Hi Rob,
This message was supposed to be posted when I posted the one to Charles - it got lost.

I take your point - this is not easy at all.

I'd probably lock the decode process inside an FPGA, use hardware to ensure that the latency is predicatable, with an ovened crystal giving pretty good short term (<10 second) frequency accuracy, lock that to the GPS PPS signal a la Charles' link, and decode a time stamp. The readout would be then a question of counting Local oscillator cycles after the latest time stamp, which I hope is locked by a very fast edge, and then deconvolving it into absolute time.

I don't remember you posting that link to me though !

Very early days yet.

Steve