Moin,

On Mon, 3 Dec 2018 18:11:30 -0800
"Tom Van Baak" tvb@LeapSecond.com wrote:

As this is becoming a topic again, and the EverSet people still
claim that the BPSK modulation of WWVB makes it more resilient
to jamming and easier to receive, I would like to restate
what I've written some years ago [1,2]:

BPSK by itself does not improve timing. At most it improves reception
by having a constant power envelope. But in case of WWVB, where
the AM modulation is still kept, this is not the case. The phase
changes do not help reception at all. In order to help reception,
one needs to send a known bit string in order for a corrolator
in the receiver to pick the signal out of the noise. The only
known bits in the BPSK signal are the first 12 bits of each minute.
Compare this to DCF77 which encodes 512 bits every second.
Ie while DCF77 gets something like 10-20dB  easier to pick out
of noise, but when using BPSK, WWVB gets... uhmm..  zero improvement.

All the BPSK modulation of WWVB does is

1. Make the signal unusable for any carrier phase tracking receiver
2. Add a second type of bit stream onto the signal for additional
   information to be encoded
3. Generate a revenue stream for companies who sell new WWVB receivers.

For more information, read [2]

		Attila Kinali

[1] https://www.febo.com/pipermail/time-nuts/2013-July/060456.html
[2] https://www.febo.com/pipermail/time-nuts/2013-July/060471.html

--
<JaberWorky> The bad part of Zurich is where the degenerates
throw DARK chocolate at you.

On Tue, Dec 4, 2018 at 9:22 PM Attila Kinali attila@kinali.ch wrote:

This seems transparently incorrect to me.  If your receiver has access
to only a tiny chunk of signal and no idea of anything else then yes,
but that isn't a realistic restriction.  Given that we know the signal
is a clock all the bits are almost perfectly predictable.  Unless I'm
confused about something about the signal.

The result is that you have some additional delay in learning the
carrier phase-- since an optimal decoder will make use of both past
and future signal to predict the bitstream, but that isn't really any
challenge against the PLL time constants that would be used.

Hi

Again to re-go over what has been said in the past:

Unless they start filling the “extra” bits on the WWVB signal with something
(are they doing this ????) the whole modulation pattern is predictable. Once
you know what time it is “now” what happens from then on can all be calculated.
I believe TVB has already done this part of it.

A “proper” receiver might take the AM and PM modulation signals and generate
an “expected” signal from them once the time now is known. Correlating against
that signal would seem to be the way to get this done.

The one thing the Everest chips do indeed do is demonstrate that initial signal
acquisition can be done under really awful conditions. Getting the same info off
of WWVB with AM only …. not so much ….At least not so much with a receiver
chip that is out on the market.

Bob

In message 20181204222100.293dc3259cf1d8683daeccdf@kinali.ch, Attila Kinali writes:

They do that in the "long" sequence, read their spec.

The fact that it is one out of 100-some known sequences does not
significantly change that situation.

And the BPSK does improve timing.

If you have access to, and tracks the carrier, you can nail the
timing all the way down to 1/your_sample_rate.

Been there, done that with DCF77 no big deal.

Even the weird phase-modulation of RDF and BBC on 198kHz can be
nailed down very precisely that way.

But the crucial thing is:  You need to digitize the carrier.

--
Poul-Henning Kamp      | UNIX since Zilog Zeus 3.20
phk@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.

On Tue, 4 Dec 2018 21:42:41 +0000
Gregory Maxwell gmaxwell@gmail.com wrote:

Yes. True. Once you aquired the signal you can guess quite a few bits
of the next minute. But this gives at best 60 bits per minute of
information for the BPSK correlator. Compare this to 512 bits per second
of DCF77. Ie DCF77 has 500 times more information to work with and
to pull the signal out of the noise than WWVB.

Or to put it differently: you could say the same for the AM signal.
Once you know what bits will be transmitted, you can do a matched filter
and get an increase in signal quality. Of course, AM is not as easy to
use this technique on and the gain is less, but at these low rates,
it's basically identical.

As I explained in one of the mails I linked to, the reason why BPSK
helps with getting signal out of the noise is the pseudo-random signal
modulated ontop of it, that gives a_lot of additional information
to the signal and de-correlates it from the usual jammers (single frequency
narrow band and wide-band noise-like jammers)[1]. The WWVB does not do
that. The additional spreading of the spectrum of the signal is minimal
and in the order of the spreading due to the AM modulation.

		Attila Kinali

[1] This is not tied to BPSK as modulation, but works with
any modulation. The key point here is making the spectrum wide
and predictable. See https://en.wikipedia.org/wiki/Spread_spectrum

It is upon moral qualities that a society is ultimately founded. All
the prosperity and technological sophistication in the world is of no
use without that foundation.
-- Miss Matheson, The Diamond Age, Neal Stephenson

Hi Bob,

On Tue, 4 Dec 2018 16:43:57 -0500
Bob kb8tq kb8tq@n1k.org wrote:

I have not seen any measurement/demonstration of this property yet.
And my back of the envelope calculations showed that BPSK modulation
shouldn't give any substantial gain. Do you have any references where
I could read up on this aquisition gain?

		Attila Kinali

--
It is upon moral qualities that a society is ultimately founded. All
the prosperity and technological sophistication in the world is of no
use without that foundation.
-- Miss Matheson, The Diamond Age, Neal Stephenson

Hey Poul-Henning!

On Tue, 04 Dec 2018 22:56:54 +0000
"Poul-Henning Kamp" phk@phk.freebsd.dk wrote:

Yes, but the same is true for pure AM modulation. If you sample
and track the carrier digitally, then you get quite a bit of
processing gain due to the long integration time. My point is that
the WWVB BPSK modulation does not give (much of) an advantage
over the AM modulation because the amount of known information
is so low.... That is, unless I miss something and there is actually
an advantage that can be exploited.

		Attila Kinali

--
It is upon moral qualities that a society is ultimately founded. All
the prosperity and technological sophistication in the world is of no
use without that foundation.
-- Miss Matheson, The Diamond Age, Neal Stephenson

Hi

There have been several posts in this thread documenting some pretty
amazing performance for the chip. The wall clock I have here (with
what I assume is the same chip) does way better than anything I’ve
had before.

Bob

In message 20181205150943.d18ce043dfd44d2731220327@kinali.ch, Attila Kinali w
rites:

No, it is not, because the AM modulation is subject to
the very narrow bandwidth of the transmission antenna which
smears the flank out over several carrier cycles.

The BPSK phase reversal is not smeared out the same way.

--
Poul-Henning Kamp      | UNIX since Zilog Zeus 3.20
phk@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.

I also ordered one of the EverSet Dev Boards.
Pricey for a one IC "glop-top" package, but I would like to see how it
performs.
But, I could not duplicate quantity one for the price, either.
I also note that they sold out about mid day yesterday, so Tom's
announcement cleaned them out.

As far as the claimed 20 (ish) dB gain of the BPSK modulation scheme, if
you read the papers, it is a system gain, not just a gain from BPSK
modulation, involving error correcting coding and extended correlation
across six minute sequences.

--- Graham