[time-nuts] 60 KHz Receiver

Poul-Henning Kamp phk at phk.freebsd.dk
Tue Oct 5 18:56:16 EDT 2010

In message <4CABA343.8C5812B1 at cox.net>, WB6BNQ writes:

>Please explain to me how spread spectrum would enhance any process
> of frequency or time recovery ?

Ok, it is late and I'm probably going to botch this, but I'll try:

The really short explanation is that your carrier transitions have
random-ish looking signs, which, if properly designed, allows you to
balance out pretty much any kind of CW or random noise.

This is, in essence, why you can separate the different GPS
sattelites, even though they all send on the same frequency.

Technically speaking, Loran-C is spread spectrum, but they botched
this aspect slightly, by not properly balancing the signs of (all)
the codes.

The Austron 2000 has a switch that allows you to disregard certain
bits in the codes to balance them, this increases the imunity to
CW interference.

So given that you can trivially get a good OCXO today, I would design
our "low-power-time-transmitter" to send one fix per hour.

For instance 127 bits of PRNG at 28 seconds per bit with a four
second gap before the next timestamp (send ID ?)

On the receiver side, you know what time it is +/- one 28sec bit,
so you digitize the signal and correlate the PRNG in a window around
your local clock.

After an hour, you pick the correlation bucket that correlated best
and have an instant estimate of the difference between your local
clock and the average of that hours transmissions.

If xmitted as NFSK at around 100kHz and digitized at 1MSPS, you
would get 1µsec resolution without resorting to interpolation.

By choosing all your magic numbers to be nonprime to normal
CW signals inside their respective periods, you supress those
by averaging.

This is why the NELS LORAN-C chains got new 4-digit GRI's: they
are imune to pretty much traditional CW interference because
they do not divide seconds or kHz on relevant timescales.


PS: DCF77 already does SS, but on a second to second basis.

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