Hello everyone,
In the circuit below bias resistor R6 does not connect to ground but to the node between C1 and R2. This seems to be common in old low power transistor amplifier circuits, can someone please explain to me what the idea behind it is?
In the circuit below bias resistor R6 does not connect to ground but to the node between C1 and R2. This seems to be common in old low power transistor amplifier circuits, can someone please explain to me what the idea behind it is?
It looks like bootstrapping. The voltage at the emitter of Q1 will follow the voltage at its base (with a DC offset due to the DC emitter-base voltage). C1 couples the AC signal voltage to the node R2-R6. As a result, there is almost no signal voltage across R6, and R6 does not affect the input impedance.
The things I don't understand are what is supposed to define the collector-emitter voltage of Q3 and what the idea behind R10 is. If you use bootstrapping to make the input impedance high, then volume control with a variable series resistance at the input should be quite ineffective.
The things I don't understand are what is supposed to define the collector-emitter voltage of Q3 and what the idea behind R10 is. If you use bootstrapping to make the input impedance high, then volume control with a variable series resistance at the input should be quite ineffective.
Q1 is the voltage amplifier and the gain is determined by R6, R10, R9 and R2. R7 holds the base voltage of Q1 just low enough to produce a Base/Emitter voltage, turning Q1 on just enough to, in turn, supply base current from the collector of Q1 to the base of Q3.
Q2 is a constant current source and Q3 is the classic class A output stage.
R10 acts in part, as a gain control in an attempt to stabilise Q1.
There is no DC control on the output so a very unstable and unreliable circuit.
Q2 is a constant current source and Q3 is the classic class A output stage.
R10 acts in part, as a gain control in an attempt to stabilise Q1.
There is no DC control on the output so a very unstable and unreliable circuit.
As Marcel mentioned, there's no AC across R6. Therefor, R6 does not play in setting the (ac) gain.
Jan
Jan
Thank you for your replies, so it seems the purpose is to raise the input impedance? I just included that circuit as an example, but not the best one I guess. It is not what I want to build, here is another example wich is very close to the circuit I am interested in, a small class B amplifier. Here you see the same arrangement, R2 does not go to ground but to the node between C4 and R3 of the feedback loop:
Yes, same concept. Because there is no AC across R2, there's no (input) current through it, so for AC it looks like a very high resistance.
Only R1 defines the AC input resistance and that can be made pretty high value.
Jan
Only R1 defines the AC input resistance and that can be made pretty high value.
Jan
The following is only important when you want to design your own bootstrapping circuit:
One thing to watch out for when designing a bootstrapping circuit is subsonic peaking. In the circuit of post #5, at very low frequencies where C4 is essentially an open circuit, the branch R2, R3 has an impedance R2 + R3, just the series value of the two resistors. At high frequencies, where the impedance of C4 is negligible, bootstrapping increases the impedance to a very high value. When you do the math, you find that in between these frequencies it is mainly inductive with an inductance L = R2 R3 C4. (In fact you get that inductance in series with a resistance R2 + R3.) When the component values are chosen wrong, that inductance can ring with C2 and cause subsonic peaking.
One thing to watch out for when designing a bootstrapping circuit is subsonic peaking. In the circuit of post #5, at very low frequencies where C4 is essentially an open circuit, the branch R2, R3 has an impedance R2 + R3, just the series value of the two resistors. At high frequencies, where the impedance of C4 is negligible, bootstrapping increases the impedance to a very high value. When you do the math, you find that in between these frequencies it is mainly inductive with an inductance L = R2 R3 C4. (In fact you get that inductance in series with a resistance R2 + R3.) When the component values are chosen wrong, that inductance can ring with C2 and cause subsonic peaking.
How would subsonic peaking manifest itself? it sound a bit scary, apparently you won't hear it coming.. Would it destroy anything?
So how do we avoid subsonic ringing? Maybe a high value of C4 of the order of 1,000uF - 2,200uF or so?
Normally you wouldn't even notice it. When you play something with substantial subsonic content, for example a warped record or an audiophile recording made in a venue with a draught, then your woofers will move more than necessary and cause some extra intermodulation distortion.
There are two ways to damp the resonance, but one of them causes a deep notch in the input impedance curve that partly defeats the purpose of having a bootstrap. The other method is choosing the values such that sqrt(R2 R3 C4/C2) is greater than or approximately equal to sqrt(2) R1. In fact I would prefer approximately equal, as it leads to approximately Butterworth pole positions.
At the same time, 1/(2 pi sqrt(R2 R3 C2 C4)) determines the cut-off frequency and has to be somewhere in the subsonic region.
On top of that, R2 should preferably be much greater than R3.
Last edited:
The circuit in the original post has an error! R3 should be connected to the output rail not the bias of the constant current transistor.
Otherwise the circuit won't work. As pointed out.
Bootstrapping was common as the OP says in the old days. I would be tempted to skip that and ground R6 to prevent subsonic bounce. 330k is still a high enough impedance even when shunted by R7 (net 200k ish) that it won't burden most signal sources.
Also R10 is not much of a volume control, needs to be rewired so that the input is connected to the high and the connection to the transistor stages taken from the wiper. That would reduce the input impedance to 10k but that again should be OK with most signal sources. As it is drawn it will mostly add noise.
Otherwise the circuit won't work. As pointed out.
Bootstrapping was common as the OP says in the old days. I would be tempted to skip that and ground R6 to prevent subsonic bounce. 330k is still a high enough impedance even when shunted by R7 (net 200k ish) that it won't burden most signal sources.
Also R10 is not much of a volume control, needs to be rewired so that the input is connected to the high and the connection to the transistor stages taken from the wiper. That would reduce the input impedance to 10k but that again should be OK with most signal sources. As it is drawn it will mostly add noise.
Bootstrapping like this may also be intended to increase the headroom before clipping (important for low voltage amplifiers) - however the more I look at this circuit the less I understand, it seems to be completely bizarre and the output is likely to be very non-linear, and ill-defined at DC with ill-defined AC gain. Its also sensitive to source impedance linearity and load impedance. My vote is this is just a "bad circuit" in the Horrowitz/Hill sense.