Message-ID: <002401beecef$c01fc900$130a063e@default>
From: "Alan Melia" <Alan.Melia@btinternet.com>
To: "rsgb_lf_group" <rsgb_lf_group@blacksheep.org>
Subject: LF: Comparison of FFTDSP4 and Specrogram 5.09 on QRS
Date: Sun, 22 Aug 1999 23:41:54 +0100
MIME-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1; format=flowed
Content-Transfer-Encoding: 8bit
Precedence: bulk
Reply-To: rsgb_lf_group@blacksheep.org
Sender: <majordom@post.thorcom.com>

             COMPARISON OF FFTDSP4y AND SPECTROGRAM 5.09
                          with some comments on HAMVIEW

Having installed what I think is probably the latest of the Spectrogram
upgrades, 5.09, I have arranged the two aux. audio outputs from my receiver
to go to the sound cards in 2 separate PCs. One runs FFTDSP4 under DOS 5,
the other runs Windows 95 and Spectrogram. Both use genuine Creative Labs
SB16 cards. I had an opportunity a few days ago to watch the effect of
severe QRN as I monitored Geri DK8KW through a local thunder storm. On that
occasion Geri would have been copiable on either system, but the effects of
the lightening crashes chopped him up very badly on FFTDSP4. It is possible
that a different setting of the colour scale would have improved things, but
unfortunately either the version I have has a small bug or there is an
incompatibility with my hardware. If I select a manual colour scale and an
integration time of greater than 1, the screen does not scroll when it
reaches the bottom. It does work well on AUTO. Geri would have been 'M'
verging on 'T' on FFTDSP4, but was a perfectly readable 'O' on 'gram with
the 4096 point display (similar resultion). For some reason Geri's morse
elements were readable right through the crashes (from strikes with 2000m!)
This exercise was they first real use of 'gram in anger and it showed up
well. I determined to leave the connection (10m of RG-58 allong the passage
from the rx to the main room!) to the Windows 95 machine and compare the
ultimate performance on a weak QRS signal.

On Saturday morning early, Mike G3XDV called Gerd DJ5BV , no trouble reading
Mike of course, but I had not heard a signal from Gerd before. FFTDSP4
showed a signal, the noise level was low and there were no static crashes,
but the signal was hardly readable. It was bobbing in and out of the noise.
I would not have been sure of the call if I hadn't heard Mike calling.
Spectrogram was set up for 16K points 60dB range and an averaging of
4....and there was a beautiful solid signal from Gerd at a full 'O' with no
compromise.  Later in the morning after Mike had called QRZ I went snuffling
around in the noise looking for the signal and realised that some weak lines
I could see at about 7Hz spacing were the dreaded Loran sidebands (abt 2-3dB
above the noise) I'd heard so much about. It is the first time I have seen
them, they are not visible on FFTDSP4, even with the 500Hz filter in the rx
and the highest resolution. ( It looks as thought the rx hardware is working
well, what I need to do now is put the same amount of effort into sorting
the aerial out and getting better sensitivity!)

In conclusion, I still like FFTDSP4 as an operating aid. I find that on a
full scan 200-2500Hz I can quickly locate a new station calling and
concentrate the hardware on it. I have a paper template stuck to the bottom
edge of the monior screen as there is no offset capability allowing the
display to indicate the correct frequency (like there is on 'gram). I can
see stations on hand keyed speeds that I cannot hear in the phones. I can
see where the QRM is and shift the filter away to to listen to a weak
station. Ok, a good CW operator could do this without the aid but it is nice
to see what is happening ( and I am not a real morse operator!). I leave my
RX tuned to 138.10 kHz LSB and use the pass-band tuning to 'isolate' the
signal and BFO shift on the AOR 7030 to adjust to a listenable tone. It is
posible to have 3 pre-programmed filter 'positions' for the 500Hz filter
within the 2.5kHz view by using the filter select, and the CW and DATA
modes.

I would normally use Spectrogram on the 4096 point resolution set to cover
the top 300Hz of the band, switching to a 16K point resolution and adjusting
the position of the display to catch the required signal for weak signals I
find averaging between 2 and 10 useful. With a timing of 400mS for 3 sec
dots. Sampling is left at 5K and range at 60dB with the standard colour
palate. I have not found ,personally, that changing the colour range of the
display makes any difference to the visibility of signals on the edge of the
noise. It would seem as though Spectrogram has up to about a 10dB advantage
over FFTDSP4, this is probably accounted for by the increased resolution of
Spectrogram 0.3Hz against 2Hz for FFTDSP4.

I also have a copy of Hamview, which I have used for a time. Hamview has the
interesting possibility of being able to drop a very narrow filter, or a
bandpass filter defined by clicking the mouse, over a signal and listen in
the computer speakers to the (delayed) filtered signal (It is essential to
turn the rx speaker down or it gets VERY confusing.) It produces a dispay
that wraps horizontally around the screen rather than scrolling and will
display about 7 minutes of signal in the highest resolution mode. The
program does not run well on my system, crashing with a variety of
breakpoint numbers, after anything from 2 mins to half an hour. I have run
FFTDSP4 for 12 hours using the longtem logging (25Mbyte file) on many
occasions. Hamview does enable a display file to be logged and a WAV file
(these are enormous after a few hours) to be stored for replay and
filtering. Its major disadvantage for QRS work is that the highest
resolution (about 2Hz I think) produces a scan of 4kHz and there is no
facility to be able to tune the averaging to suit the signal. It is about
the same as FFTDSP4 on sensitivity. I might use it more if I had a version
that didn't crash on my hardware. It has taught me that audio filters, no
matter how good, will not put a weak signal out of the noise and make it
more readable! Narrow IF filters do help but not as much as I used to
believe. Audio filters are good for reducing the effect of nearby qrm on
signals that are otherwise readable. I have found the notch of an external
Daiwa filter ,most useful for removing the fatigue of listening to the
screech of the Greek RTTY station at night.

On the use of filters I have found that putting the required signal near the
edge of my 500Hz IF filter, sometimes makes the tone of a weak signal stand
out more from the sound of the
filtered noise. Is this a normal technique known to morse men?

Alan Melia G3NYK
Alan.Melia@btinternet.com