Message-ID: <5A833C84.9000600@posteo.de>
Date: Tue, 13 Feb 2018 20:29:08 +0100
From: DK7FC <selberdenken@posteo.de>
User-Agent: Mozilla/5.0 (Windows; U; Windows NT 6.1; de; rv:1.9.1.8) Gecko/20100227 Thunderbird/3.0.3
MIME-Version: 1.0
To: rsgb_lf_group@blacksheep.org
References: <5A82E06D.23523.562F00A9@mike.dennison.ntlworld.com> <8DFB18F5-5CEE-45C3-8AB7-BD478AA62A30@hurontel.on.ca> <18A42D73771E4616AE008148DE17C78B@mal> <302E518F-74A4-4F3C-82AB-B5D547057DC3@hurontel.on.ca> <5A833B06.5050206@posteo.de>
In-Reply-To: <5A833B06.5050206@posteo.de>
Subject: Re: ULF: On the way to results... | vlfrx tools - advice needed...
Content-Type: multipart/alternative;
 boundary="------------010908000901040406030508"
Sender: owner-rsgb_lf_group@blacksheep.org
Precedence: bulk
Reply-To: rsgb_lf_group@blacksheep.org

This is a multi-part message in MIME format.
--------------010908000901040406030508
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 7bit

PS: I know, it would have been more clever to split the EbNaut messages 
into daytime and nighttime messages instead of choosing a 24 hour 
transmission time. And i know, it is more useful to concentrate on the 
carriers first and then see the day/night phase difference. Later i can 
correst the day/night phase difference to get better results in EbNaut. 
All this will be necessary if i want to have a decode of the 5 char 
message. So all this is a truely excellent tutorial to learn how to use 
vlfrx-tools :-) A tutorial with a reward in the end...

73

Am 13.02.2018 20:22, schrieb DK7FC:
> Hi Paul,
>
> Well, i'm on the way to produce the results from the data recorded on 
> the recent *970 Hz experiment*. Meanwhile i managed to actually decode 
> one of the EbNaut messages but it is quite a struggle. So that's the 
> first good news, but later more.
>
> Now, There is a problem with the E field reception: The trees were 
> higher than expected so the active E field antenna was not high enough 
> above them, i.e. it has a very low effective height. I knew about that 
> effect but i didn't think that it is so expressed: When looking on a 
> 'wideband' (the word developed during the kite experiments time) 
> spectrogram you can see that the noise floor is almost completely 
> given by the preamp noise floor, it almost looks like white noise and 
> some frequency response. One would not expect that the message can be 
> found in that kind of noise background but it is in there! So it could 
> have been much better...
>
> Here is a first image showing the E field in the upper part and the H 
> field in the lower part: 
> http://www.iup.uni-heidelberg.de/schaefer_vlf/ULF/E+H_first2days.png
> The spectrogram covers about the first 2 days of the recording, 
> without filters and blankers, just the raw recording.
>
> I learned how to apply a filter and a noise blanker and i played with 
> the threshold level. For the E field i found it is best at -a25. The H 
> field is best at -a10.
>
> First i concentrated on the 2 character EbNaut message. This message 
> was transmitted two times, so there was the chance to stack it. From 
> the E field i found the message at -5.0 dB and -6.5 dB. E field 
> stacked gives *-3.5 dB!* So far so good.
>
> Next i tried to decode the message from the H field. But the H field 
> has a problem too. It looks like is some lack of sensitivity and i can 
> also hear a Spanish broadcast station at night, i.e. some AM 
> demodulation. This also happens on the E field but just slightly.
> Anyway, i need to improve the ULF reception system and i need to find 
> a better location for the next experiment.
>
> The 2char message was found at -5.9 dB and -5.3 dB on the H field. 
> These two days stacked give *-2.8 dB!*
>
> Just for fun i stacked E and H now, ignoring the phase. And suddenly i 
> got a prediction of *-0.1 dB!!!!* This is quite an unconventional way 
> to combine E and H i guess :-) But i put this stacked file to the 
> decoder and tried a few variations with the -a option. A screenshot is 
> in attachment but a summary of the results follows later. So i decoded 
> the 2 character message with _*Eb/N0 = 0.2 dB*_.
>
> -----
>
> So far so good. But now i'm trying to combine E and H before applying 
> the noise blanker. This should help because when building a cardioid 
> with a good F/B, the sferics from the back side are rejected by say 30 
> dB, so the noise blanker will not be triggered when the E, H amplitude 
> weighting is set correctly. I already played a bit arround with vtmix 
> -c1,1 or vtmix -c1,j or vtmix -c1,-1 (1st channel is E, 2nd channel is 
> H) and watched how the result changes. But if i want to get the best 
> results i will first have to remove the mains hum from the H field 
> channel. Mixing E and H by using vtmix may be easy when both antennas 
> work well but in this special situation the H field has a high hum 
> component whereas the E field is almost white noise. When mixing first 
> and applying a hum filter then, i think the result gets worse.
>
> I would tend to select E or H and apply a filter and noise blanker and 
> then save this to a file, i.e. one for E, one for H. Next i would need 
> to combine/mix these files again. Will the phase and timing still be 
> as in the original file or will it change, depending on the filter 
> poles and so on?
> How can i mix E and H again from separate files? Is it done with 
> vtjoin? Do i have to write a script to do that? Starting to read both 
> files and send the output to a buffer. Then i call both buffers, join 
> them, and save them into a 2-channel file again? Then i can read that 
> preapred 2 channel file and use vtmix to build the cardioid antenna, 
> as well as the resampling to SR=1, converting to base-band etc. to 
> generate the file which can then be moved to the EbNaut decoder?
> Sounds quite complex.
> Do you have a better idea?
>
> 73, Stefan
>
> PS: Still not watched for the carrier, i guess it will be detectable...

