Hi,
I have a pre-designed audio power amplifier to analyse and test. The circuit is attached.

I have got several values but during test some of them were not quite right.

My results
__________

differential (input) stage of about (-) 65.2 - that is for Q1
second stage gain of about 1224 - that is for Q4
feedback ratio, about 1/25 i.i 0.04
Overall gain of about 26

Input impedance of about 4.4KΩ
Output impedance of less that 1Ω

Low frequency limit of about 32Hz due to C2
High frequency limit of about 133kHz due to C3


at the output stage, we assume a load of 8Ω
Maximum operational output voltage was about 3V peak (positive)

which gave me an operational power delivered of about 0.56W

Power supply is about 1W
so effieciency gives me about 56%

The maximum power dissipated at each output transistor was about 0.51W
So heatsink not really needed.


TESTING
________
When testing my amplifier, The low and high frequencies limits were not right. I got about 25Hz and 4MHz for low and high frequency limits.

Can you please help me check my results. Especially for high and low frequency limits, and also power analysis?

 

The voltage divider of R5 and R6 create as voltage at point A of only 0.48V which is not enough to turn on Q3 and Q5 so the amplifier WILL NOT WORK.

 

The voltage divider of R5 and R6 create as voltage at point A of only 0.48V which is not enough to turn on Q3 and Q5 so the amplifier WILL NOT WORK.

With Vcc = 9V, the voltage divider of R5 and R6 creates a voltage of about 1.86V.

The amplifier work.

 

Low frequency limit of about 32Hz due to C2

Why you "skipped" C5?
F1 = 0.16/(C2*R8) = 33Hz and F2 = 0.16/(C1*Rin) = 1.8Hz and F3= 0.16/(C5*RL) = 20Hz
so the Fd ≈ 1.1 * √(F1^2 + F3^2) = 42Hz. Simulation in LTspice show similar result.
And are you sure that in real life you use C2 = 22μF?
Because if C2 >> 22uF then Fd approaching to F3= 0.16/(C5*RL) = 20Hz

High frequency limit of about 133kHz due to C3

How do yo get this result? Maybe this is a open loop high frequency limit?

 

And are you sure that in real life you use C2 = 22μF?
Because if C2 >> 22uF then Fd approaching to F3= 0.16/(C5*RL) = 20Hz

I used a 22uF, and I had about 25Hz when measuring on CRO.

How do yo get this result? Maybe this is a open loop high frequency limit?

I got the high frequency by applying Miller's theorem on C3. But it gave me about 133kHz. Which is different from about 4.5MHz observed on CRO. I believe my approach was not right. Do you have a way to calculate the high frequency limit pls?

 

I need value of a β for the BJT that you used in your small signal analysis.
Also what voltage you assume at R4 resistor maybe you assume 1V?
As for Fd I don't know why you get 25Hz.

 

I need value of a β for the BJT that you used in your small signal analysis.
Also what voltage you assume at R4 resistor maybe you assume 1V?
As for Fd I don't know why you get 25Hz.

I used an hfe of 330 for all BC547/BC557.
Voltage across is VA - Vbe = 1.86 - 0.7 = 1.06V.

 

Ok let's do some simplified approximation analysis.
Input stage gain is equal to

Av1 = (R3||rinQ4)/2re = (1.5K||1.2K)/14.85Ω = 666.666Ω/14.85Ω = 45[V/V] LTspice show 33[V/V]

rinQ4 = (hfe + 1)*re4 = 331 * Vt/Ic = 331 * 26mV/7mA = 331 * 3.71Ω = 1.23KΩ

2re = 4VT/Iee = 4*26mV/7mA = 14.85Ω

Second stage (VAS) voltage gain is equal

Av2 = (hfe+1*RL)/re4 = (91 *8Ω)/3.71Ω = 196[V/V] LTspice show 235[V/V]

I assume hfe = 90 for BD139

So the overall open loop gain is equal

Aol = AV1 * AV2 = 8.82K [V/V] LTspice show 7.78K [V/V]

Additional if we assume Cbe = 10pF and Cbc = 3pF we can find open loop roll-off frequency

F = 0.16/ ( R3||rin4 * (cbe + Av2*(cbc+C4) ) ) = 0.16/(666.7 *9.86nF) = 24.3KHz
And gain bandwidth product is equal to

GBP = 8.82K * 24.3KHz = 214MHz 71MHz in LTspice

So if close loop gain is equal to 1+R7/R8 = 26.5 the high frequency cut-off is equal to Fc = 214MHz/26.5 = 8MHz LTspice show 5.6MHz

As you can see our calculations are as good as our model we used.
And this is why in real life we add additional capacitor+resistor to ensure and clearly determine high frequency cut-off.