We would like to synch the read time of (MSP430 based) data loggers to within about 10 ms over a 2-3 day period . However, the crystal oscillators used for the MCUs do not have this kind of stability, and any oscillator that might would not be within our power budget.
One idea for synchronization is to read the time from the GPS system; I note that the Sparkfun GPS Buying Guide says the time is given to “within several milliseconds.” We don’t care about position information; our important issues are low power requirements and short latency to reporting the time (maybe from the first satellite acquired?).
My questions:
Has anyone used one of the Sparkfun GPS devices for time synchronization?
Can anyone recommend one of the devices from the long Sparkfun GPS list?
Does anyone know of websites/papers/books that address this particular use of the GPS systme?
The problem with GPS is that if you’re indoors you could easily loose signal… I don’t know about the actual time accuracy, but this product seems to prove that the concept could work… http://www.endruntechnologies.com/gps-time-server.htm
You probably need a GPS with external antenna if if you are indoors to mount it on a window or the roof!
Any GPS module that provides a 1PPS output should provide the accuracy you require.
GPS is widely used for time synchronization purposes. The national ranges (White Sands Missile Range, Wallops Island, etc.) need to accurately sync up data recorded at widely separated locations. This used to be done with a master clock that transmitted a RF time signal which of course had to be corrected for propagation delays. Now they just plop down a device at each location that receives GPS.
For minimum power you can read time from the GPS infrequently and use it it to correct you local clock. You might find, for example, that your 32,768 Hz crystal is oscillating at 32,767.1 Hz.