z trigger

A fast $z$ trigger is important for discriminating charged tracks produced at the interaction point from background tracks from interactions of spent electrons with the material around the beam pipe, beam gas events, and cosmic rays. The $z$ trigger was designed by utilizing the $z$-position information from the three cathode layers and the axial/stereo layers of CDC [85]. The cathode hit information provides direct information on the $z$-coordinate close to the interaction point. The accuracy of the cathode $z$-position is determined by the cathode-strip width of 8 mm and the cathode-strip-cluster size which is typically from 1 to 3 strips for a normally incident track. Since these cathode layers are located at the innermost CDC region, the additional $z$ information provided by the stereo wires is necessary for good trigger performance. Fig. shows the trigger tower map One axial/stereo z-trigger layer consists of consecutive two axial and two stereo layers. Each pair of neighboring axial and stereo wires yields a calculated $z$-position. The coincidence of two axial or stereo layers is formed to reduce accidental trigger signals due to uncorrelated noise hits. The accuracy of calculated $z$-coordinates is around 50 cm for a single pair of axial stereo cell hits.

Figure: Trigger tower map. The $z$ positions of each layer are calculated in 8 azimuthal segments individually. When a track passes through CDC, as shown in the upper figure, it yields the trigger tower map shown in the lower figure. If the $z$ positions of the middle and outer layers of the cathode line up with the same $\theta$ value, it is regarded as being a track from the interaction point.

The $z$-position is calculated in 8 azimuthal $\phi$ segments individually. In order to avoid the inefficiency for tracks around the $\phi$ segment boundary, the calculated $z$ signals of adjacent $\phi$ segments are ORed. The tracks in the $r$-$z$ plane ($z$-tracks) are reconstructed by seven sets of $z$-positions as shown in Fig. . A pattern yielded by tracks is represented by a "trigger tower map." When the $z$-positions in the inner and outer layers line up with the same tower bit, which presents the same polar angle from the $z$-axis ($\theta$), it is regarded as being a track from the interaction point. In order to reduce the effects of inefficiencies of CDC wire and cathode hits, we require at least two hit layers among the three cathode layers and at least one hit layer in each of the middle and outer layers with the logic shown in Fig. .

The schematic diagram of the $z$-trigger logic is shown in Fig. . There are five processes for the wire logic and four for the cathode logic. The number of $z$-tracks formed by the trigger tower process is counted in the final decision process. Finally, a two-bit signal corresponding to the number of $z$ tracks originating from the interaction point is transmitted to GDL.

Figure: Schematic diagram of the $z$-trigger algorithm.

The $z$-trigger system is constructed with three types of modules. The detail of hardware implementation is described in Ref. [85]. The main trigger module is a 9U-VME containing 7 Xilinx FPGA chips (XC4005HPG223) with 384 I/O's. All $z$-trigger logics are implemented in the FPGA chips using 53 of these modules. The $z$-trigger processes are run in a pipeline mode synchronized with a 16 MHz clock in order to avoid any deadtime losses. The trigger signals from the $z$-calculation and trigger-tower logic are read by TDC to monitor the $z$-trigger logic. The $z$-trigger logic run in a pipelined mode with 11 steps synchronized with a 16 MHz clock requires 687.5 ns. Including the drift time in CDC and propagation delay in the cables, the maximum latency of the $z$-trigger system was found to be about 1.45 $\mu$s.

Figure shows the trigger efficiency as a function of the distance from the IP ($dz$) for single tracks with 500 MeV/c (left figure) and for events with the "at least one $z$-track" condition (right figure) for cosmic rays. The tracks originating from cm are effectively rejected. The efficiency is greater than 98 % for single tracks with 300 MeV/c.

Figure: Trigger efficiency as a function of $dz$ for a single track (left) and an event (right) with the " at least one $z$-track" condition. The dots are the efficiency for cosmic data of the $z$-trigger logic, and the histogram is the efficiency for the reconstructed logic using CDC hit data as the input.

In order to keep the efficiency high, we require at least one $z$-track in the actual trigger condition. Because of quite high hit rates at the innermost layers of the CDC due to the beam background, the rejection power of the $z$-trigger is reduced. Fig.  shows the $z$ distribution of the track for two-track events with and without the $z$-trigger requirement, in which we require at least one $z$-track. The $z$-trigger reduces about one-third of background events without losing beam interaction events.

Figure: $z$ distribution of two-track events with (single-hatched histogram) and without (blank histogram) $z$-trigger requirements. The cross-hatched histogram shows the $z$ distribution of rejected events.