Date: Sun, 26 Feb 2017 04:34:21 -0500
From: Markus Vester <markusvester@aol.com>
To: rsgb_lf_group@blacksheep.org
Message-Id: <15a79c5598b-5011-ee81@webprd-m104.mail.aol.com>
In-Reply-To: <alpine.DEB.2.11.1702260842210.642@mailn.lipkowski.org>
References: <alpine.DEB.2.11.1702260842210.642@mailn.lipkowski.org>
MIME-Version: 1.0
Subject: Re: LF: ebnaut & speclab question
Content-Type: multipart/alternative; 
	boundary="----=_Part_74184_2034344039.1488101661065"
Sender: owner-rsgb_lf_group@blacksheep.org
Precedence: bulk
Reply-To: rsgb_lf_group@blacksheep.org
Status: RO

------=_Part_74184_2034344039.1488101661065
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: quoted-printable

Hi Jacek,


short answer:=20


Yes (if you have a soundcard without dropouts on the input). It will work o=
n many modern PC's. For example, RN3AUS has achieved successful decodesd fr=
om DK7FC using samplerate sync to GBZ.


Longer answer:


The samplerate detector will track the frequancy and the phase of the refer=
ence signal. However if a buffer overrun occurs, an unknown number of sampl=
es may be missing in the data stream. Then the phase of the reference frequ=
ency will be reconstructed, but the phase on other frequencies will still b=
e wrong. For example, if you are using a 10 kHz reference, a 1 ms dropout w=
ill swallow exactly one cycle, so the SR detector will simply not see it. O=
n the other hand, the same 1 ms step will introduce a 97.2=C2=B0 phase step=
 on 8270 Hz, which would ruin an EbNaut decode. With a non-integer kHz refe=
rence (e.g. 22.1 kHz), a phase change would be detectable on the reference,=
 but the correction would still not scale correctly to the receive frequenc=
y. So  to be able to apply a correction to any frequency, we need to know t=
he absolute duration of the outage.


One theoretical solution would be to track multiple different off-air refer=
ences simultaneously, which will experience different phase steps, which co=
uld then be used to solve for the length of the data gap. However such a fu=
nction is not quite trivial and not available in SpecLab.


The ambiguity is conveniently resolved using 1pps pulses from a GPS, allowi=
ng correct identification and elimination of glitches which are shorter tha=
n +-0.5 s. In addition with a PC clock, the 1 pps pulse can be allocated to=
 it's time of day, so absolute phase can be retrieved on all frequencies af=
ter arbitrarily long outages or after a restart. If PC time cannot be kept =
within +-0.5 s of UT (eg. due to non-availability of NTP time, or multiple-=
second latency variation on an Icecast stream), serial time from the GPS (N=
MEA) can be fed to SpecLab to identify the pps second.=20


However while 1pps can be superimposed on antenna signals and eliminated a =
posteriori by the noise blanker, feeding NMEA to the soundcard requires ded=
icating a second soundcard channel solely for this purpose. This is the mai=
n reason why only one of two loops is available on Stefan's tree-grabber. T=
o solve the issue, several mnonths ago I have found and developed a reliabl=
e way to identify the pps second using signals available from the antenna, =
i.e. either the 3.6 s Alpha sequence, or the DCF77 minute marker (which is =
conveniently downsampled by the soundcard to 2*48-77.5 =3D 18.5 kHz). I sti=
ll haven't given up the hope that Stefan may some day decide to make use of=
 these scripts...


Best 73,
Markus


-----Urspr=C3=BCngliche Mitteilung-----=20
Von: Jacek Lipkowski <sq5bpf@lipkowski.org>
An: rsgb_lf_group <rsgb_lf_group@blacksheep.org>
Verschickt: So, 26. Feb 2017 8:53
Betreff: LF: ebnaut & speclab question


Is it possible to decode an ebnaut transmission using only the sample rate=
=20
correction using external VLF transmitters in Spectrum Lab?

