Should you use AGC in FSK receiver?

In a FSK receiving system(superhetrodyne) after first IF is it better to use AGC and apply a constant power to IF2 and limiter,quadrature modulator sections.(with SA636 IC)?

if followed by a limiter, i don't see why you would use an AGC unless you are using it to control the front end of the receiver.

I could probably think up a reason to have a front end AGC, but it would be a real stretch! Maybe if you can only do minimal bandpass filtering up front and do not want stray intermods coming into band from strong interferers.

In an event, except for those unique situations, a simple limiter at the end is all that is needed.

If you have the option, you want to do the final IF bandpass filtering as close to the limiter as possible, for "click noise" reasons.

In FSK superheterodyne receiver,
It is better to use AGC in some applications.
If you want to limit your first stage LNA output then you need AGC.
Set a threshold & when your output crosses threshold change the gain of amplifier(LNA).

the Third-order input intercept point of SA636 RF Input is -10dBm,SA636 is IF2 and limitter of the receiver,I hesistae to add AGC after IF1 and before IF2 to apply a known maximum signal level (ex. -25dbm) to SA636,is this method ok?

one another is add AGC at RF section before first mixer(IF1) and after a 15dB gain LNA,is this method better?

What is distance between your transmitter & receiver?
Why dont you try different IC such as TDA 5211 or 5210 from Infenion for your application?
They have inbuilt LNA which will control your Gain with respect to your input power?

one another is add AGC at RF section before first mixer(IF1) and after a 15dB gain LNA,is this method better?

To find the better method, have a budget calculations for both the methods. find which one is higher IP3 etc.

I want to do AGC at first RF stage before first mixer.

I want to use this config:

Low Loss filter + 15dB LNA(NF=1,OIP3=35dbm) + 15 db LNA(NF=1,OIP3=35dbm) + 40dB voltage Variable Att.

I detect the output of ATT and control its attenuation.

Is this have problem?

I have not noticed any problem as I do not have the mixer data. So If you are sure enough that the maximum input signal will be handled by the attenuator( takes few u Sec to milli second for detect and correct)and no impact on mixer input level.