Results

Three separate Monte Carlo simulations were run for this study. Run 1998 was done under the conditions of the experiment during parasitic running the summer of 1998 at 4GeV end point energy. These numbers may be directly compared with the rates measured during that period, and provide a consistency check of the simulation. Run 1999a was done under the conditions expected for summer 1999 parasitic running at 5.5GeV end point energy. Run 1999b was done under similar conditions as 1999a, except that the Radphi target was moved 2m downstream from its nominal position. This places the target beyond the front face of the glass, and so provides an estimate of the rates component that does not come from the target. The target was not removed entirely for run 1999b because the flux at the face of the target is used in the simulation to normalize the bremsstrahlung flux in lieu of simulating the tagger. The major results of these runs are shown in Table 1. For each run two numbers are given for the rates in each detector: the first is the total number of hits and the second is the total number of events with at least one hit. The former corresponds to the sum of scalers connected to each channel and the latter to the scaler counting the logical or for the entire detector.

The UPV is a plane of scintillators located upstream of the target that serves to veto beam halo events. The CPV is a plane of scintillators located between the target and the forward calorimeter which serves to veto any events containing forward-going charged tracks. Both the CPV and UPV have a rectangular hole in the center where the beam passes. The barrel scintillator detector [BSD] consists of three concentric cylinders of 5mm thick scintillator paddles. The three layers are labelled in Table 1 as layers 1 (inner) 2 (middle) and 3 (outer). The data in Table 1 show several important things regarding the background.

1.

Increasing the beam end point energy from 4GeV to 5.5GeV has very little affect on the CPV rate. It will run in 1999 at substantially the same rate as it did in 1998, for comparable beam intensity.

2.

The background is largely low-energy particles. This can be seen from the fact that the outer layers of the BSD are effectively shielded by the material of those inside, even though the layers are composed of only 5mm of plastic scintillator.

3.

Roughly 90% of the background within the detector acceptance is coming from the target. This can be seen by comparing column 2 (target in) with column 3 (target out). The UPV is located upstream of the target, and so its rates are largely unaffected when the target is removed, but the CPV and BSD rates go down by about a factor of 10.

4.

The residual background in the detector when the target is removed is coming from the lead collimator, and has a harder energy spectrum than the target-related background. This can be seen from the fact that rates are roughly the same in all three BSD layers when the target is removed. That these particles are coming from the region of the collimator was seen by examining events during the simulation.

5.

With the UPV moved upstream in the 1999 runs relative to its position during 1998, its rate is substantially decreased. It still has a role even in its far-upstream position, however, because about half of the 640 halo tracks that fire the BSD triple coincidence also hit the UPV. In the following discussion I will show that it will be possible to reduce the BSD triple-coincidence rate down to a level where the UPV veto may be useful.

Table 2 compares the calculated rates from run 1998 with those measured during summer 1998 parasitic running. The relative normalization was fixed by scaling both measured rates and Monte Carlo to a common beam intensity of $10^7$ tagged $\gamma$/s. This value corresponds to the tagger-OR, not the rate of tagged photons that hit the Radphi target per second. The rates in the CPV and UPV can be directly compared with the Monte Carlo. The Monte Carlo did not include a description of the RPD, but a rough comparison can be made between the measured RPD rate and the Monte Carlo rate in the barrel scintillators between the polar angles of $40^o$ and $60^o$. This value is shown in parentheses in Table 2 because the equivalence to the RPD is only approximate. The comparison with the measured RPD rate is also affected by the fact that the discriminators on the thick (E) RPD counters were set somewhat above the minimum ionizing peak. This was done to reduce the trigger rate from electromagnetic background while retaining good efficiency for the more heavily ionizing recoil protons. With these considerations taken into account, the Monte Carlo and measured RPD rates in Table 2 are compatible.