VY 73

Jacek / SQ5BPF
------=_Part_74184_2034344039.1488101661065
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;">Hi Jacek,<br></div><div style=3D"color: black; font-fami=
ly: arial,helvetica,sans-serif; font-size: 10pt;">short answer: </div><div =
style=3D"color: black; font-family: arial,helvetica,sans-serif; font-size: =
10pt;"><br>Yes (if you have a soundcard without dropouts on the input). It =
will work on many modern PC's. For example, RN3AUS has achieved successful =
decodesd from DK7FC using samplerate sync to GBZ.</div><div style=3D"color:=
 black; font-family: arial,helvetica,sans-serif; font-size: 10pt;"><br>Long=
er answer:</div><div style=3D"color: black; font-family: arial,helvetica,sa=
ns-serif; font-size: 10pt;"><br>The samplerate detector will track the freq=
uancy and the phase of the reference signal. However if a buffer overrun oc=
curs, an unknown number of samples may be missing in the data stream. Then =
the phase of the reference frequency will be reconstructed, but the phase o=
n other frequencies will still be wrong. For example, if you are using a 10=
 kHz reference, a 1 ms dropout will swallow exactly one cycle, so the SR de=
tector will simply not see it. On the other hand, the same 1 ms step will i=
ntroduce a 97.2=C2=B0 phase step on 8270 Hz, which would ruin an EbNaut dec=
ode. With a non-integer kHz reference (e.g. 22.1 kHz), a phase change would=
 be detectable on the reference, but the correction would still not scale c=
orrectly to the receive frequency. So&nbsp; to be able to apply a correctio=
n to any frequency, we need to know the absolute duration of the outage.</d=
iv><div style=3D"color: black; font-family: arial,helvetica,sans-serif; fon=
t-size: 10pt;"><br>One theoretical solution would be to track multiple diff=
erent off-air references simultaneously, which will experience different ph=
ase steps, which could then be used to solve for the length of the data gap=
. However such a function is not quite trivial and not available in SpecLab=
.</div><div style=3D"color: black; font-family: arial,helvetica,sans-serif;=
 font-size: 10pt;"><br>The ambiguity is conveniently resolved using 1pps pu=
lses from a GPS, allowing correct identification and elimination of glitche=
s which are shorter than +-0.5 s. In addition with a PC clock, the 1 pps pu=
lse can be allocated to it's time of day, so absolute phase can be retrieve=
d on all frequencies after arbitrarily long outages or after a restart. If =
PC time cannot be kept within +-0.5 s of UT (eg. due to non-availability of=
 NTP time, or multiple-second latency variation on an Icecast stream), seri=
al time from the GPS (NMEA) can be fed to SpecLab to identify the pps secon=
d. </div><div style=3D"color: black; font-family: arial,helvetica,sans-seri=
f; font-size: 10pt;"><br>However while 1pps can be superimposed on antenna =
signals and eliminated a posteriori by the noise blanker, feeding NMEA to t=
he soundcard requires dedicating a second soundcard channel solely for this=
 purpose. This is the main reason why only one of two loops is available on=
 Stefan's tree-grabber. To solve the issue, several mnonths ago I have foun=
d and developed a reliable way to identify the pps second using signals ava=
ilable from the antenna, i.e. either the 3.6 s Alpha sequence, or the DCF77=
 minute marker (which is conveniently downsampled by the soundcard to 2*48-=
77.5 =3D 18.5 kHz). I still haven't given up the hope that Stefan may some =
day decide to make use of these scripts...</div><div style=3D"color: black;=
 font-family: arial,helvetica,sans-serif; font-size: 10pt;"><br>Best 73,<br=
>Markus<br><br><br>-----Urspr=C3=BCngliche Mitteilung----- <br>Von: Jacek L=
ipkowski &lt;sq5bpf@lipkowski.org&gt;<br>An: rsgb_lf_group &lt;rsgb_lf_grou=
p@blacksheep.org&gt;<br>Verschickt: So, 26. Feb 2017 8:53<br>Betreff: LF: e=
bnaut &amp; speclab question<br><br></div>Is it possible to decode an ebnau=
t transmission using only the sample rate <br>correction using external VLF=
 transmitters in Spectrum Lab?<br><br>VY 73<br><br>Jacek / SQ5BPF<br><br><b=
r></div>
------=_Part_74184_2034344039.1488101661065--