Date: Tue, 29 Nov 2016 07:58:47 -0500
From: Markus Vester <markusvester@aol.com>
To: rsgb_lf_group@blacksheep.org
Message-Id: <158b02a87b2-1f7f-112fd@webprd-m75.mail.aol.com>
In-Reply-To: <583C799E.4090608@posteo.de>
MIME-Version: 1.0
Subject: VLF: SpecLab GPS-phaselock with faster PSK
Content-Type: multipart/alternative; 
	boundary="----=_Part_88865_350577255.1480424327087"
Sender: owner-rsgb_lf_group@blacksheep.org
Precedence: bulk
Reply-To: rsgb_lf_group@blacksheep.org
Status: O

------=_Part_88865_350577255.1480424327087
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: quoted-printable

The SpecLab GPS-PLL script can overcome soundcard glitches and provide abso=
lute phase of the antenna voltage, but PSK modulation has been limited to s=
low symbol rates due to the ~ 3 second time constant of the phase swing. It=
 appears that the PLL latency is mainly due to the digimode terminal genera=
ting the output signal, which responds to frequency commands from the inter=
preter script only with some delay. I have now tried to circumvent the lag =
by using the L3 frequency mixer which can be controlled faster, allowing ph=
ase-switching in less than 200 ms.=20


A couple of tricks were necessary: A constant 2 kHz signal is generated in =
the L1-L2 blackbox, which is then converted to the desired VLF output frequ=
ency. In addition, instantaneous phase switching is implemented by switchin=
g the bypass factor in the L4-L5 delay line. Alternatively, the phase can b=
e smoothly ramped by setting a "hardkey" flag to zero.  =20


The configuration file is here:
 http://df6nm.bplaced.net/VLF/fec_tests/vlf_tx_pll_fastpsk.zip


For EbNaut transmissions, you will need to use Paul's software and prepare =
a file "psk.txt" (containing zeroes, ones and spaces, similar to the provid=
ed template), and set up start times etc. in scheduled actions. You can als=
o produce a simple 2-second test pattern by activating the second line in p=
eriodic actions. As before, the output frequency is defined by the second t=
est signal generator, which provides the GPS-controlled reference.


Hope this may be useful,
Markus (DF6NM) =20


-----Urspr=C3=BCngliche Mitteilung-----=20
Von: DK7FC <selberdenken@posteo.de>
An: rsgb_lf_group <rsgb_lf_group@blacksheep.org>
Verschickt: Mo, 28 Nov 2016 7:40 pm
Betreff: Re: VLF: EbNaut transmissions on lower frequencies?, pre-tests: 6.=
47kHz


Thanks Paul,

There could be another reason, see attachment. So far the phase is=20
inverted within SpecLab, not by e.g. keying a relay by the serial=20
interface. Due to the PLL and time delay in the sound card chain this=20
takes some time. With a short symbol length of 12 seconds the effect may=20
become relevant. A phase turn of 180 deg is done by varying the=20
frequency for a certain time as you can see.
So maybe this costs a dB of SNR here?

For a 7 second symbol length in the 75 character message i will need a=20
different technique, taking 'no' time for the phase inversion.

Another thing could be the moving sunrise / sunset time on our path. We=20
already saw a wider phase variation in the 20 character message than in=20
the 7 day 'TEST' message before. Maybe it is even worse now.

Am 16.11.2016 19:48, schrieb Paul Nicholson:
>
> For 1uW ERP, DK7FC to Todmorden
>
> http://abelian.org/vlf/tmp/1479317903_9356.png=20

That plot must have been valid 2 weeks ago. What does it show now? Maybe=20
it is a good idea to start 08:15 UTC instead of 08:00 UTC...
Could it (it =3D the move of the positive green slope towards later times)=
=20
be the reason for the higher phase variations that we can see?

