Content-Transfer-Encoding: 8bit
Message-ID: <1.5.4.32.19990806013130.005fee98@pop.sci.fi>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Date: Fri, 06 Aug 1999 04:31:30 +0300
To: rsgb_lf_group@blacksheep.org
From: "Paul Keinanen" <keinanen@sci.fi>
Subject: Re: LF: eclipse
Precedence: bulk
Reply-To: rsgb_lf_group@blacksheep.org
Sender: <majordom@post.thorcom.com>

At 23:02 5.8.1999 +0200, Johan Bodin wrote:
>Hello,
>
>Forwarded from Rik Strobbe:
>
>>A software team from the Belgien amateur radio society UBA has developed a
>>programme that can be used to register the fieldstrength measurements. All you
>>need is a PC with Win95/98 and a soundcard. The audio output of the receiver
>>is fed to the input of the soundcard.
>
>AGC (which is present in most amateur receivers) will make such measurements
>almost useless (unless the AGC can be disabled).

>From http://www.uba.be/zon/SoftGB.htm it is stated that the receiver is set
to  a constant gain, i.e. AGC off. 

In my opinnion, the problem is that the receiver gain should not fluctuate
much during a several day period. A typical IF amplifier strip does not use
feedback resistors to set the gain, but the gain depends mainly of the
transistor parameters, which are known to fluctuate a lot with temperature,
operating voltage etc., causing a gain variation of several decibels. Thus,
the receiver should be kept at constant temperature for several days and the
IF strip operating voltages well regulated in order to give reliable results.   

In order to compensate for the gain variation, in radio astronomy some form
Dicke radiometer is used, in which the receiver input is switched several
times a second between the antenna signal and a very well known noise sourse
(e.g. a resistor at 300 K or an auxiliar antenna pointing to a cold spot in
the sky). The receiver output is a square wave altering between the
reference level and signal of interest level and the difference between
these two levels are recorded.


Looking at the HF test signal for this experiment:
 
           a long dash (14 seconds) 
           CW-identification (25 wpm ON4UBA/B) 
           Idle period (5-6 sec)  

it would be very tempting to use the CW-id to compensate for any gain drift,
but unfortunately that would only give the difference between the received
beacon signal  (mark) and local background noise (space). Unfortunately, the
local noise level varies a lot, so this would only give a measure of the
signal to noise ratio, not the absolute signal level.

To get accurate measurements, a noise source of very precisely known output
would be required so that the receiver input could be switched several times
a second between the antenna and the noise reference. Unfortunately the
noise level at LF/MF/HF is so high that a resistor at 300 K would be useless
and some other form of calibrated noise source is required. A well shielded
pseudo random noise generator (CMOS or TTL) followed by an attenuator might
be useful.
 
If the a.g.c. can not be disabled, the IF-strip acts like an amplifier with
very fast gain drift :-). With fast enough switching between the reference
noise level and antenna signal, the a.g.c. time constant is so long that the
gain variation would be neglectable, thus the Dicke radiometer principle
could be used even if the a.g.c could not be disabled.

But as I said, using this method would require a calibrated noise source and
some extra hardware to switch the receiver input between the antenna and the
noise source, so it is understandable that this is not used in this experiment.

Paul OH3LWR