Table 1: Hit statistics from Monte Carlo simulation. Run 1998 simulates conditions during parasitic running in the summer of 1998 at 4GeV end point energy; run 1999a simulates conditions expected during parasitic running in the summer of 1999 at 5.5GeV end point with the Radphi target in place, and 1999b the same but with the target effectively removed.

		1998						1999a						1999b
total simulated		10.6		M						10.6		M						10.6		M
total hit target		8.00		M						8.89		M						8.65		M
total tagged		172		K						167		K						167		K
tagged hit target		142		K						149		K						145		K
UPV rate, or		6850				6530				1660				1360				1750				1380
CPV rate, or		238		K		182		K		234		K		178		K		16.7		K		13.7		K
BSD rate, or for layer 1										29.2		K		26.1		K		1830				1290
BSD rate, or for layer 2										15.2		K		13.7		K		1700				1230
BSD rate, or for layer 3										9210				8140				1710				1190
BSD rate for triple coinc.										7370								640

Table 2: Comparison between rates in the trigger counters predicted by Monte Carlo run 1998 with measured rates from summer 1998 parasitic running. Both data and Monte Carlo have been normalized to 10 MHz in the tagger range from 78% to 95% of the bremsstrahlung spectrum. The data have been corrected for electronics dead time.

		Monte Carlo			measurement
UPV or			380		K			720		K
CPV or			11		M			14		M
RPD or			(150		K)			75		K

The measured rate of the UPV or shown in Table 2 is an average value over the 1998 run period. It varied from the value shown by a factor of about 2, depending on the tune of the machine. The value shown in Table 2 as measurement represents an effective average over periods of stable operation. Because it is located upstream of the Radphi target, the UPV is more sensitive to the quality of the beam than are the other elements of the detector. The difference between measured and simulated rates in the UPV might be explained by non-ideal alignment and transverse profile of the electron beam. The simulation assumes ideal alignment and zero emittance for the electron beam. The Monte Carlo value should be taken as a lower bound for the UPV (under 1998 conditions) while the CLAS target is full.

The CPV or provides the most rigorous test of the Monte Carlo. The thresholds for these counters were set low to obtain good efficiency for all charged particles. In the Monte Carlo energy loss spectrum, the minimum ionizing edge is readily seen and summing the count rate is not ambiguous. The rate is dominated by soft $e^+/e^-$ tracks coming from beam interactions in the target, and so is not very sensitive to effects related to beam quality of the kind discussed above for the UPV. I would expect the simulation and measured rates to agree within about 5%. Small variations in the Monte Carlo CPV rate at the level of a few percent are obtained by adjusting the parameters of the simulation (changing the low-energy cutoff in the tracking, generating bremsstrahlung all the way down to the KeV regime, enabling discrete delta-ray production) but at the 5% level the Monte Carlo prediction appears solid.

If, instead of taking the CPV or rate from Table 1 to compare with the measurement, we take the total CPV rate then the agreement is excellent. The difference between the total rate and the or rate is that the total rate counts all hits separately while the or only counts the number of events with at least one hit. On paper, the electronics that generated the CPV or during 1998 running should count only once no matter how many tracks there are, provided that all of the hits are within 5ns of each other. The simulation shows essentially all hits associated with a single beam photon as coming within 5ns of each other. The signals from individual CPV paddles were synchronized to within 3ns of each other, and dispersion from light propagation delays down the paddles is not a large enough effect to explain the difference. Double-pulsing in the CPV discriminators might have been responsible, but that will have to be checked during the next run.

For the purposes of this report it suffices that the numbers from the simulation agree to within 20% with those that were measured during 1998 parasitic running. Based upon that we are able to forecast the trigger rates for 1999 with sufficient accuracy to show that the level 1 rate with the upgraded detector will be the same or even lower than was obtained with the RPD, for more than twice the acceptance. With the trigger and data acquisition improvements in place this year, we expect to demonstrate efficient operation of Radphi at the design intensity during the summer 1999 g6 run period.