re: #297 Eric The Fruit Bat

The cell towers could also get their time source from their internet backbones as well, via NTP - it would certainly be cheaper than equipping each cell tower with a specialized WWVL receiver that feeds a digital time source into the tower (especially if the tower is hosting multiple cellular providers.)

That presumes the cabled network lag from the NPT Stratum 1 source can be precisely characterised. For most desktop and mobile phone applications etc. being half a second out in either direction is not critical, for a lot of niggly technical purposes that’s unacceptably sloppy.

GPS time signals can be assumed accurate to within fractions of a microsecond since the transmission path length is known to within a metre or two and compensation applied. If better than that is required then further calculations involving airpath changes to the speed of the radio transmission can be made to correct the time data even further. This of course can be further improved by using a dozen or more GPS downlinks together.

A long time back (not quite pre-GPS but close) I was involved in designing an accurate time clock system for financial transactions recording with a required accuracy of plus/minus 0.5 milliseconds. That used a broadcast time signal with geographical corrections to allow for speed-of-light delays from the transmitter to fine-tune the clock.