Re: [Discuss-gnuradio] AGC loop

[Eric Blossom]

On Thu, Sep 21, 2006 at 12:20:50PM -0400, Robert W McGwier wrote:
> Eric Blossom wrote:
> 
> >On Tue, Sep 19, 2006 at 01:36:55PM -0400, Robert W McGwier wrote:
> > 
> >>Tom:
> >>
> >>A good agc has at least two time constants.  One for attack and one for 
> >>decay.  Your attack is much too slow.
> >>
> >>tmp = (reference - sqrt(real(y)^2 + imag(y)^2));
> >>rate = rate1;
> >>if   tmp > gain    rate = rate2;
> >>
> >>gain += tmp*rate;
> >
> >Shouldn't this read:
> >
> >
> >>if   tmp > reference    rate = rate2;
> >>
> >            ^^^^^^^^
> >
> 
> I don't think so.   If your noisy  current observation, tmp,  which is 
> an instantaneous measurement on the needed gain to attain reference,  is 
> below the current smoothed state,  gain,  then you are in decay mode.  

Thanks.  Thinking about it as the instantaneous measurement of gain makes sense.

> If your instantaneous needed gain observation is less than the current 
> smoothed gain,  you are in the attack condition and a different time 
> constant on the smoothing is needed.  I am pretty sure what I told him 
> is right.  Does it make more sense now?
> 
> Bob

Eric