Re: Transistors

On Nov 19, 11:58 am, Jan Panteltje <pNaonStpealm...@xxxxxxxxx> wrote:
On a sunny day (Mon, 19 Nov 2007 08:42:09 -0800 (PST)) it happened Winfield
Hill <h...@xxxxxxxxxxx> wrote in
<869fdae0-823c-4d47-b987-bab35c0b9...@xxxxxxxxxxxxxxxxxxxxxxxxxxx>:



This is with lower emitter io resistors:
ftp://panteltje.com/pub/amp1_ac2.jpg

Small-signal spice tests (frequency sweeps) of power-
amplifier circuits like that, at high-frequencies,
are close to meaningless - useful for no-signal loop
stability perhaps, but certainly not for evaluating
how an amplifier can do at delivering power at high
frequencies. You can try transient full-power tests
with spice, but I'd want to verify the component
models on the bench before giving it much credence.
And of course, when the failure mode is thermal, and
your spice model isn't set up to handle self-heating,
well... What you can do with spice, after you've got
models you can trust, is to evaluate the continuing
emitter current in a class AB transistor after it's
supposed to have gone off or nearly off each cycle.
Doing this will help teach you what's badly wrong
with that circuit for high-frequency high-power use.

OK, you do have a point, and I have not build this one
with these transistors (that would be the real test,
I trust spice more then in the past, but reality rules),
but this begs the question to *you*:
How would you do it?
You mentioned class A, and this was an attempt to go towards
class A from my side (decreasing the emittor resistors
decreases the phase shift).
I am curious how you would solve the 20V pp 10MHz say
50 Ohm load and open drive?

50 ohms isn't so hard, with the sourcing resistor
it's really 100 ohms or only +/-100mA for +/-10V
--> +/-5V at the terminated destination.

Anyway, there's always brute-force class A ---

But first let me say, I'm not an IC designer and
I don't have access to a fab, etc., at least not
past the usual student stuff, so I don't spend
time worrying about that scene. So the low-
capacitance low-inductance parts and connections
that are needed for efficient 10 to 20MHz power
opamp circuits are a past my interest. However,
I do use basic class-AB design techniques for my
discrete circuits up to say 5MHz.

Looking at your classic output stage, with the
class-AB bias voltage, driving output transistors:
Imagine instead a design where you have such an
output stage running at high class-AB currents,
for high fT and slew rates, but with relatively
small BJTs to keep the capacitance down, imagine
two sets of those, one for the high side and one
for the low side of the class-AB bias voltage.
Now imaging using these two fast medium-power
stages to drive two final output transistors.
This way you are providing serious turn-on and
turn-off currents separately to the two bases,
out of phase with each other. That's the type
of circuit, with extra internal stuff to drive
each output component, that has possibilities
to work to higher frequencies and slew rates.
.