Thanks for your reply, Hero999!
1) Polarised capacitors can be used for AC coupling providing the DC bias is higher than the negative peak voltage. For example if you have a capacitor with a 1V DC bias and a 1V peak AC signal the minimum voltage across the capacitor will be 0V and the maximum voltage will be +2V.
Understood. so the question is, how we can tell that the DC bias imposed on the coupling cap is a forward bias which is big enough? In this case, for C3/C4 before o-amp input, why the circuit designer can assume that an enough forward bias is there? the input is come from JA or JB, which is connected the output of CD player, or MP3, or sound card. browsing the web, I also noticed the same polarity configuration, e.g., in <http://www.dself.dsl.pipex.com/ampins/webbop/opamp.htm>, all the input coupling caps is polarized the same way. i am guessing this may due to some characteristics of the op-amp chip which i am not familiar with (I will study these in consequence).
2) You're right audio amplifiers don't need a regulated supply. The supply to the NE5532 is regulated because it has an absolute maximum voltage rating of +/-22V so powering it from +/-25V would destroy it.
i also read in some article talking about PSRR of an op-amp, it seems that the basic facts is that the output voltage of an op-amp is also related to the voltage (or change of the voltage) of power supply (in a way related to PSRR somehow). so i guess that might be one of the reasons why op-amp needs a regulated power supply.
For the power-amp, i guess it would be suffice for me at the moment to remember that it does not require a regulated power supply (as compared to op-amp). i will give a closer look of the reasons when it crosses.
3) The resistors help to reduce the power dissipated by the regulators. C9 and C10 should really be after the resistors to help improve the quality of the supply to the regulator.
Check the pinout to the LM7912, it looks wrong to me.
So what would be the common practise in choicing the resistance values (as well as power ratings) of R11/R12, particularly in this case? i.e., why there are 100 ohm in this case?
For C9/C10, thanks for point this out. I will double check the PCB to verify that.
4) W1 forms a filter with R9, R10, C5, C6 and W2; it's a dual ganged potentiometer one side is for the left channel and the other is for the right.
Let's just look at one channel for now. The filter relies on two things:
The impedance of a capacitor decreases with increasing frequency.
The impedance of a resistor remains constant regardless of the frequency.
C5 and W1 are connected in parallel.
R9, C5 and W1 and W2 form a potential divider.
As the frequency increases the impedance of C5 decreases, therefore the output of the potential divider with increase with increasing frequency.
W1 damps the effect of C5, the lower W1's setting the less difference C5 makes, therefore W1 must be a bass boost control.
This is an excelent/detail description, which gives me some insights. i also found another brief comments on this circuit (talking about the same speaker), at <http://bbs.ustc.edu.cn/cgi/bbsanc?path=/groups/GROUP_6/HiFi/D7D414A5B/M.1030306415.A>. Assuming you may be not versed in reading chinese, the translation of the comment "NE5532构成的前置放大器为0db放大器,音调控制旋钮通过一个RC滤波器控制NE5532输出信号的反馈量达到增减低频分量的目的." reads: "the pre-amp by NE5532 is a 0db gain amplifier, followed by the tone control, which is implemented by a RC filter to control the amount of feedback to the op-amp, for boost or damp the bass."
here the "feedback" is mentioned. i guess i also need to get familiar with op-amp to fully understand this.
6) The relays delay connection of the speakers so it doesn't make a thump noise when the power is turned on. I would guess that it makes a small thump noise when it's turned off because the one power rail's filter capacitor will discharge at a slightly different rate to the other rail's causing a small current pulse through the speaker.
By "power rails's filter capacitor", which capacitor you are refering to? C108/C110, or else?
Thanks again,
/bruin