The GOES time code service, 1974-2004: a retrospective

Journal of Research of the National Institute of Standards and Technology, March-April, 2005 by Michael A. Lombardi, D. Wayne Hanson

[FIGURE 8 OMITTED]

The original receiver operated in uncorrected mode. NBS Technical Note 1003 [37], published in 1978, contained the plans for a GOES "smart clock" that operated in corrected mode, making automatic path delay corrections. This clock added a second Intel 4004 microprocessor and a math chip removed from an early scientific calculator to the previous decoder clock shown in Fig. 8 [35]. The new hardware was used for the calculation of the free-space propagation delay from the CDA to the clock via the satellite. This delay value was used with a delay generator to compensate for the free-space path delay.

The published NBS receiver designs quickly generated interest in the commercial sector. It was noted in 1978 that NBS had been contacted by "more than 18 manufacturers" who were interested in building GOES time code receivers and by then, several commercial models were already available [38]. At least three manufacturers, including Arbiter Systems, TRAK Microwave, and True Time, made long term commitments to manufacture and sell GOES time code receivers; and it is estimated that over 10 000 commercial receivers were sold. The cost of a GOES time code receiver and antenna generally ranged from about $2000 to about $7500 U.S. dollars, depending upon the number of features included. Arbiter and True Time produced several different models and continued to sell and support GOES receiver and antenna products well into the 1990s (Fig. 10).

[FIGURE 9 OMITTED]

[FIGURE 10 OMITTED]

9. Time and Frequency Control and Uncertainty of the GOES Time Code Service

The frequency and time reference for the GOES service was usually a cesium oscillator, but rubidium oscillators were sometimes used. These oscillators were originally owned and maintained by NIST, but in the later years they were provided by NOAA. The TCGs were referenced to the atomic oscillator frequency and set to be 260 000 [micro]s ahead of UTC(NIST), with an uncertainty of [ or -]10 [micro]s throughout the lifetime of the service [32], and [ or -]1 [micro]s during the last nine years (1996-2004). The TCGs were adjusted from Boulder in time increments of 0.2 [micro]s, the period of the 5 MHz time base.

GOES time at the CDA was synchronized by NIST personnel using a number of techniques throughout the years, including the use of television signals from nearby Norfolk, Virginia, LORAN-C transmissions from Cape Fear, North Carolina [29, 30], portable clock carries from Boulder to Wallops Island [39], and eventually through the use of GPS signals [32]. The performance of the station clock was continuously monitored from Boulder using a remote controlled data logger [40], which was replaced by more modern equipment in 1999.

The quality and age of the orbital elements always limited the timing uncertainty of the GOES service more than the performance of the station clock. In the 1980s, budget cuts forced NOAA to drop the trilateration ranging network and to make position estimates using images of the Earth transmitted by the satellites every 20 min. This resulted in poorer quality orbital elements [41]. Other incidents that increased the timing uncertainty were satellite maneuvers, where NOAA sometimes moved the satellites to a new position before new orbital elements were available and the position data in the TCGs could be updated; and solar eclipse periods, when the time code was sometimes moved to a spare satellite for periods as long as two hours each day [32]. Despite these challenges, the stated [ or -]100 [micro]s time uncertainty was usually met throughout the lifetime of the service, from the early days [33, 39] until the end. Figure 11 illustrates this by showing the results of a 3 month comparison between GOES-West as received in Boulder and UTC(NIST) in 2003. The noise visible in the received data is not random, and was mostly caused by errors in the delay computations due to imperfections in the predicted satellite orbit as discussed in the next section.