Look carefully those input circuits are the same. The only difference is NPN vs PNP input transistor.
In circuit from #19 fig3 the Q1 DC current start flowing from Vcc --- TR2 (MJ480) --->R3----> enters emitter of a 3906----> 8.2K to GND.

Now for the circuit from post 16 we have this path for DC current:
Vcc--->R2---->VT1---->R6--->R9---->Q2---GND.

PS I find another example
http://www.zen22142.zen.co.uk/Circuits/Audio/2wamp.html

 

Look carefully those input circuits are the same. The only difference is NPN vs PNP input transistor.
In circuit from #19 fig3 the Q1 DC current start flowing from Vcc --- TR2 (MJ480) --->R3----> enters emitter of a 3906----> 8.2K to GND.

Now for the circuit from post 16 we have this path for DC current:
Vcc--->R2---->VT1---->R6--->R9---->Q2---GND.

The reason I already mentioned, if you doing that then the R5 and C5 were quite strange when they at that position.
Please check the structure of amplifier, it's quite normal.

The circuit was linked from the worksheets of AAC, from here.
Another -- Design of a Three-Stage BJT Amplifier.

Please give me some other examples as what you said from internet.

 

Look here
http://www.allaboutcircuits.com/vol_3/chpt_9/3.html
And again we have exactly the same input stage structure. The Q1 emitter DC quiescent current path to ground is via R6-->R10-->Q4-->GND.
And R2 (22Ω) together with R6 (2.2kΩ) sets the closed loop voltage gain. C3 is used to reduces the DC gain to one.
Exactly the same as in circuit from post #16 and #19

 

The problem, as I see it, is that rio found or was given the circuit and they seemed to populate it with some randomly chosen parts (that is why all the resistors are 1 kOhm).

rio has the design parameters for the circuit... but so far I have not seen any actual attempt to DESIGN the circuit to meet the design parameters. So all resistors are the same, all capacitors are the same and there are some transistors and a diode. Don't ask me why diode is there because I have no idea.

At this point rio is plugin in random parts and seeing what happens. They can spend the next hundred years or so trying to create a circuit that will meet the design parameters given to them.

 

Look here
http://www.allaboutcircuits.com/vol_3/chpt_9/3.html
And again we have exactly the same input stage structure. The Q1 emitter DC quiescent current path to ground is via R6-->R10-->Q4-->GND.
And R2 (22Ω) together with R6 (2.2kΩ) sets the closed loop voltage gain. C3 is used to reduces the DC gain to one.
Exactly the same as in circuit from post #16 and #19

Thanks Jony.
Except this circuit, I can't find the same circuit or similar from internet, I think i need to find some more about the circuit to study.