[time-nuts] Antique 5061A option 004

[David I. Emery]

On Mon, Oct 15, 2012 at 06:15:15PM -0400, paul swed wrote:
> David
> Yes indeed I picked mine up for $125. Many hours of fun though the tube
> really was bad. But sure learned a lot and as mentioned here on time nuts
> actually have it working with a 5060 tube. John mentioned your location and
> we are actually reasonably close. Like you I was at nearfest.
> So before you fire it up perhaps we should talk. Part of the weight may be
> the nicad pack and who knows what condition its in... Hope it did not vent.

	I am about to look at the batteries.   I think they likely are
still there from the weight.   Those old nicads are less of a problem
than some new ones... I've got some old spook gear with those things in
it and while some were shorted none leaked badly.

	As you may have learned John Forster ran into me as I was
carting the thing off... he offered to help move it on his motorized
wheel chair but I thought it too heavy and clumsy to try that - I'm not
very coordinated and of course since he had his diabetes problem he is
not agile either - might have made for a dicey situation.  In the end a
good Samaritan helped with one handle before I got to the parking area
outside the fence...

	I live right near the business center of Weston on Boston Post
Road... (west of the center).   If you know the area you might have
noticed a house with some discreet satellite dishes in the back yard...
and antennas on the roof...

	I'd be glad for advice before powering up, though I am eager to
see if it pumps down and beam I is reasonable.   It does seem to have
some mods for setting the C field - a stereo 8MM phone jack on the front
panel with a little label marked "M" and of course the C field output on
the back panel.  Neither appears original.

	I did follow your adventures with yours, and of course on
another topic have been injecting my 0.02 into the discussions about
WWVB phase modulation on and off as well.

	The current spec for that seems pretty rational to me at the
moment in fact... though I agree that with low signal levels and QRM and
QRN you will never be able to lock the signal reliably without some kind
of very low bandwidth PLL (eg a Costas loop).

	It appears that one can predict many of the bits (and thus
phases), but not all of them - especially if they add various random
messages interspersed with the BPSK time info and given that one could
produce a de-PSK'd signal with a balanced mixer with some dropouts for
the early part of  unknown-in-advance bits.   Whether this would work OK
with antique receivers or not I don't know.   Would seem just as easy to
do a Costas loop and simply feed the antiques the carrier recovered in
the Costas loop though of course that does introduce another source of
phase error and phase drift.

	The one liability of a Costas loop approach is long fades. You
would need either to provide a squelch of some kind that would stop
phase tracking or lock it to some kind of local reference based on the
local standard during fades.   The latter is quite possible, if you have
an accurate standard you are comparing with - all you need is to use
some digital processing (on a processor these days of course) to
synthesize a carrier in phase with WWVB and when the signal fades  feed
that to the loop until the signal returns.

	If real time time of day operation is not required, another
technique suggests itself... simply delaying the signal by one symbol
time (in this case 1 second) by digitally sampling it with an A/D at
some convenient rate (like 1 MHz) and stuffing the samples into a
circular buffer before outputting them via a D/A.   During that bit time
it should be possible to correlate with a local digital generated 60 KHz
and determine the best estimate of the bit phase state.   And when one
gets to the best estimate of the 1 second splice, one simply adjusts the
phase of the sample play out by multiplying the samples by 1 or -1
depending on which phase state seems to correlate best with the signal.

	A slightly more sophisticated version of that would use the
known time code most of the time and just use the correlation when the
bit state was ambiguous due to the possibility of messages.   That would
ensure that the 60 KHz phase in the reconstructed signal matched the 0
degree phase of the transmitted signal.

	The 1 symbol time  delay technique ought to be able to handle a
very very very long fade given a decently accurate digital time base
since it is not necessary that the local synthesized (numerically inside
the software I would expect) 60 KHz have any particular exact phase
relationship to the WWVB carrier... clearly one would want to correlate
in I and Q space (eg with the carrier and the carrier shifted 90
degrees) and make ones best guess where on the unit circle the 0 phase
carrier lies...

	When one returned from a fade one would want to try to
re-estimate the phase relative to the local time base of the WWVB
carrier (by knowing what phase state most symbols were) - and clearly
one do this for a few bits before attempting to do any kind of
estimation of unknown bits.

	But given a normal accuracy digital clock (very few ppm) once
one had acquired time of day lock and bit (symbol) timing it would be
tens of thousands of seconds before a despreader deBPSKing  known bits
of the time code - all one second long - would be off enough to make
much difference for a narrow band loop tracking the signal - and this
could be reduced further by blanking the beginning and end of the bits
(the splice) so only the middles were actually output via the D/A at
full amplitude.


-- 
  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."