[time-nuts] WWVB Response
Clint Turner
turner at ussc.com
Thu Sep 27 14:57:08 UTC 2012
While there could have been a few things to make the WWVB transmissions
easier to recover with low S/N, keeping them compatible with the
"legacy" time-only receivers was somewhat of a hindrance.
Unlike the DCF77 signal - which has a digital phase modulation that does
NOT really lend itself to improved SN (it's at a much higher rate, but
since a WWVB-ish signal is much stronger in Germany owing to its
small-ness) and was really intended for use with strong signals, anyway
- the vast majority of users ignore the BPSK, anyway and rely on the AM.
As for how the new WWVB format allows improved recovery of frequency
reference and time signals:
- I really don't see how this new scheme makes it worse - just
different. A very narrow bandwidth loop on the carrier recovery such as
a Costas (e.g. as slow as you want!) allows the carrier to be recovered
at very low S/N. At least in part, it depends on how long you want to wait!
- With typical antennas and receivers, the fairly narrow bandpass is
still wide enough to allow reasonable recovery of the signal. Note
that, at worst, the phase changes at a rate of 0.5 Hz and to have
recovered the 1-second time signalling at all with any degree of
accuracy, it would be wider than that 0.5 Hz!
- The phase change occurs in a pre-determined manner. Once you recover
the carrier, you *know* that you can window the filtering on 1 second
blocks and that there are only two possibilities of what the phase could
be within that block. On a computer, you can use "past" data and slide
back and forth until good correlation is received: This would then get
one close to 1-second timing in that same operation! (A bit harder to
do this on very low-power receivers, but easy on a PC.)
- The use of the same preamble (to indicate time or message) can allow
one to review many minutes of past-stored data to achieve a correlation
of the minute frame. Using multiple blocks of past data can greatly
improve the effective S/N of that particular aspect - the degree
depending on how long you are planning to wait! (Again, tougher on a
low-power receiver, but fine on a PC.)
- Once you have "minute" frame correlation (and second-timing as well)
you can then use several techniques to further-recover the time data:
- You *know* that the DST/Leap Second information isn't going to
change much so that can be used for further correlation.
- The time data is incrementing at a known rate. In the event that
the data is consistently "ratty" (e.g. you don't get the full time frame
without errors) you could further-correlate those bits that are NOT
changing very quickly over several minute frames.
- Furthermore, if you have an "approximate" time (which can be known
from the user if not from long-term correlation of that time data that
changes very slowly over several frames) you can then go back in old
data and apply assumptions based on best guesses of time and see if that
data "fits" as well.
- Don't forget that the use of the Hamming code (the ability to
correct one bit of data and to detect if two bits are in error) also
provides a degree of recoverability - particularly if applied over a
number of minute frames.
With a low-bandwidth antenna (assuming that it's not too narrow - that
is, it has 2-3 Hz bandwidth or more) the difficulty will be determining
when, exactly, the 1-second intervals are occurring to within a few 10's
of milliseconds - even knowing the amount of ("fixed") delay in the
filter/detection scheme, but this uncertainty is almost guaranteed with
varying noise and signal level conditions and the vagaries associated
with a very narrow filter.
If, however, the filter within the receive system is extremely narrow
(<1-2 Hz) then the BPSK does make things more difficult.
73,
Clint
KA7OEI
> The AM just makes the situation in low S/N areas worse. The BPSK wipes out
> the possibility of any very narrow band prefiltering, because of filter
> time response.
>
> I suspect, although have not tested, that active antennas with either
> mechanical or crystal filters in their preamps will be rendered useless.
>
> -John
More information about the time-nuts
mailing list