Feedback artifacts, cars and semantics

[Charles Hansen]

I think he means this:

http://www.diyaudio.com/forums/showt...524#post333524

[Steve Eddy]

Originally posted by Charles Hansen
I think he means this:

http://www.diyaudio.com/forums/showt...524#post333524

I assume you mean this:

Responding to an earlier comment, the gain device doesn't
know what mode (Common Drain, Source, or Gate) it is being
used in. It only sees the varying voltages and currents, and
these can easily be identical from the device's point of view.

Ok.

Essentially the same can be said of an opamp.

So what does this tell us?

se

[andy_c]

Originally posted by Steve Eddy
(...)So what does this tell us?

Life is like a beanstalk...isn't it?

[Steve Eddy]

Originally posted by andy_c
Life is like a beanstalk...isn't it?

Hehehe. Yeah. I'm beginning to feel like I'm in a Kung Fu flashback.

Weedhopper

[andy_c]

Originally posted by Steve Eddy
Hehehe. Yeah. I'm beginning to feel like I'm in a Kung Fu flashback.

Sorry to be so cryptic. It's a Procol Harum quote from one of my favorite albums, "Shine On Brightly". Words are here: http://www.procolharum.com/w/w0208.htm. The guy who transcribed it made a lot of run-on sentences out of it that probably weren't intended by the author.

[Nelson Pass]

I don't mean to be cryptic, so I'll say also that off the top, the
circuit driving a Sziklai pair sees the Vgs or Vbe variation of the
first device, and with the darlington it sees the variation of the
sum of both transistors, thus it is not the same, and the
darlington is seen to have less voltage gain, all other things
being equal.

[SY]

So, then, you're implying that if both connections are degenerated to unity voltage gain, the Sziklai has more feedback. Interesting!

[Steve Eddy]

Originally posted by Nelson Pass
I don't mean to be cryptic, so I'll say also that off the top, the
circuit driving a Sziklai pair sees the Vgs or Vbe variation of the
first device, and with the darlington it sees the variation of the
sum of both transistors, thus it is not the same, and the
darlington is seen to have less voltage gain, all other things
being equal.

Hmmmm. How does the source driving the pair see just the Vgs/Vbe variation of the first device in the Sziklai but the Vgs/Vbe variation of both devices in the Darlington?

The source is referenced to ground and is in series with the source/emitter resistor. Since the current through the source/emitter resistor (I assume we're talking about an emitter follower configuration here) is the sum of the currents through both devices whether Sziklai or Darlington, then the source must be seeing the Vgs/Vbe variations of both devices whether Sziklai or Darlington, no?

se

[Charles Hansen]

Hello SY -

Well, you don't have any choice about this when it comes to the Darlington emitter follower. It will always be (slightly less than) unity gain.

Now the CFP is a different beast altogther. Instead of a common collector feeding a common collector (where there is no choice about what the gain is), we now have a common emitter feeding a common emitter. Each of these common emitter stages will have a gain that (before closing the feedback loop) will depend on what resistor values are chosen for that circuit.

To illustrate this, let's look at a circuit taken from Self's book on amplifiers. In one figure, the first transistor in a CFP has a 100 ohm collector resistor. There is no emitter resistor for the second transistor. The speaker load (nominally 8 ohms) forms both the emitter resistor for the first transistor and the collector resistor for the second transistor.

Now if we "break" the feedback loop in our minds eye, the first transistor will have a gain of 100/(re + 8) where re depends on the bias and is probably a couple of ohms. Let's just say that re = 2 for conveniece's sake, so that the gain of the first transistor is 10. The second transistor will have a gain of 8/re, where re is now around 0.5 (based on a likely bias current in the output stage). This gives a gain of around 16 for the second transistor.

So the composite gain with the loop broken is roughly 160 or 44 dB. Once the loop is closed to achieve unity gain, we now have 44 dB of feedback. This is just within the short 2-transistor loop created by the CFP and is in addition to any other (e.g., global) feedback loops that may exist in the amplifier.

The CFP will have lower measured steady-state distortion (into a resistive load) than the Darlington emitter follower because it has more feedback. Furthermore, it will be less stable.

Best regards,
Charles Hansen

[Nelson Pass]

Originally posted by Steve Eddy
Hmmmm. How does the source driving the pair see just the Vgs/Vbe variation of the first device in the Sziklai but the Vgs/Vbe variation of both devices in the Darlington? The source is referenced to ground and is in series with the source/emitter resistor

I did not assume such a resistor for either case, and adding one
creates a new ball game.

As far as the estimate of internal feedback of the Sziklai, I think
40 dB is way too high if you consider that there is invariably
a resistor Base-Emitter of the second transistor. Depending on
that value, a more typical figure is maybe 10 dB, very dependent
on the impedance of the source and the load.

This doesn't negate the central conclusion - is there more gain
with a Sziklai? Voltage yes, current not really. After that we
are arguing amounts.