TEXAS On-the-Air Confirmed: Texas lowfer copied in Canada, 1623 miles!

[Cantrell Bill-QA0057]

Hello All,
I just got in to work today and saw the flurry of email regarding Bill de
Carle's accomplishment.  Dave Riley is almost there too!  

>From Dave - AA1A:
06:39am: <Lrt[09]f[0D]-,TXh2\A/\AsXuiXsYp)Yu):R<[09]EE,TU

Hi Bill de Carle,
Congratulations and Great Job!  Yes, I was on the air at that time, and
continue to be on the air 24 hours, running good antenna current right now.
The forecast is clear dry weather for this part of the country for at least
a week -- perfect for lowFER transmissions.  My coordinates are:  Lat:
32.92598N, Lon: 97.32581W, so I calculate the distance at _ 1623 miles _ as
the crow flies using:

 http://www.indo.com/distance/

Thanks much for trying for me, and thanks ever so much for your AFRICA
program in general.

I am still running continuous BPSK, 24 hours:  TX(space) , run length of 3,
MS100, ET1, 189.700 kHz, +/- 1 part per billion,  1 watt DC input for ERP ~
5 milliwatts.

TX TX TX TX TX TX TX TX TX ...

Regards,
Bill
"TEXAS" & WD5CVG

-----Original Message-----
From: Bill de Carle [mailto:bill@ietc.ca]
Sent: Tuesday, December 28, 1999 10:52 AM
To: bpsk@qth.net; lowfer@qth.net
Subject: [Lowfer] Texas lowfer copied in Canada


Yesterday morning I awoke to find a couple of hits of TX<sp> in the
overnight trace-file.  I was using AFRICA V2.5, MS100, ET1, with SYNC
time of 250 seconds, GRAB run-length of 3, GRAB depth of 31.  The receiver
was a Kenwood TS850S with external frequency standard tuned to 189.700
Khz with AFRICA's frequency-tracking loop disabled.  Antenna was a
rectangular loop approximately 20 feet high and 30 feet wide, looking
towards the west.

The momentous event was a long time coming - three years or so.
Conditions have been improving over the past few days and I sort
of figured I might have a chance.  The main obstacles are some local
powerline-related noise I've never been able to get rid of completely
and that powerhouse broadcast station from Iceland on 189.0 Khz.
Iceland is on an extension of the bearing from TEXAS to here, so I can't
do much to null him out by turning the loop.


I started my log on Sunday Dec 26th a few minutes before 11 pm Eastern
standard time, and it ran through dawn the following morning.  When
I went to bed the Icelandic station was pounding in more annoyingly
than usual.  Even with my narrow CW filter in, with the rig tuned to
TX's frequency of 189.700 Khz all I could hear was music from Iceland!
The local noise seemed somewhat abated, and static crashes were infrequent
though loud.  I figured that at some point during the night Iceland would
fade out and I might have a shot at TEXAS.

That apparently happened around 6 am local time.
Here is an excerpt from the log:

06:00:09 *=ps[00][18]$[1F]t;[00][18]ehts=tgW9gu9|u9|u9g"`y"`e"texto3t[0F]
06:01:22 3t[0F]xT[09]3T[09]3TkrTXrTXwTXwTXymXyTX TX TXyTXymX mX2TXsTXsvX
06:02:49 svXsmXs[15]XsTXsTXyTX*vX*vX*TX*TX[1E]TX>TXtTXyd2>T8>4[08]>4;[0F]

As you can see, there are lots of TX's but only two complete TX<sp>'s.
Turns out, over the entire 10.8-hour trace file, which has some 24,379
decoded characters, there were 19 hits on "TX" and only the 2 on "TX<sp>".

Yes, I know that if you use enough chimpanzees sitting at enough
typewriters for long enough they will eventually reproduce the complete
works of Shakespeare :-) - so why do I think these two hits on TX<sp>
are due to anything other than random noise?

Well, the chances of finding the sequence TX<sp> somewhere in 24 thousand
random characters are 1 in 29.15 - and the chance of it happening *twice*
is one in 850!

However, I was using grab, which makes things more complicated.  We can
see that if a particular character sequence occurs once, using GRAB makes
it more likely to occur again close by - more likely than if GRAB were not
enabled.  Just how much more likely is difficult to say - at a first guess
I'd say the probability for the first occurrence would be one in 29.15,
same as if GRAB were not enabled, but the probability of a second occurrence
within the next few characters would be one in 29.15 divided by the square
root of the number of samples added up, the GRAB depth.  Any statisticians
out there want to comment?  In this case I was only using a GRAB depth of
31, so the probability of second occurrence would be around one in 5.235.
Taken together, we can say the chance of getting two (2) complete TX<sp>
decodes in 10.8 hours of logging works out at 1 in 150 or so, even with GRAB
enabled.

There is more here to convince me I was actually seeing Bill Cantrell's
signal and not just random noise.  Clearly while that Icelandic station
was pounding away most of the evening, no copy would have been possible.
I know Iceland fades away just before local dawn - so the only real
opportunity for me to copy TEXAS would be right around 6 am local time,
while it is still dark between here and Haslet TX.  That is exactly
when the hits occurred.  With almost complete confidence, I could have
started recording only after the Icelandic station faded and terminated
the listening session just after dawn here - which would have reduced the
number of decoded characters to a fraction of those in the log.  With the
smaller sample set, the odds of finding two complete copies of TX<sp> would
be much slimmer.  One last test:  I looked for instances of "XT" instead
of "TX" - if pure noise were generating those hits you'd expect to find
approximately as many instances of XT as of TX.  Well, there were no hits
on "XT" over the entire 10.8 hour log.  So it sure seems like some of the
energy from the transmitter in TEXAS was influencing my receiver, hi!

I'm claiming good copy!

The co-ordinates of Haslet are 32:58:29N, 97:20:51W - and my co-ords here
in the Laurentian Mountains of Quebec are 48:56N, 74:18W - so I calculate
the distance as 1510.76 miles.  That's the furthest lowfer signal I've
ever copied.

Now all I need is confirmation from Bill Cantrell that he was really on
the air, hi!

73 de Bill VE2IQ