The basic philosophy underlying the trigger design is that the hardware must be mounted on the spectrometer, for the fastest possible trigger decision, but it should also be remotely controllable and programmable from the counting house, to optimize flexibility. These considerations have driven a design which utilizes currently available ``emitter-coupled-logic'' (ECL) modules with remotely programmable parameters. The trigger was further designed to run in one spectrometer in singles mode, or both spectrometers in coincidence mode.
In this document, we will divide the trigger system into three (intertwined) subsystems: the physical hardware of the trigger (crates, electronic modules and cables), the logical arrangement of the trigger (crate/slot assignments and wiring) and finally the software control of the trigger. The hardware will dominate our discussion of safety; not only the safety of the user, but also how to safe guard the hardware. The logical arrangement can, in principle be modified. However since it is presently working well we expect it to be stable. We will record here its present (March 1997) configuration. The software which controls the trigger is called XTrigMang. A detailed manual for the use on XTrigMang is available online (ftp://ftp.cebaf.gov/pub/halla/doc/TrigMang.ps). Nevertheless, we will also include a basic manual for operating the software in this document.
The hardware of the trigger system consist of two nearly identical set of electronics mounted in the detector package of each spectrometer, as well as three cables which run between the spectrometers for coincidence logic and re-timing. Each spectrometer contains three CAMAC crates and part of a VME crate which houses the Trigger Supervisor, the interface between the event and ``CODA'' (see CODA manual for details), and the crate controller (the interface between the crates and the either-net). The CAMAC crates are stuffed full of discriminators (LeCroy 4413), delays (LeCroy 4518), logic gates (LeCroy 4516) and memory-lookup-units -- MLUs (LeCroy 2373).
The trigger will simultaneously make a decision in each spectrometer based on local information. If it is running in coincidence mode it will then essentially ``and'' these two signals to produce the coincidence trigger. A simplified overview is sketched in the figure below.
The details of the electron-arm singles trigger, which can be used by itself or feed into the coincidence trigger, is sketched below.
Each module has a logical name, such as s12e_l_disc , which is the discriminator for the left side of the scintillators (layer 1 and 2) in the electron-arm. The logical name is the name a user will generally need when operating the trigger.
One of the important considerations for the user of the trigger is getting signals from the electron-arm and the hadron-arm to arrive in the coincidence logic at essentially the same time. A slow hadron, such as a 40 MeV proton, could arrive as much as 200 ns after the electron, which is essentially traveling at the speed of light. On the other-hand, Hall A can also perform experiments with kinematics in which the proton can travel at roughly 0.98c, ie. with essentially no time difference.
The way in which the trigger synchronizes these signals is through a series of delays which can be increased or decreased as need be. It is important that all the delays from the hadron arm through the ``Trigger Supervisor'' to the ``Fastbus Transition Module'' (FTM) in the Hadron arm should add up to match the 380 ns fixed delay (top of figure 3). Likewise the delays from the electron arm, through the two 237 ns transit times between the two spectrometers, and through the programmable delays, the ``Trigger Supervisor'' to the ``Fastbus Transition Module'' (FTM) in the Electron arm should add up to match the 680 ns fixed delay. However, all of these delays should be calculated and modified in the trigger settings files before the run starts.
The details of the coincidence trigger are sketched below.