The ALMA Telescope Control System

The Atacama Large Millimeter Array (ALMA) is a joint project between North America, Europe and Japan. ALMA is an aperture synthesis radio telescope consisting of 50 12-meter antennas located at an elevation of 5,000 meters in Llano de Chajnantor, Chile. These antennas will operate at frequencies ranging from 31.3 GHz to 950 GHz. The antennas can be moved and placed in different configurations, with baselines between the antennas varying from 150 meters to 20 km. The 50 antennas are supplemented by sixteen additional ones, known as the ALMA Compact Array (ACA): 12 7-meter antennas and 4 12-meter antennas. The ALMA control system will consist of over 70 computers separated by distances of over 20 km. Two aspects of the system are apparent: its distributed nature and its need to accurately synchronize events across many computers separated by large distances. In this paper we describe key features of the architecture of the ALMA Control System, focusing on its properties as a distributed system and on the mechanisms employed to achieve its time synchronization goals. This control system is a distributed system that uses the ALMA Common Software (ACS) as a middleware system layered on top of CORBA. The architecture of the control system extensively employs the component/container model in ACS. In addition, the use of CORBA allows us to employ Java in the higher levels of the control system, leaving C++ to the lower time-critical levels. Python as a scripting language is used by astronomers, to craft standard observing programs, and engineers, in a testing and debugging mode. Key to the concept of an aperture synthesis telescope is a special purpose hardware system known as a correlator, responsible for making various delay model corrections and correlating the signals from the antennas. There are two correlators in ALMA, one for the array of 50 antennas and one for the ACA. This entire system operates under a control system that must synchronize events across the entire system to an accuracy of better than one microsecond in order to achieve its scientific objectives. Time synchronization is accomplished by sending precisely timed electronic signals, which are derived from a Maser, to all hardware that requires accurate time synchronization. The control system preloads time-critical commands and monitor requests that are triggered by these timing signals. Propagation delay times to remote hardware must be taken into account. The slowest of these timing signals, the 48ms timing event, is also connected to all real-time computers. This timing event is used for synchronizing time internal to the array to external time standards. It is part of a hierarchy of timing signals that, in some cases, measures time to subnanosecond accuracy.