Because the AGC is supposed to be a closed feedback loop, the varying response of the upstream gain stage shouldn't have too much of an effect - provided that it isn't so sensitive at one end of the band that you run into loop stability issues, or so insensitive at the other end that you run out of control range.

The ARRL and RSGB bookshelves should have some books with chapters on the subject.

A related topic is gain distribution.

"...for an FM receiver amplifier..."

By the way, this AGC isn't for your FM amplified antenna project, right? The AGC is for a FM receiver, right? I just to make sure on that point.

  

Hi,

that´s right, the AGC is for the antenna amplifier, so the input of the FM receiver isn´t overloaded.

You are right, I think I´m having some instability issues with the closed loop. I´ll have to spend more time in the lab!

thanks.   

I was afraid of that.

In short, you can't have an AGC circuit in a wideband RF amplfier. Well you can if you want, but you probably won't like the results.

The issue is when you're listening to a weaker station at 90.1 MHz (for example), and you drive past an extremely strong station at 107.1 MHz (for example), and the antenna AGC backs off the antenna gain making the desired signal sink into the receiver's noise floor.

With AGC circuits, it's critical to ensure that the feedback signal is a product of the desired signal, AND ONLY the desired signal. And this means that the feedback control signal must be derived from the filtered IF signal within the receiver.

Even with IF-derived AGC, it's considered to be a receiver flaw if the receiver desensitizes due to signals that are close enough to affect the AGC, but outside the narrower downstream IF and AF filters. The operator is left wondering why the receiver is desensitizing. Good receivers take extremely measures to minimize this issue.

The better approach is to ensure that the antenna amplifier provides only minimal gain (proper gain distribution), and provides an exceptionally wide dynamic range (low noise combined with high limit for linearity).


There may be a moderate amount of antenna AGC that would be operationally acceptable. Where the designer assumes that the user is not likely to be listening to signals very close to the noise floor. But then one might as well reduce the antenna gain by that fixed amount. Otherwise the gain will go up and down seemingly randomly.

 

Me: "...you can't have an AGC circuit in a wideband RF amplfier."

To clarify this, you can certainly control a wideband RF amplifier, but the AGC control signal should be derived from downstream narrow band stages.