[time-nuts] The USFS Frequency Standard...

Poul-Henning Kamp phk at phk.freebsd.dk
Sat Feb 11 11:20:19 EST 2017

In message <2126B261-3E4E-46D4-9181-1FB368244304 at n1k.org>, Bob Camp writes:

>One *could* make a WWVB “new modulation” receiver with some sort
>of MCU demo board and a handful of parts. It would be fine for a basement
>lab / learning sort of project.

With reasonable OCXO as timebase, it would do much more than that.

SRS sold the SR700 Loran receiver as a "Cesium replacement".

>the longer you wait to start that project, the better a board you will have
>as the base of the project. 

Current boards are more than capable of this, a 12 ADC at 1MSPS is plenty.

>How many people want to spend more than a year on that sort of thing

Unless you're a total programming beginner, you will have carrier lock in
a week:

* Configure ADC for 1MSPS

* Interrupt routine: exponentially average the samples into a 50[1]
  bucket circular buffer

* Control-Loop (not time-critical):

  * Multiply 50 buckets with synthetic sine/cos function,

  * Average the 2x50 results to get I/Q phase signal.

  * Feed phase into PLL to steer OCXO.

  * Bonus: Decode timegram.

I did that with Loran-C, which is a *much* harder signal, I did it
8 years ago, and I did it in two weeks.


[1] If you use a platform with enough memory, for instance a BBB,
use a one million (=full second) buffer instead, you will be
able to pull any and all VLF signals out of it.  For instance
the US navy runs a couple of stations with frequency stability
comparable to WWV.  On a BBB you could even add a FRI length
buffer also, to receive the Wildwood Loran-C.

Poul-Henning Kamp       | UNIX since Zilog Zeus 3.20
phk at FreeBSD.ORG         | TCP/IP since RFC 956
FreeBSD committer       | BSD since 4.3-tahoe    
Never attribute to malice what can adequately be explained by incompetence.

More information about the time-nuts mailing list