A Method of Fast Data Transfer From FASTBUS

Nuclear engineering, Computational physics

Oussena, Baya

University:University of Glasgow

关键词: Nuclear engineering, Computational physics;   

Others  :  http://theses.gla.ac.uk/78373/1/11011440.pdf

来源: University of Glasgow

【 摘 要 】

One major factor which affects the counting efficiency of a nuclear physics experiment is the dead time of the detectors and the data acquisition system. Experiments performed by Glasgow University photonuclear group typically involve the readout of ~ 1000 ADC's and ~ 1000 scalers which contain information on the products of a photo-disintegration event. These require fast readout to minimise dead time and to this end a method of programming the model 1821 FASTBUS Segment Management Interface (SMI) to increase data throughput coming from FASTBUS has been developed. The electronic hardware used is comprised of VMDE-bus, CAMAC, and FASTBUS systems. The VME-based CPU is the heart of the data acquisition system. FASTBUS is mainly used for ADC's and TDC's while CAMAC is mostly used to control the experimental parameters such as detector thresholds, trigger logic, high voltage etc. Each FASTBUS crate is controlled by a LeCroy 1821 Segment Manager Interface (SMI), and the interfacing to the VME CPU is accomplished either by using the VME fast memory module type HSM8170 or the slower CAMAC interface type LeCroy 2891A. The HSM8170 is connected to the SMI using the 32-bit LeCroy ECL bus. The VME CPU runs the OS9 operating system, and the data acquisition software has been written almost entirely in C. Software for the sequencer in the 1821 SMI is written in machine code, although it is hoped in the future to develop a simple assembler. Two different SMI codes have been developed. These are called CODE1 and CODE2. The first attempt, CODE1, uses the slow, CAMAC connection at the front panel of the 1821 SMI for module initialisation and data readout. To improve the data throughput, it was decided to develop CODE2 which uses the rear panel ECL bus connection to a fast VME memory, and require no intervention from the VME host CPU to initiate data readout. Associated C routines written for the VME CPU handle downloading of the code to the SMI and create FASTBUS module addressing SMI instruction words. Finally, the performance of the two FASTBUS readout methods has been compared on a test setup where more than 100 ADC channels are read for each event. Under these conditions, the dead time for a CODE2 readout was found to be approximately a factor of 8 less the dead time for CODE1.