Comments on "requirements for start counter"

[Elton Smith <elton@jlab.org>]

Comments on "requirements for start counter"
--------------------------------------------

Here is my assessment of what we "decided" regarding the function of the
start counter. My guess is that there may be different conclusions about
the discussion even though collectively people thought we had arrived at a
"consensus". I wanted to put some notes down before the discussion
evaporated from my own head. Here is what I took away from the discussion:

1. The main function of the start counter is to serve in the first level
trigger, with timing "jitter" of about 5 ns, to help in reducing the
trigger rates during initial stages of low intensity operation for GlueX.

2. In software, the start counter would provide sub-ns (sigma ~ 300 ps)
timing resolution by using hit based tracking information. This corrected
time would be a very good determination of the vertex time to be used in
further event reconstruction.

Discussion:

Two general concepts are available: a) single cylinder with double-sided
readout, presented by Werner, b) cylinder with cone and single-ended
readout (Hall B design). My understanding is that concept b) was chosen to
be presented at the detector review, implemented with Bfield-tolerant
pmts. There is no simple way of having two-sided readout in this design,
and therefore timing of the detector which is smaller than the transit
times in the scintillator is not possible. [Aside: clearly studying
options of providing two-sided readout usign, for example SiPMs is
encouraged, but should not be considered a requirement.]

In Hall B, we have generally not relied on the start counter to produce
sub-ns resolution in the offline. [Up until the latest running period, the
start counter was segmented into three paddles.] The reason for this is
that accidentals are a serious issue there. Estimates of the accidental
probability of finding a hit in the start counter if there is a track in
one sector is approximately ~ 2 MHz x 100 ns = 20%. However, for Hall D
the situation is improved due to the lower background rates and
segmentation, giving accidental probability ~ 20 kHz x 100 ns = 0.2 %.

This implies that one should be able to successfully associate a hit-based
track in the CDC (FDC) with a hit in one of the start counter segments
with a resolution of a few cm. This is adequate to make a transit time
correction and obtain a sub-ns time from the start counter.

Note that questions about the details of start counter design and
operation can be provided by Youri Sharabian and Valeri Koubarovski.

I hope my understanding of the discussion is shared by (at least a
majority) of everyone else.

Cheers, Elton.


Elton Smith
Jefferson Lab
elton@jlab.org
(757) 269-7625