I've got time on my mind

[Dr Sardonicus]

Hmm, it seems GPS time has some idiosyncracies of its own.

First, GPS time zero is 00:00:00 UTC 6 January 1980. It continuously counts time in seconds since then, using a system of atomic clocks. The system guarantees a "time transfer accuracy" of 40 nanoseconds.

GPS time progresses uniformly - it does not insert "leap seconds." It is therefore offset from UTC by the (net) number of leap seconds inserted in UTC. This offset is presently 18 seconds.

GPS counts weeks. Currently, this count uses a 10-bit register. So the weeks count "rolls over" every 1024 weeks (a bit under 20 years). Some GPS receivers start the count with the firmware release date "hard wired" in.

The system is being upgraded, including using a 13-bit register instead of a 10-bit register to count weeks. Then the weeks count will only roll over about every 157 years.

In addition to GPS time, here in the US there are UTC time signals broadcast by radio station WWV. According to the FAQ,

WWV broadcasts on 2.5, 5, 10, 15 and 20 MHz from a location near Fort Collins, Colorado. WWVH broadcasts on 2.5, 5, 10 and 15 MHz from Kauai, Hawaii.
...
The time is kept to within less than 0.0001 milliseconds of Coordinated Universal Time (UTC) at the transmitter site, but the signal is delayed as it travels from the radio station to your location. This delay increases the further that you are from the station, and the delay can vary by as much as 1 millisecond if the signal is bouncing between Earth and the ionosphere. However, for most users in the United States, the received accuracy should be less than 10 milliseconds (1/100 of a second).

This is not accurate enough for some things, as already indicated. But it is accurate enough for many applications, and the signals are freely available, courtesy of US taxpayers. Many devices, from ordinary clocks to agricultural equipment, are equipped with receivers for these signals, and depend on them.

In 2019, the Administration proposed "zeroing out" the budget for the NIST facilities providing these signals. Congress, in its wisdom, has kept the system funded.

[chalsall]

Originally Posted by Dr Sardonicus

Hmm, it seems GPS time has some idiosyncracies of its own.

Yup. Including the fact that relativity has to be taken into account...

[EdH]

Hmmm. . .

I don't often purchase all the new gadgets when they appear and often accept hand-me-downs from others that do stay on top of technology. A couple of examples were older GPSs. When I first energized one of them a few years ago, it wouldn't operate until I supplied the date/time. I found this odd, since it should know more precisely what the time is where it is located, than I could possibly enter. Perhaps Dr. Sardonicus has explained it.

Note: Older GPSs are quite fun. They often show me driving out through fields and get confused in rebuilt intersections, such as where a circle has replaced a four-way stop.

[xilman]

Originally Posted by chalsall

Yup. The NTP developers *seriously* understand time! Careful measurement and correlation of network latency, jitter, etc. For those who don't get their time from a GPS data stream.[/URL].

Indeed.

I first used NTP on a 486sx-class machine running Slackware. Most of the time it was a free runner. Once or twice a day I would connect to the internet with a 38k4 dialup modem.

After a run-in period of a week or two it was rare that the clock was more than a tenth of a second away from reality, even if I hadn't connected for several days.

[Uncwilly]

Originally Posted by EdH

When I first energized one of them a few years ago, it wouldn't operate until I supplied the date/time. I found this odd, since it should know more precisely what the time is where it is located, than I could possibly enter.

https://en.wikipedia.org/wiki/GPS_week_number_rollover

[EdH]

Originally Posted by Uncwilly

https://en.wikipedia.org/wiki/GPS_week_number_rollover

Thanks! Quite interesting. I must study this more. . .