Next Previous Contents

4. NCCS Layout

4.1 Physical Network Structure

The physical structure of the New Control Computer System is a network of dedicated machines, connected together on a single thin ethernet spine as an independant subnet of the HartRAO network. Currently this control network consists of the following:

with nccs1 and nccs2 operating as gateways with two network interfaces each, the second being connecting them to the main HartRAO network as well. Each runs a secondary nameserver service for the HartRAO domain and can service any nameserver requests on the control subnet. Both machines are also equipped with 16-port serial and HPIB interface cards to which most of the envisaged NCCS hardware will be connected. All power is supplied by a 230V UPS derived via a 110-230V transformer from the main 110V UPS.

See also the NCCS Layout for an earlier view.

4.2 Reference Time Distribution

The fundamental reference for timekeeping within HartRAO is the Truetime 900 Series Time Code Unit driven by the 5 MHz frequency reference "A" output of the OscilloQuartz EFOS 6 Hydrogen Maser (comprising the master HartRAO Station Clock.) This clock is routinely monitored against a Magellan OEM 5000 GPS receiver to track its behaviour relative to Universal Coordinated Time (UTC), the international fundamental reference for timekeeping. Power to all the clock units is supplied from the standard 110V UPS with a further battery-based DC backup supply to ensure the local fundamental reference will survive even UPS failure.

In order to provide accurate time to all (U)nix machines within the local HartRAO network, a Truetime NTS-100 Network Time Server has been slaved to the standard IRIG-B timecode output of the Truetime TCU to provide a local stratum 1 Network Time Protocol (NTP) time server tick. In addition the Linux Steer interface cage device driver has been configured to slave the operating system clock of the steer machine to the cage's own timecode receiver card (in turn supplied with NASA 36 timecode by the |rueTime TCU). By using the LOCAL clock driver within the stock xntp3 distribution as a (fudged) stratum 0 reference clock,steer then runs as a backup stratum 1 NTP time server. Further redundancy is provided by a HP 6065A Rubidium Vapor frequency standard plus a Systron Donner 8120 Time Code Generator (the Station Backup Clock) and a prototype NASA 36 timecode transmitter.

All NCCS data acquisition is referenced to the HartRAO Station Clock (ie. the local fundamental reference) by direct hardware clocking within each of the acquisition system, namely the Steer Interface Cage, the Australian LBA Data Acquistion System (DAS) and the Mark I Pulsar Timer. Both nccs1 and nccs2 run as stratum 2 NTP time servers (thus ensuring that both operating system times are correct) referred primarily to tick with both steer plus a basket of external NTP servers providing redundancy. All other HartRAO (U)nix machines run as NTP clients of nccs1 and nccs2, such that the entire network has a closely linked idea of operating system time which typically should track UTC to within 10ms.

Next Previous Contents