A time of flight (TOF) development effort has been initiated to
assess the performance of various detector components planned for the E864
TOF hodoscopes. The immediate goal of these efforts has been to select a
scintillator - photomultiplier combination which will deliver the desired
timing resolution at a reasonable cost. Three types of plastic scintillator
and two types of photomultiplier tubes (PMT's) were evaluated in these tests.
A few characteristics of these scintillators and PMT's are summarized in
Tables 
and .
The range of sizes and types of scintillators studied are
listed in Table .
Timing resolution was studied as function of scintillator
size and type, PMT high voltage, dynode voltage ratio, discriminator threshold
and source position.
Tests were performed using either cosmic ray muons or a 
source, which
yields 1.0 MeV mono-energetic electrons. A schematic drawing of the test
apparatus is shown in Fig. . As a trigger, a triple
coincidence was formed between TRIGA, PMTA and PMTB. The trigger logic was such
that the time-to- amplitude converters (TAC's) for PMTA and PMTB received a
common start initiated by TRIGA. Each TAC was stopped by the respective logic
signal generated by the discriminated PMTA and PMTB pulses. TRIGB could not be
included in the trigger for the source tests since the 1.0 MeV
electrons are stopped in the test scintillator. In the tests using cosmic rays,
the trigger rate was sufficiently small that any TRIGB requirements could be
made off-line if desired.
Table: Summary of test scintillator specifications.
Table: Summary of test PMT specifications.
Table: Sizes and types of scintillators tested.
Figure: Schematic of the experimental apparatus
used in the TOF tests.
To determine the timing resolution of the PMT-scintillator combinations, time spectra were gathered corresponding to the time differences between TRIGA and PMTA and PMTB. The measured time interval for each channel, TOFA and TOFB, was corrected for slewing based on the ADC information recorded for each signal. The slewing corrections were determined by fitting the raw spectra to the functional form:
where Q is the integrated charge recorded by the ADC.
Figure
shows an example of the raw and slew-corrected time spectra obtained. Only
the difference between the actual recorded time interval and 
is plotted. As can be seen, this procedure largely removes any dependence
of the recorded time on pulse height.
Figure: Recorded time (ns) versus pulse height (arbitrary units) for (top)
raw and (bottom) slew-corrected time spectra. Also shown is the projection
of the data onto the time axis for each plot.
The width of the TOFA and TOFB distributions include
contributions from timing jitter in the start counter as well as jitter in the
PMTA and PMTB channels themselves. For these tests, the timing resolution of
the start counter TRIGA was rather poor, being in the neighborhood of 140 
160 ps. This was due in part to the PMT used for TRIGA. To exclude the
contribution of the start counter to the measured widths of the time spectra,
the time difference, DELTOF, between TOFA and TOFB was used to quantify the
timing performance in each test. Assuming that the time jitter in each channel
is statistically independent and that the timing resolution for PMTA and PMTB
are equal, the intrinsic time jitter of either channel should be smaller than
that of DELTOF by a factor 
. Independent of the assumption of equal
time resolution for each channel, the jitter on the mean time, defined as
, will be one half that of DELTOF.
Proceeding in this manner, a number of tests were performed on various
scintillator - PMT combinations. Unless otherwise noted, the numbers given
refer to the implied mean time resolution (= 
)
for the hodoscope
element as determined by the measured DELTOF resolution.
Figure shows the relative timing performance as a function of
operating voltage of the two photomultiplier tubes considered. Since the two
PMT's have different maximum operating potentials (-1500 V and -1800 V for the
R1635 and R3478, respectively), they are compared as a function of the fraction
of their recommended maximum operating voltage. As can be seen, the timing
resolution obtained using the R3478 is better by approximately a factor of 1.4.
A study of the timing characteristics of the R3478 was made as a function of the
magnitude of the applied high voltage and discriminator threshold. These
results are shown in Fig. . For the 5L1-404 scintillator
tested, the timing resolution appears to be rather insensitive to the operating
voltage over the range examined, but degrades by about 15%
when the magnitude of
the threshold was changed from 25 to 100 mV.
Figure: Relative timing characteristics of the
Hamamatsu R1635 and R3478
photomultiplier tubes using the 5L1-408 scintillator and a source.
Figure: Relative timing characteristics of the Hamamatsu R3478
photomultiplier tube as a function of (a) operating voltage (discriminator =
-25 mV) and (b) discriminator threshold (HV = -1400 V). For this test, the
5L1-404 scintillator was used with a cosmic ray source.
The behavior of the timing resolution of the R3478 was further investigated as
a function of the voltage ratio applied to the dynodes of the PMT. To obtain
the best timing resolution, the voltage difference between the photocathode and
the first dynode is typically made several times the voltage difference between
subsequent dynodes to reduce jitter in the electron transit time. In an attempt
to optimize the timing performance of the R3478, the resolution was studied as
a function of the voltage difference between the photocathode and the first
dynode, for a fixed operating potential. This was achieved by varying the
resistance of the first stage in the voltage divider of the PMT base.
Figure shows the behavior of the time resolution measured as a function of this
resistance. Based on these measurements, it appears that the factory
recommended voltage distribution ratio of 7:1:1.5:1:1:1:1:1 (corresponding to
a cathode - first dynode resistor value of 1.68 M 
) is nearly optimal for our
timing purposes.
Figure: Relative timing characteristics of the Hamamatsu R3478
photomultiplier tube as a function of the photocathode - first dynode
resistance (HV = -1400 V, discriminator = -25 mV). For this test, the 5L3-404
scintillator was used with a source.
Having clearly established the photomultiplier of choice, a number of tests
were performed to investigate the relative timing properties of the various
scintillator sizes and types under consideration. The mean time resolution
determined for each scintillator size and type tested with cosmic rays is
listed in Table . A subset of these results is shown in
Fig. .
Table: Mean time resolution (RMS) for various sizes and types of
scintillators tested with cosmic rays.
Figure: Mean time resolution as a function
of length of the test scintillator.
Finally, a study of detector timing resolution as a function of position along
the scintillator was performed. For this test, the trigger scintillators were
positioned at various distances measured from the PMTA end of the 5L3-404
scintillator. The timing resolution, shown in Fig. ,
demonstrated a very
slight (5 
3 %) degradation as the trigger scintillator position was moved
from PMTA to PMTB.
Figure: Mean time resolution as a function of trigger counter position
along the length of the test scintillator (measured from PMTA). For this test,
the 5L3-404 scintillator was used with a cosmic ray source.
As described above, our TOF tests demonstrate that for the 3 hodoscope
stations we should be able to achieve mean time
resolution for minimum ionizing particles in the neighborhood of 60 - 80 ps
using R3478 photomultiplier tubes and almost any of the scintillator types
considered. Assuming a timing resolution of 50 ps for the trigger counter
which will be used in the actual experiment, it is possible that we can build a
time-of-flight measurement system of the size and granularity required for
E864 with intrinsic time resolution in the neighborhood of 80 100 ps. Such
resolution will be degraded in the actual operating environment of the
detectors, e.g. due to dispersive losses in the signal cables, but is
still expected to be superior to the
200 ps resolution proposed for the experiment. The resolution will also be
affected by the required addition of light guides.
We plan to design and test appropriate light guide -
scintillator combinations in the near future but anticipate that such light
guides will not drastically degrade the timing abilities of the proposed TOF
hodoscope system.