In playing with my simple emitter follower circuit trying to figure out why there's very faint output when connected to my 8 Ohm speakers, I came upon some important realizations.
My circuit is a simple emitter follower with one transistor. The base is biased to midpoint through a voltage divider and the input is capacitively coupled. The output is also capacitively coupled through a large capacitor to block DC. The emitter resistor is 5K.
I expected that with a large capacitor, even an 8 Ohm speaker would be suitable to place the corner frequency of the high pass filter low enough to pass audio frequencies and high enough to block DC. However I was faced with two anomalies. Whenever I connect my speaker after the capacitor, I hear a very faint and distorted sound. Whenever I connect the speaker before the capacitor (the NPN's emitter), I get an almost exact replica of the input (which is what the emitter follower is supposed to do). Simulation confirmed this behavior but I was puzzled with the result. Do I throw out of the window all I know about high pass filters and frequency response. Also do I throw out the importance of bias? Here we have a high pass filter that is not acting like a high pass filter. And we have a low resistance emitter resistor (after connecting the speaker in parallel with the 5K emitter resistor) that should throw the bias point way off midpoint.
After playing with simulation for a while, and replacing the transistor with equivalent circuit, etc, I hit upon one of my biggest realizations in electronics (mind me if it's too obvious for you). The models are correct so long as the transistor is ON and operating in its linear region. In such a case the high pass filter will act nicely and block DC and pass the audio signal. However, with such low resistance, If the output to follow the input faithfully, huge current will have to pass in the capacitor for the slightest negative output voltage, and since the current in the emitter resistor (5 KOhm) is very small compared, this current would have to follow in the transistor in the opposite direction, so the transistor turns off. That's my reasoning for why the transistor would turn off.
So now we have another mode of operation. Now it's another circuit. It's not a high pass filter but a capacitor discharging through a resistor. At a certain point the capacitor charge will be large enough to have the transistor turn on only very briefly and the output voltage change so little.
What caused my confusion is that in most electronic circuits, there are transistors but often the circuit description doesn't tell if that transistor will be ON all the time or not. Then analysis follows that replaces caps with shorts at signal frequencies etc. But that's valid only if the transistor stays on and hence acts as a linear device. I wonder if circuit designers even with all their experiences and rules of thumb take the extra step of analysis (or simulation) to make sure the transistor is always operating in the linear region. If they do, it's often not mentioned to us, less experienced people. In other applications, such as rectifiers, the charging and discharging is explicitly mentioned and used. This dual role of the capacitor always puzzled me, when to think of it as a charging/discharging device, and when to think in terms of frequency response and frequency dependent impedance.
The other question was why connecting the speaker directly to the output is OK. Why no blocking capacitor is needed? It turned out that the small impedance shifts the DC bias point very close to ground, and hence the output is a replica of the input but shifted up in DC level just a few millivolts above ground, which seems to not alter the speaker response much. That seems to be a problem if the gain is high and clipping will start to happen as you move away from midpoint, but when the gain is 1, no problem since the input signal is already a few millivolts around 0.
A possibility here is that the speaker also responds to changes in magnetic field and hence the DC component wouldn't matter, but i'm not sure about that one.
I apologize for anyone who took the effort and went through my long post. Am I right in my conclusions?
Thank you.