Roland AF-60 Bee Gee question

[Lino22]

Guys what is that cap for? They seem to have split the feedback resistor and inserted the 10u cap.

[antonis]

it's exactly for what you said..

Without that cap, there is both DC (bias) and AC (signal) feedback..
The former is desirable to stabilize the DC operating point but the later isn't desirable due to feedback resistor apparent value much lower than its actual (resistive) one due to Miller effect..
By splitting and decoupling the feedback resistor, signal sees only the resistor between Base and decoupling cap hence there is input impedance improvement..

e.g.
for a stage gain of x19 say, 940k undecoupled resistor is seen as 940k / (19 +1) = 47k, severely dominating stage input impedance..

P.S.
Of course, here might be reduntand 'cause input impedance is primarily dominated by grounded Emitter..

[Rob Strand]

In this case the devil is in the details.

It can also increase the gain.   The simplified gain is 2*470k/15k = 63.   But the transistor without feedback is going to produce a gain of 170.  Since 170 isn't enormous compared to 63 we don't expect to get 63.   With the cap the transistor input impedance will form a divider with the 15k.  So it's difficult to compare without getting to precise calculations.   It's close enough that in this case the gain increase will be small.

In fact perhaps the input impedance isn't that different. For example, 

bypass cap: collector at 4.5V => IC =4.5/15k = 300uA,   re = 26mV/IC = 87,  rpi ~ hFE * re = 300*87 = 26k.  Then rin ~ 15k + 26k = 39k.
no bypass cap (approx):   gain 172,  2*470k => 2*470k/172 = 5.4k,   rin = 15k + 5.4k = 20k

A secondary thing the cap does is reduce the hum from the power rails.
Here the details for a hFE of 300

Ckt                         Gain           Input Z
Bypass cap              34.5dB       37k
No bypass cap        29.5dB       22k

In reality not a big difference because the transistor is a non-ideal amplifier.

Hum from the supply rail is worse with the bypass cap because the gain is higher.

[antonis]

The simplified gain is 2*470k/15k = 63.  But the transistor without feedback is going to produce a gain of 170.  Since 170 isn't enormous compared to 63 we don't expect to get 63.

I'm pretty sure OP is trying to understand the above approximation..

@Lino22: Consider the stage above as an inverting amplifier (although transistor Base could hardly be considered virtual ground..)..
Gain formula is Ao / (1 + Ao*β), where Ao = Open-loop gain and β = feedback fraction (15/470+470).. For Ao = very high, closed-loop gain is simply 1/β (940k/15k=63)
For Ao = 170, closed loop gain is about 46..
(actually, is lesser due to other "imperefections"..)

P.S.
Although one should be tempted to place a cap series resistor to GND (T-network feedback) for preserving gain with simultaneous lowering feedback resistors values (for noise and phase margins reasons), it should be avoided for input impedance reasons (said above..)