DomainKey-Signature: a=rsa-sha1; c=nofws;
        d=googlemail.com; s=gamma;
        h=mime-version:in-reply-to:references:date:message-id:subject:from:to
         :content-type:content-transfer-encoding;
        b=YW6CA73g+cKc1JbiBMIZgfOSKEiytsat19KDCgftEMilWe7yv0rdJLzEJRSBpC64SZ
         YOZQBagj53VEWBy+9CwQD4l9hQ5ThkUCGIy5KsBHUCHnnwTLjOLbj5NvRnfnW7jarLUs
         pUHIzrvFd5kwLWmR4KXOtt7OauTSmpT1VEaWU=
MIME-Version: 1.0
In-Reply-To: <7.0.1.0.1.20090305091934.0181aca8@magma.ca>
References: <000a01c99a91$a54a1180$1402a8c0@e7010>
	 <FA25C6BCA43E4B89B3C17DA7F30F45EE@DR2>
	 <7.0.1.0.1.20090305091934.0181aca8@magma.ca>
Date: Thu, 5 Mar 2009 16:01:52 +0000
Message-ID: <fc7eccec0903050801o1fb3b07fk220e1faec2e4e291@mail.gmail.com>
From: Andy Talbot <andy.g4jnt@googlemail.com>
To: rsgb_lf_group@blacksheep.org
DomainKey-Status: good (testing) 
Subject: Re: LF: Deep copy...
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable
Sender: owner-rsgb_lf_group@blacksheep.org
Precedence: bulk
Reply-To: rsgb_lf_group@blacksheep.org


The effect of AGC is even worse than amplitude modulation, it can add
a non linear element to the signal processing.
The amplitude of the receiver output is proportional to the input
power multiplied by a factor (an attenuation) that depends on the
near-enough instananeous signal level.  So Pout =3D G * Pin * K.* Pin
where G is Rx gain with no AGC, and  K is the AGC function, which
we''ll assume for now is linear (it isn't but...)

So, for this very simplistic Rx,    Pout =3D Pin^2 {G.K}.   Since K
won't be simple and linear in practice, even higher power terms in Pin
will appear, including teh odd order ones that give rise to close in
intermod products.

So, for signals that are modulated, and / or vary at a rate slower
than the AGC action, we will see intermodulation products, harmonics
and propably other cross-modulation effects as well.

I have seen very pronounced AGC induced intermod products when doing
dopplergram plots of HF signals.  Where two very-nearly-identical
signals were present via multiple propagation paths, sidebands could
clearly be seen either side of the pair due to AGC action in the
receiver following the fading that was the apparant effect of the two
signals combining.

Andy  G4JNT
www.g4jnt.com



2009/3/5 Bill de Carle <ve2iq@magma.ca>:
> At 09:25 PM 3/4/2009, you wrote:
>
> I seem to get best results while trying to pull signals out of the noise
> when the AGC of the receiver is OFF as well as the Noise Blanker and Noise
> reduction features being OFF...
>
> Do you concur??
>
> Also setting a good audio filter to the passband of interest seems to bypa=
ss
> some heavy static hits as well...
>
> I've noticed that when doing HF frequency measurement tests (working to th=
e
> nearest milliHertz) - turning off the AGC under high static conditions see=
ms
> to improve the accuracy of the measurement, at least with the software I
> use.=A0 One plausible explanation is that AGC action necessarily introduce=
s
> amplitude modulation (on all signals in the passband).=A0 When I process t=
he
> AGC'd signal with what amounts to a very narrow DSP filter the added
> amplitude modulation shows up as apparent sidebands close-in on the signal
> I'm trying to measure.=A0 If the power in those sidebands is comparable to
> that of the signal whose frequency I'm looking for, the FFT algorithm (whi=
ch
> assumes the real signal has the largest amplitude) gets confused and comes
> up with an estimated frequency somewhere between the correct value and tha=
t
> of a nearby sideband.=A0 The effect is small however because the AGC pumpi=
ng
> action doesn't occur very fast so the added sidebands are seen to be only
> some milliHertz away from the signal.=A0 The sidebands occur on both sides=
 of
> the "real" signal, so one might expect them to cancel out but in practice
> they don't because the amount of error depends on where the "real" signal
> falls with respect to the fixed frequency bins of the FFT.=A0 It should be
> possible to model the AGC action of a particular receiver and compensate f=
or
> it in the software.=A0 No doubt the phenomenon becomes less significant wi=
th
> smaller FFT's or shorter integration times.=A0 Changing the AGC setting
> between SLOW-FAST-OFF might help under some conditions.=A0 With QRSS-60
> signaling rates it can take a long time to find out which setting is
> optimum, especially when band conditions are changing or the QSB period is
> close to a bit time, hi!
>
> VE2IQ
>