--------------010908000901040406030508
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: 7bit

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>
  <meta content="text/html; charset=UTF-8" http-equiv="Content-Type">
</head>
<body bgcolor="#ffffff" text="#000000">
PS: I know, it would have been more clever to split the EbNaut messages
into daytime and nighttime messages instead of choosing a 24 hour
transmission time. And i know, it is more useful to concentrate on the
carriers first and then see the day/night phase difference. Later i can
correst the day/night phase difference to get better results in EbNaut.
All this will be necessary if i want to have a decode of the 5 char
message. So all this is a truely excellent tutorial to learn how to use
vlfrx-tools :-) A tutorial with a reward in the end...<br>
<br>
73 <br>
<br>
Am 13.02.2018 20:22, schrieb DK7FC:
<blockquote cite="mid:5A833B06.5050206@posteo.de" type="cite">
  <meta content="text/html; charset=UTF-8" http-equiv="Content-Type">
Hi Paul, <br>
  <br>
Well, i'm on the way to produce the results from the data recorded on
the recent <b>970 Hz experiment</b>. Meanwhile i managed to actually
decode one of the EbNaut messages but it is quite a struggle. So that's
the first good news, but later more.<br>
  <br>
Now, There is a problem with the E field reception: The trees were
higher than expected so the active E field antenna was not high enough
above them, i.e. it has a very low effective height. I knew about that
effect but i didn't think that it is so expressed: When looking on a
'wideband' (the word developed during the kite experiments time)
spectrogram you can see that the noise floor is almost completely given
by the preamp noise floor, it almost looks like white noise and some
frequency response. One would not expect that the message can be found
in that kind of noise background but it is in there! So it could have
been much better...<br>
  <br>
Here is a first image showing the E field in the upper part and the H
field in the lower part:
  <a moz-do-not-send="true" class="moz-txt-link-freetext"
 href="http://www.iup.uni-heidelberg.de/schaefer_vlf/ULF/E+H_first2days.png">http://www.iup.uni-heidelberg.de/schaefer_vlf/ULF/E+H_first2days.png</a><br>
The spectrogram covers about the first 2 days of the recording, without
filters and blankers, just the raw recording.<br>
  <br>
I learned how to apply a filter and a noise blanker and i played with
the threshold level. For the E field i found it is best at -a25. The H
field is best at -a10.<br>
  <br>
First i concentrated on the 2 character EbNaut message. This message
was transmitted two times, so there was the chance to stack it. From
the E field i found the message at -5.0 dB and -6.5 dB. E field stacked
gives <b>-3.5 dB!</b> So far so good. <br>
  <br>
Next i tried to decode the message from the H field. But the H field
has a problem too. It looks like is some lack of sensitivity and i can
also hear a Spanish broadcast station at night, i.e. some AM
demodulation. This also happens on the E field but just slightly.<br>
Anyway, i need to improve the ULF reception system and i need to find a
better location for the next experiment.<br>
  <br>
The 2char message was found at -5.9 dB and -5.3 dB on the H field.
These two days stacked give <b>-2.8 dB!</b><br>
  <br>
Just for fun i stacked E and H now, ignoring the phase. And suddenly i
got a prediction of <b>-0.1 dB!!!!</b> This is quite an unconventional
way to combine E and H i guess :-) But i put this stacked file to the
decoder and tried a few variations with the -a option. A screenshot is
in attachment but a summary of the results follows later. So i decoded
the 2 character message with <u><b>Eb/N0 = 0.2 dB</b></u>. <br>
  <br>
-----<br>
  <br>
So far so good. But now i'm trying to combine E and H before applying
the noise blanker. This should help because when building a cardioid
with a good F/B, the sferics from the back side are rejected by say 30
dB, so the noise blanker will not be triggered when the E, H amplitude
weighting is set correctly. I already played a bit arround with vtmix
-c1,1 or vtmix -c1,j or vtmix -c1,-1 (1st channel is E, 2nd channel is
H) and watched how the result changes. But if i want to get the best
results i will first have to remove the mains hum from the H field
channel. Mixing E and H by using vtmix may be easy when both antennas
work well but in this special situation the H field has a high hum
component whereas the E field is almost white noise. When mixing first
and applying a hum filter then, i think the result gets worse. <br>
  <br>
I would tend to select E or H and apply a filter and noise blanker and
then save this to a file, i.e. one for E, one for H. Next i would need
to combine/mix these files again. Will the phase and timing still be as
in the original file or will it change, depending on the filter poles
and so on?<br>
How can i mix E and H again from separate files? Is it done with
vtjoin? Do i have to write a script to do that? Starting to read both
files and send the output to a buffer. Then i call both buffers, join
them, and save them into a 2-channel file again? Then i can read that
preapred 2 channel file and use vtmix to build the cardioid antenna, as
well as the resampling to SR=1, converting to base-band etc. to
generate the file which can then be moved to the EbNaut decoder? <br>
Sounds quite complex. <br>
Do you have a better idea? <br>
  <br>
73, Stefan<br>
  <br>
PS: Still not watched for the carrier, i guess it will be detectable...<br>
</blockquote>
</body>
</html>

--------------010908000901040406030508--