[opendtv] WWVB signal change
- From: Albert Manfredi <albert.e.manfredi@xxxxxxxxxx>
- To: "opendtv@xxxxxxxxxxxxx" <opendtv@xxxxxxxxxxxxx>
- Date: Sat, 4 May 2013 19:20:08 -0400
Funny how such really important stories don't ever make the news. I think there
was a short piece on TV Technology magazine about this.
So WWVB is a 60 KHz station that belongs to NIST, in Fort Collins, CO, which
transmits a 70 KW ERP time signal. Where the WWV transmitters also have an
audio track, and transmit in the SW band, WWVB is only a digital signal,
intended to cover all of the continental US.
For years now, WWVB has used a combination of 2-ASK and PWM, to transmit a
BCD-encoded year, day of year, hour of day and minute of day signal, where time
is UTC. Each time update is sent at 1 baud, and takes one minute to complete.
The modulation scheme is weird. The full strength carrier is attenuated by 17
dB for either 0.2 or 0.5 seconds, to denote a 0 or a 1 respectively. and by 0.8
seconds to transmit a "marker." Seconds are derived by the receiver, because
the minute signal is always transmitted at precise times. Also transmitted are
DST pending, DST active, leap second info, and the correction between atomic
time and UTC.
This odd signal was apparently devised in the mid 1960s, and has not changed
much. The attenuation was increased from 10 dB to 17 dB, when the ERP was
raised to 70 KW a few years ago. It's kind of difficult to receive in parts of
the US, e.g. the East Coast (although my clock does a decent job of it after I
added a long wire antenna, capacitively coupled).
Big change happening as we speak.
In addition to keeping the same old ASK-PWM signal as is, WWVB is now also
layering a BPSK signal on top of the full power portions of the existing
carrier. BPSK is the most robust form of n-PSK, using the full 180 degree
phase shift to go from a 0 bit to a 1 bit. According to NIST, this new
modulation should do a much better job of covering the continental US, avoiding
previous (abortive) plans to locate another transmitter in Alabama. Also
avoiding any power increase.
The message content is all different too. The scheme transmits a 26-bit binary
"minutes of century" signal, plus the DST, leap seconds, etc, and a 5-bit
parity check code. Time is still UTC, and it still takes a minute to get one
update out.
A constant bit sequence replaces the "marker" of the old scheme.
http://www.nist.gov/pml/div688/grp40/upload/NIST-Enhanced-WWVB-Broadcast-Format-2012-12-07-33.pdf
Given that they made this change to a more NTP-like "minutes of century," as
opposed to explicitly transmitting year, day, etc., I kind of wonder why they
didn't just use, say, the top 32 bits of the NTP time stamp. At one baud, that
would take 32 seconds per minute to transmit, which I would think would have
fit in the full power portions of the ASL carrier? Maybe it was too tight a fit.
Anyway, this was an interesting discovery for me. You'd a thunk that a brief
mention in the news would have been called for.
Bert
Other related posts:
