Re: BCAL Threshold -- Input Needed

[Elton Smith <elton@jlab.org>]

Hi Blake,

I meant to get back to you with a few comments. Your analysis is correct
as long as electronic noise plays no role. When the dark rate noise is so
low that it does not impact the width of the pedestal, it is likely that
electronic noise will be the dominant contribution. The width of the
pedestal is hard to estimate from first principles, but the beam test data
can give you an empirical value for it. Since we used vacuum pmts for
the beam test, the dark rate was presumably negligible in that case
too. So I suggest you take a look at the pedestal widths from the beam
test an consider something like 3-4sigma as an estimate for the threshold.

Cheers, Elton.

Elton Smith
Jefferson Lab MS 12H5
12000 Jefferson Ave
Suite # 16
Newport News, VA 23606
elton@jlab.org
(757) 269-7625
(757) 269-6331 fax

On Mon, 18 Aug 2008, Blake Leverington wrote:

> Hi all,
>
> I'm trying to come up with a number for the the threshold in the case of
> a 3x3 geometry with FM PMTs as readout.  Since we do not have the dark
> rate (less than 1 kHz) it seems to me that the threshold for the
> detector is then only the energy needed to generate one photoelectron or
> about 0.61 MeV. Does this seem reasonable or is there other noise I am
> not taking into account?
>
> -Blake
>
> Matthew Shepherd wrote:
> >
> > Hi all,
> >
> > Blake and I are now working on retuning the BCAL MC.  As has been
> > noted many times in the past, the BCAL cell detection threshold is an
> > *absolutely critical* input to modeling and performance.  It will
> > "make or break" the the reconstruction in the < 200 MeV range.
> >
> > Since it seems to be breaking more than making with our latest round
> > of revisions I wanted to go through how we determine this once more.
> > If some of you could give a little thought to this and see if it is
> > reasonable, that would be great.  There are some assumptions made;
> > however, we are really just trying to get within +- 30% or so of a
> > reasonable threshold number.
> >
> > Assumption 1:  One will need to do some zero suppression on a per
> > channel basis on the FADC.  Let's assume that the pedestal alone
> > cannot cause more than 5% of all BCAL channels to exceed the zero
> > suppression threshold.
> >
> > Assumption 2:  The main contributor to upward fluctuations in the
> > pedestal is single PE dark pulses from the SiPM.  This assumption is
> > optimistic since there will undoubtedly also be electronics noise that
> > will broaden the pedestal also.  The rate of single PE dark pulses is
> > 40 MHz.
> >
> > Assumption 3:  The readout window for the FADC is 100 ns.  This means
> > SiPM pulses must be fully contained in this window including any
> > forward backward shifts due to differences in transit time in the
> > readout chain (hopefully minimized) and transit in the module (about
> > 25 ns for full length).
> >
> > With a 100 ns readout window this implies there will be an average of
> > 4 single PE pulses in every event.  According to Poisson statistics
> > this means that 95% of events will have 7 or less pulses.  Therefore
> > we set the threshold at 8 photoelectrons.
> >
> > Now we need to make a careful translation between photoelectrons and
> > energy deposited in a cell.  For this we use the latest and greatest
> > GlueX-doc 1069.  The first bullet point at the top of p.9 gives
> > results for SiPM array.  Scaling for the non-operational cells in the
> > sensor one estimates 11 photoelectrons per MeV deposited in the fiber,
> > when attenuation effects are removed.  That means 1 PE = 91 keV
> > deposited in the fiber.  The sampling fraction for cosmic rays is
> > about 15% -- therefore 1 PE = 91 keV / 0.15 = 0.61 MeV deposited in
> > the module.
> >
> > The simulation "attenuates" the energy deposited in the cell
> > (fiber+lead+epoxy) down to each end.  At that stage the hits are only
> > saved if they exceed a threshold of 8 PE * 0.61 MeV / PE = 4.9 MeV.
> >
> > Do we agree that this seems reasonable?  One might hope that signal
> > processing in the FADC could suppress some of the 1 PE pulses.  This
> > would effectively lower the 1 PE rate and allow a lower threshold.
> > However, I'm sure overlooking things like electronics noise in this
> > estimate, make the estimate somewhat optimistic.  As I said we would
> > like to get as close as reasonable.
> >
> > We are forging ahead with this threshold and will generate new plots
> > for each of the two segmentations.
> >
> > -Matt
> >
> >
>