The choice of photomultiplier tube is motivated by the desire to find
a fast tube with good gain and and minimum time jitter. The area to
be readout sets the scale for the size of the photocathode. The cross
sectional area of the edge of the inner scintillator tile is 6.35
cm2 which suggests a 28 mm (1-1/8 inch) diameter head-on type
photomultiplier. A large variety exist at this size built by a number
of commercial vendors. At the time of this submission, the final tube
choice remains open and will be subject to an open bid process in the
event that several PMTs meet the required specifications. However, it
is useful to describe an affordable PMT which we are presently using
for the 2 experiment which has passed many critical timing and gain
tests related to our similar needs there. In 2 the average rate
changes from early-to-late times by 105, and the initial rate can
exceed a few MHz. In the µLan detector, the peak rate is nearly
100 times lower. Furthermore, in 2, an intense initial burst of
particles due to pion contamination in the beam creates such a large
light output in the detectors that the tubes have to be gated off for
up to 20 
s. This environment is particularly hostile for good
timing and gain stability; our PMTs and custom bases have thus been
subjected to a rather severe test and have performed quite
well [21]. In this development, we have created test
systems at BU and Illinois to measure rate and gain dependence of PMTs
using LEDs and lasers.
An excellent candidate PMT presently in use by us is the Hamamatsu R6427. It is a 28 mm diameter 10-stage head-on tube with a gain exceeding 5 x 106 and a maximum quantum efficiency of approximately 25%. The 10-90% risetime is 1.7 ns and the transit time jitter for single photoelectrons has a standard deviation of about 200 ps. We have designed a base for this tube with a gating circuit. It will not be difficult to produce simpler ungated bases following this design, or to use the one offered by Hamamatsu as a package which already matches the tube. The cost of the tube, base and magnetic shielding which we have budgeted is $500 per unit. Over the next year we will test other tubes in order to find the best final product. Our main goals are fast risetime, low jitter, and short-term stability (burst mode) for the timing and gain, all within a moderate cost envelope.
The individual equilateral scintillator tiles of 6.35 mm and 3.175 mm
thicknesses can be obtained either as large sheets from which we will
do the cutting and polishing, or cut to form. In either case, the
bulk cost of the tile represents a minor cost of the detector. We
will make a final selection between the two fast-timing choices from
Bicron [19], namely BC-404 and BC-418. Their main
properties, along with those of the standard BC-408 bulk scintillator,
are compared in table 2. The BC-418 has the shortest
decay constant which we want for the best pulse-to-pulse separation.