[time-nuts] TNS-BUF update
csteinmetz at yandex.com
Sun Aug 19 04:48:24 EDT 2018
> The bias
> circuit resembles the ones used for applications where
> the transistor's emitter is connected directly to ground
This is often called a "wrap-around" bias circuit.
> since there is this 68 ohm resistor,
> I don't see why it isn't sufficient to simply
> connect a fixed bias of about 1V to the base. You
> could even temperature compensate the voltage
> I'm not
> saying the circuit won't work, just suggesting it
> is needlessly complicated.
That is more or less correct, although see below re: noise.
The wrap-around circuit creates an approximately temperature-compensated
current with an LED, the bias transistor, and the bias transistor's
emitter resistor. This current runs through a 1k resistor to ground to
generate an approximately temperature-compensated voltage that is used
to set the amplifier transistor's base voltage.
That said, the amplifier transistors run quite hot, particularly the
last one, which operates at about 30mA -- nowhere near the much lower
temperature of the LEDs and bias transistors -- so the temperature
compensation does not really do much to stabilize the amplifier current
(note that the same would be true of your posited ~1v voltage source,
even if it were temperature-compensated). The wrap-around circuit *is*
relatively quiet, so your voltage source would likely need to be at
least as complicated to match its bias noise.
> Can I make a high power version of this by
> simply changing to 2N3566/2N5109/2N5943, etc.
In principle, yes. However, note that the maximum DC current rating for
the specified transformers is 30mA. For more power, you would almost
certainly want to pass more quiescent current through the amplifiers.
You would need a different transformer at least for the third stage,
since that stage already draws about 30mA.
> Is the transformer feedback a poor man's Norton
> amplifier scheme?
It is a so-called "noiseless feedback" circuit, yes.
More information about the time-nuts