estimation of
the effiency by a factor 2 !!
On the next figure a plot of the Spring run :
Figure 1.11 - same kind of analysis with a run made in spring (
notice that the elastic locus was not define the same way ).
ARS analysis - to do list and remaining issues :
- Also try to understand why there is a pedestal in
the "cer_trig_tdc - trigger_tdc " gate where it should be only
electrons ( as far as I understand this trigger ) :
-> at least the pe's are within 26ns which is good !
Update :
Figure 2.1 : ARS signals vs trig gate ( in ns )
- Define the contamination.
-> cannot possibly do with those data.
- search for "real" pions and "real" electrons.
- check if consistent with physics/simulations , etc.
- look at the pedestal more carefully for 2ns runs. - > maybe it was
wrong to cut events that comes too early ,I am not sure.
Work in
progress :) (last update : Aug 29):
- Try to do a threshold scan using ARS :
Update : I did the threshold scan and I didnt see much for now - >
need to fix the loss of efficiency before I can get something nice.
- Try to plot the ARS for different ced / fpd regions -> need more
statistics .
-> done for the FPD regions
- calibrate all the timing variables in terms of "ns" : done + I made
some correction using Pillot's program.
- programs to look at all the f_mt's versus IARS ( Integrated ARS
pulses : equivalent to ADC ) : basic functions done.
More on the efficiency loss
in the ARS :
Sumary :
- Electrons are recognized as pions (up to 50% depending on
where I look at) and pions recognized as electrons by the ARS.
- Strong correlations with the Cer_trig_tdc ( which define Mult1 or 2,
I m not sure ) .
- Time evolution : NA octants seemed to be fine before and start to
look like french octants now for the ARS.
Figures 2.2 - on figure a) : spring run for octant6 (32ns)
figure b) : spring run for octant7 (32ns)
figure c) : July - Aug run for octant6 (32ns)
figure d) : July - Aug run for octant7 (32ns)
Here I put some cuts on each cer_tdc X for each pmt X I ploted . That
mean that I require a Cerenkov trigger on each pmt that I look at.
And I plot the sum of 2 IARS in order to have good separation.
a) __________________
b) _____________________
c)____________________
d)____________________ Figures 2.3 - Correlation with Cer_trig_tdc :
