To: rsgb_lf_group@blacksheep.org
References: <1601c0521a8-1726-4ca9@webjas-vae030.srv.aolmail.net>
 <3ae286fb-7ba7-8683-62f2-3e230e9b4ccf@abelian.org>
 <76ec2250-2206-1c8d-b775-d68f8978cacb@abelian.org>
 <8f91a741-a2da-296f-6410-83d4002483b5@tele2.se>
From: Paul Nicholson <vlf0403@abelian.org>
Message-ID: <27d67172-8512-8661-ffa5-b304b8f64452@abelian.org>
Date: Sun, 3 Dec 2017 14:34:32 +0000
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:52.0) Gecko/20100101
 Thunderbird/52.3.0
MIME-Version: 1.0
In-Reply-To: <8f91a741-a2da-296f-6410-83d4002483b5@tele2.se>
Content-Language: en-US
Subject: Re: LF: W4DEX at 8270 Hz - and VO1NA
Content-Type: text/plain; charset=utf-8; format=flowed
Content-Transfer-Encoding: 7bit
Sender: owner-rsgb_lf_group@blacksheep.org
Precedence: bulk
Reply-To: rsgb_lf_group@blacksheep.org


Two T/A signals at the same time, and on almost the same
bearing:

  VO1NA 280.1 deg at 3575.4 km

  W4DEX 284.6 deg at 6194.2 km

This couldn't have happened any better if we'd planned it.
Perfect timing by Dex to come on to 8270 when he did.

Markus wrote:

 > The directional lobe seems rather sharp

The bearing of each Fourier bin is calculated (mod 360) and
any bins which have bearing outside a +/- 20 degree band around
the chosen azimuth, are suppressed by setting their magnitude
to a small value.   This is the 'filtered bearings' part of it.

That gives the impression of sharpness as you swing the compass.

The bins which survive the filtering then have their magnitudes
re-calculated by synthesizing a cardioid aimed on the bin's
bearing.  Thus we obtain some extra gain in the forward direction
and a null in the opposite direction.

It is possible to run without the bearing filter stage, then you
just get the cardioid response.  This may be better in some
situations.  I've got a slicker version of the spectrogram
in the pipeline and that will have the option to filter bearings
or not.    But I won't try to finish that until I've upgraded
the H receiver.

 > What is the latency between the end of a one-hour buffer and
 > the time the latest spectrogram comes online?

About an hour.

 > are you still using three separate stereo soundcards with
 > individual timing corrections,

Still three M-Audio 192 PCI cards 24 bit 192k/sec. Each has
VLF signal into one channel and a shaped PPS into the other.
Three instances of vtcard and vttime, and a vtjoin to align
them by timestamp.  The result is a 3-channel stream.

Various things T-off that.   A vtresample to 48k/sec and
a vtwrite records /raw continuously and the spectrogram
dips into /raw with vtread to grab one hour segments.

 > Could downconvert-and-decimate be used to speed up FFT
 > calculations for the narrow spectrogram?

The one hour segments of 3-axis stream are passed through a
vtfilter which applies the phase EQ for each channel, then
through the sferic blanker, then mixed down and resampled
to 240 samples/sec before the bank of Goertzel filters.
I don't think it would be any quicker to replace the Goertzel
bank with a fast Fourier transform.

I'm very keen to eliminate the three PCI sound cards and replace
them with one capture device but so far I haven't found one
good enough.

--
Paul Nicholson
--