73, Stefan
------=_Part_88865_350577255.1480424327087
Content-Type: text/html; charset=utf-8
Content-Transfer-Encoding: quoted-printable

<div style=3D"font-size:10pt;font-family:arial,helvetica,sans-serif;color:b=
lack;"><div style=3D"color: black; font-family: arial,helvetica,sans-serif;=
 font-size: 10pt;">The SpecLab GPS-PLL script can overcome soundcard glitch=
es and provide absolute phase of the antenna voltage, but PSK modulation ha=
s been limited to slow symbol rates due to the ~ 3 second time constant of =
the phase swing. It appears that the PLL latency is mainly due to the digim=
ode terminal generating the output signal, which responds to frequency comm=
ands from the interpreter script only with some delay. I have now tried to =
circumvent the lag by using the L3 frequency mixer which can be controlled =
faster, allowing phase-switching in less than 200 ms. <br></div><div style=
=3D"color: black; font-family: arial,helvetica,sans-serif; font-size: 10pt;=
">A couple of tricks were necessary: A constant 2 kHz signal is generated i=
n the L1-L2 blackbox, which is then converted to the desired VLF output fre=
quency. In addition, instantaneous phase switching is implemented by switch=
ing the bypass factor in the L4-L5 delay line. Alternatively, the phase can=
 be smoothly ramped by setting a "hardkey" flag to zero.&nbsp;&nbsp; </div>=
<div style=3D"color: black; font-family: arial,helvetica,sans-serif; font-s=
ize: 10pt;"><br>The configuration file is here:<br>&nbsp;http://df6nm.bplac=
ed.net/VLF/fec_tests/vlf_tx_pll_fastpsk.zip</div><div style=3D"color: black=
; font-family: arial,helvetica,sans-serif; font-size: 10pt;"><br>For EbNaut=
 transmissions, you will need to use Paul's software and prepare a file "ps=
k.txt" (containing zeroes, ones and spaces, similar to the provided templat=
e), and set up start times etc. in scheduled actions. You can also produce =
a simple 2-second test pattern by activating the second line in periodic ac=
tions. As before, the output frequency is defined by the second test signal=
 generator, which provides the GPS-controlled reference.</div><div style=3D=
"color: black; font-family: arial,helvetica,sans-serif; font-size: 10pt;"><=
br>Hope this may be useful,<br>Markus (DF6NM)&nbsp; </div><div style=3D"col=
or: black; font-family: arial,helvetica,sans-serif; font-size: 10pt;"><br><=
br><br>-----Urspr=C3=BCngliche Mitteilung----- <br>Von: DK7FC &lt;selberden=
ken@posteo.de&gt;<br>An: rsgb_lf_group &lt;rsgb_lf_group@blacksheep.org&gt;=
<br>Verschickt: Mo, 28 Nov 2016 7:40 pm<br>Betreff: Re: VLF: EbNaut transmi=
ssions on lower frequencies?, pre-tests: 6.47kHz<br><br></div>Thanks Paul,<=
br><br>There could be another reason, see attachment. So far the phase is <=
br>inverted within SpecLab, not by e.g. keying a relay by the serial <br>in=
terface. Due to the PLL and time delay in the sound card chain this <br>tak=
es some time. With a short symbol length of 12 seconds the effect may <br>b=
ecome relevant. A phase turn of 180 deg is done by varying the <br>frequenc=
y for a certain time as you can see.<br>So maybe this costs a dB of SNR her=
e?<br><br>For a 7 second symbol length in the 75 character message i will n=
eed a <br>different technique, taking 'no' time for the phase inversion.<br=
><br>Another thing could be the moving sunrise / sunset time on our path. W=
e <br>already saw a wider phase variation in the 20 character message than =
in <br>the 7 day 'TEST' message before. Maybe it is even worse now.<br><br>=
Am 16.11.2016 19:48, schrieb Paul Nicholson:<br>&gt;<br>&gt; For 1uW ERP, D=
K7FC to Todmorden<br>&gt;<br>&gt;  <a href=3D"http://abelian.org/vlf/tmp/14=
79317903_9356.png" target=3D"_blank">http://abelian.org/vlf/tmp/1479317903_=
9356.png</a> <br><br>That plot must have been valid 2 weeks ago. What does =
it show now? Maybe <br>it is a good idea to start 08:15 UTC instead of 08:0=
0 UTC...<br>Could it (it =3D the move of the positive green slope towards l=
ater times) <br>be the reason for the higher phase variations that we can s=
ee?<br><br>73, Stefan<br><br>
<br></div>
------=_Part_88865_350577255.1480424327087--