[Hallc_sidis_analysis] EDTM correction, exclusive model, electronic dead time

[bosted at jlab.org]

Hi all,
   Thanks to Garth mentioning that online checks during KLT
showed that param3000 for exclusive pi+ seemed to work quite
well, I went back to my code and found an error: xptar and
yptar were reversed in SIMC for the exclusive files (but not
the SIDIS files). When I finally fixed this problem, I found
that param3000 works well (within 15%) for most runs, and in
the worst case is about 25% low. The old param04 fit is too
large by factors of 1.5 to 3. So for now, I suggest to stick
with param3000. I plan to continue working on the pi- channel
as well as the fermi-smearing model for aluminum.
   Here is how I calculate the live time:
       rmeas = float(edtmtrigaccptd) / float(edtmtrig)
       f = current / curcut
       newlt = (rmeas - (1. -f)) / f
The number of accepted and generated EDTM events come from
the report file, as does the ratio of current with a 5 muA
cut to average current without the cut. I use BCM1 but
it doesn't matter really which one you use.
Basically, the assumption is that the beam is either on or
off, and that the live time when the beam off is 100%.
Then with algebra you get the above formula.
   I should mention that there is an extra amount of electronic
dead time that is not included in the above. This is due to
the fact that  the reference time TDC (used to get the
"startime" from the hodoscopes) had two inputs: 3/4 scint
(which is the one we want), followed by elreal, which involves
the TOF (2/4 scint) signal plus a Cerenkov signal (pretty
much always fires for P>3 gev) and a small shower counter
signal, coming 200 nsec later. For P>3 GeV, we very often
see this double pulse in the ref. time.
So, if there is a random trigger
coming of order 50 to 200 nsec earlier than the coin trigger
that we are interested in, the 50 nsec wide (?)
that elreal pulse blocks the TDC
from getting the 3/4 signal that we are interested in. The
extra dead time should be proportional to
rate_el_real * x / 1.e9, where x is something of order of
magnitude 50 to 100 nsec. Over runs 3400 to 8300, the SHMS elreal
rate was typically 100 kHz, but sometimes as big as 500 kHz.
So if x=100, el-real rate=500kHz,
the dead time would be 5e5 *100 / e9 = 0.05
or 5%, which is quite significant! I will try to find out
what the effective optimal value of x is. What makes it
even more complicated is that when the true ref. time is
blocked, the software will usually pick either the 3/4 or
the elreal time from the earlier random, which in turn
makes the cointime bigger than it should be, which means
we have moved events out of the cointime peak over to one
side, and may then subtract it from the true signal. This
part depends on the realaccidental ratio. It could in
principle be measured by an asymmetry n the accidental rate
on one side of the cointime peak compared to the other. This
is complicated by the protons and kaons in positive polarity,
but more clear in the negative polarity runs. More on this
later, I hope.
   Peter


> Hi All,
> We will have the regular CSV analysis meeting tomorrow (Friday) at 1:00 PM
> EDT
>
> The Meeting URL is
> https://urldefense.proofpoint.com/v2/url?u=https-3A__bluejeans.com_341030850&d=DwIBaQ&c=CJqEzB1piLOyyvZjb8YUQw&r=6UVi9h7XDe4YpNVl1Pr7kJfw2N_A9rNcjJ8jEvLZw2M&m=kHxCl_AVIYz13AqZ193UNjcv0OEj-w2N2Vx63Vz0Tc4&s=wyoqD4UI7Hnh-oHH69Zqiqf29RrCwmnK2iZJ5-Eanxw&e=
>
> Please plan on showing a few slides on your work and please upload the
> slides to the SIDIS elog at:
> https://hallcweb.jlab.org/elogs/SIDIS+Pt+and+CSV+Experiments+Analysis
>
>
> Cheers
> Dipangkar
> _______________________________________________
> Hallc_sidis_analysis mailing list
> Hallc_sidis_analysis at jlab.org
> https://mailman.jlab.org/mailman/listinfo/hallc_sidis_analysis
>