[time-nuts] HP 117/10509a...

paul swed paulswedb at gmail.com
Sat Jul 7 15:30:36 UTC 2012


Pretty sure NIST will not do anything. Just to set expectations.
We are fortunate that to some extent John Lowe is responding to questions.
But we are on our own.
I think the big lesson I have already learned is that there are lots of
standard approaches to solving the problem Micros FPGAs dpll pll.....
But the fun comes in when you account for the 17 db amplitude variation for
modulation. With propagation, with BPSK and sprinkle in noise thats higher
in level then the signal that contains impulse and random crud.

Regards
Paul


On 7/6/2012 11:28 AM, Bob Camp wrote:

Hi

My *guess* is that $50 is in the ball park for parts cost of a pretty
good receiver for the new format. That does not include things like
the external standard, antenna, frequency comparison stuff, power or
case. I'd bound the range of the guess as $25 to $100.

Bob

On Jul 5, 2012, at 11:56 PM, J. Forster wrote:


 On Thu, Jul 05, 2012 at 04:19:25PM -0700, J. Forster wrote:

 If propagation goes south, you loose track of the carrier phase, the
basis
of the system. If your local standard is stable and close to right,
that's
not a big deal. If not, you can easily go down the garden path.

 	If I read this correctly, you mean you have a 180 degree
ambiguity due to the BPSK - obviously losing track of the carrier phase
in general with a significantly wrong local standard loses...

 David,

Most of what has been tried is an analog squareing, then a divide by two.
No additional PLLs in the system, beyond what is already in the Rx.


 	I have not devoted enough time to this to be absolutely sure but
it sure sounds like from what I read that if you know the accurate time
to one second it should be possible to unambiguously predict the carrier
phase sequences simply because you know the message format exactly, AND
you know the exact time of day message that is being transmitted or most
of it.

 The BPSK rate is 1 bit per second, There are 120,000 half cycles in that
time. Fades can last seconds, minutes, or hours. It comes down to how long
does it take your local standard take to drift roughly 4 uS.

At the moment we are not looking at the message at all.

Certainly a correlating receiver that uses the message as well as the
carrier could be built. But, IMO, that'd be a whole lot easier done from
scratch with a micro. The object here is a small, fairly simple, retrofit
for the existing receivers. The message format may not be fully defined as
yet. The squareing approach is message independant.


 	There are of course two forms of encoding in PSK modulations -
absolute, and differential (or transition) ... naively to me it would
seem that if absolute encoding is used for this and you know most of the
bits of the message most of the time you could predict which phase will
be used a lot of the time, and also know when you don't know (message
bits you might be uncertain about)...

 If you used the signal to set your local clock, and knew the format, it
should be easy to predict at least a good part, if not all, of the
message.


 	Differential encoding has the down side for this that UNLESS you
know all previous message bits accurately starting from some phase
reference datum you cannot predict what phase is in use at a particular
moment.   Absolute encoding (eg 0 phase for a 0, 180 for a one) doesn't
have that liability and if the time of day message is aligned to, well,
the time of day if you know that with reasonable accuracy (and you do
since you are being sent it in the first place) you should be able to
predict a very large percentage of phases used accurately.

	Again, deferring to those who have done the experiments (which I
have clearly not), it would seem that the ability to predict the phase
most of the time would allow creation of a reliable local 60 KHz
reference which could be used to disambiguate those bits you don't know
apriori

	My naive scheme would be to drive a balanced modulator on the
output of the 60 KHz loop antenna with either two or maybe three values
(1 and -1 or 1,  0  and -1) using some cheapie micro (Arduino, PIC etc)
with a software PLL to keep the bit timing in sync with the signal.

 This is what Equatorial did, in TTL, 30+ years ago.


 	For bits that one could not predict, one could either output 0
to the balanced modulator for the entire bit interval  which would
produce a drop in the 60 KHz carrier, or do a fast timed fraction of a
bit look at the output of a synchronous detector and choose the most
likely value for the bit and use that, maybe after a brief 0 no carrier
interval to avoid a detectable phase glitch.

	Of course the other approach is to start with the assumption you
have a pretty good stable source of clock or you would not be doing this
to begin with, and simply A/D the 60 KHz with the stable clock (say at
10 MHz), delay it by storing samples in RAM for one bit time of the low
speed code  and use that entire interval to decide which phase you were
seeing and suitably adjust the output phase accordingly when you spit
out the samples delayed by one bit time.

	This later approach would certainly be doable with modern
processors mostly in software, certainly so if you could live with say 1-2
MHz sampling of the 60 KHz or so... and quite possibly also pretty
nicely with a modest FPGA complete with the sample storage in the chip.

	Both approaches would be helped a lot if the architecture of the
system allows prediction of absolute phase (eg not differential encoding
of unpredictable messages)... and AFAIK that is not yet set in stone and
could be changed to allow this.

	The intent of both of these schemes would be to ultimately
output a De-psk'd signal that older equipment could process using its
antique analog circuitry without serious issues.   Thus the output
would be an attempt at a phase stable corrected version of the original
signal...

 This is what NIST is planning, I think.  Make a new receiver, then
synthesizing 60 kHz from the internal locked clock. It's kinda like a TV
'Converter Box'. It will continue to provide the functionallity, but at
what price? At $50 it would be a good deal; at $5000 not so much, IMO.

-John

=================




 	Certainly using a lab reference stable 10 MHz derived 960 Khz
or whatever sampling clock to delay the signal one time code bit time
should not produce significant 60 KHz phase wanderings at all...

--
 Dave Emery N1PRE/AE, die at dieconsulting.com  DIE Consulting, Weston, Mass
02493
"An empty zombie mind with a forlorn barely readable weatherbeaten
'For Rent' sign still vainly flapping outside on the weed encrusted pole -
in
celebration of what could have been, but wasn't and is not to be now
either."




_______________________________________________
time-nuts mailing list -- time-nuts at febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

 _______________________________________________
time-nuts mailing list -- time-nuts at febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.



More information about the Time-nuts_lists.febo.com mailing list