next up previous contents
Next: Report on Recent Up: Computers Previous: Computers

New interferometer control

With the construction of the 5th antenna, and after our evolution towards the use of VME and Ethernet to control the correlators and receivers, we have redesigned the interferometer control and especially the antenna control system.

The new antenna control system is based on a hierarchy of VME microprocessors and microcontrollers. The VME microprocessors may communicate with each other or with the UNIX station (bure01) in charge of monitoring, driving the operations but also of collecting the data. The task to task communications are based on TCP/IP and use Ethernet links, optical or coaxial cables.

Two kinds of data transfer are needed: periodic and heavy traffic from the correlator micros to the UNIX station, and asynchronous and medium traffic for antenna and receiver control. These transfers have characteristics so opposed that we foresaw to use two different subnetworks. The subnet dedicated to the correlators is attached to bure01 only. The antenna and receiver microprocessors are on the general subnet, on which one finds also the second UNIX station (bure02), foreseen for near real time data reduction, and the router for communications to/from other sites.

In each antenna there are 2 VME microprocessors, one in the receiver cabin, in charge of the receivers and the subreflector, and the second in the pedestal, mainly in charge of driving the Az and El axes while avoiding the sun. In the antennas, the housekeeping, including the change of station, the survival configuration and the de-icing, are performed by microcontrollers which are linked to bure01 via the VME microprocessors.

The distribution of tasks has been largely modified, since VME micros based on Motorola 68030 chips operating at 25MHz can perform all coordinate transformations and sun tests for an Alt-Az mount. The central UNIX station (bure01) is still in charge, for each new observed source, of computing the true equatorial coordinates, taking into account nutation, precession and all possible additional corrections. These calculations should be repeated every hour for long observations. Let us note here that all this computing does not require a very precise knowledge of the universal time (UT).

The VME microprocessors (in the pedestals) calculate the azimuth and the elevation. This transformation from equatorial to horizontal coordinates needs the UT, which is set by reading a universal time server, and synchronized via a time bus. This time bus is distributed all over the site; the antenna control uses only its one second pulse. The universal time server on the site is a microprocessor (named `clock'), located in the computer room, also connected to the one second pulse, and which can read the time from our UT master clock, in seconds integral or in fraction of seconds.

At boot time the ``antenna control'' VME micros, on the first occurrence of a one second pulse, request from the time server (clock) the time, in integral seconds, corresponding to that pulse, and check that the answer is received well before the next one second pulse occurs. Besides, the ``antenna control'' VME micros obtain time with higher resolution using interrupts at 64Hz, incrementing a counter, which is resynchronised every 1s on the time bus pulse.

As stated before, the equatorial coordinates sent by bure01 to the ``antenna control'' VME micros vary only slowly with time; just like offsets, pointing parameters and homology coefficients, they need not be applied on a pre-determined second. However, drift scans require a synchronisation between bure01 and the ``antenna control'' VME micros. To solve this specific problem, periodic operations on bure01 are synchronised on the reception of periodic (synchro) messages issued by the time server VME micro (clock). The period of these synchro messages is one second and each message is dated. This time is used to stamp each request sent to the ``antenna control'' VME micros and triggered on the synchro messages.

The control of the subreflectors is performed in the ``receiver control'' VME micros, since these micros are located close to the subreflector motor amplifiers. These micros monitor and control the motors upon requests from either a local test program or a task communicating with the companion ``antenna control'' VME micro, which provides the motor positions dependent on the current antenna elevation, the homology parameters and extra subreflector tilts and translations, as requested.

The tests so far have been performed with one antenna disconnected from our current control system (CAMAC based). The antenna was fully equipped with VME modules, some of which were specifically developed at IRAM, such as the time bus interface, the incremental encoder and resolution enhancement interface, and the subreflector 5-motor interface. In observation mode, the commands concerning the antenna position, prepared on the VMS system for the current interferometer control system, were reformatted to the new command syntax and forwarded to an UNIX station. This station, synchronised on the universal one second pulse, executed the commands and transferred the resulting requests to the VME micro in the antenna. With the subreflector and the main pointing axes operating in this way, and pointing to quasars and planets, we could get interferometric correlations with the other antennas.


next up previous contents
Next: Report on Recent Up: Computers Previous: Computers

Robert Lucas
Mon Sep 18 09:50:01 METDST 1995