AGC Detector Circuit: How Do D2, D3, D4 Affect Transistor Collector Voltage Drop?

Hello to everybody reading this question! I am trying to figure out how does the circuit i attached at the bottom of this post works. I guess that , from the input we have a diode envelope detector with R47 and C79 setting the time constant. My question is how does the colector voltage of the transistor drop , what is the limit of this drop and what is the role of those 3 diodes (D2,D3,D4)

Try to install LTspice to simulate that circuit, could be a good help Please put here the simulations to discuss

In the first picture is the circuit in Pspice. In the second is "AGC" for 100mV ,500mV,1V,3V,5V

With no input signal, Q16's collector current is zero. As signal peaks rise above 0.9 volts, Q16 starts to draw collector current. More signal gives more collector current and lowers AGC_Out.
The 1N270 germanium diode is likely obsolete.
D3 prevents Q16's collector voltage rising above 6.7 volts. D4 prevents AGC_Out voltage from rising above 6.7 volts. D2 prevents C81 being discharged by resistance between AGC_Out and ground.
What is the part number of the dual-gate MOSFET's you are using? I used long-obsolete RCA 40673's in radios thirty years ago.

Thank you very much for your support , Peter ! I am using 2 J310 JFET transistors in casocode configuration  instead of a dual gate MOSFET because it is kinda hard to find a dual gate these days. The 0.9 voltage peak is the VBE voltage of Q16 + the voltage that drops across the diode i guess, right ?Another question is what is exactly the role of those two 1M resitors R49/R50 ? And how does changing those values influence my circuit ?

Referring to your Pspice schematic (I first used Pspice 25 years ago), I may misread a part-designator, as the image is small.R22 provides collector current to Q1 and charging current to C23. R21 does this and also keeps diode D12 or D11 turned on. C23 delays the gain increasing when the signal goes away. Changing R22 or R21 or C23 will change the time delay.I recommend you think through the circuit behavior manually by thinking where the currents flow.What frequency are your J310's amplifying?