Barkhausen must be wrong.

[The Phantom]

At this site:

http://www.4qdtec.com/singen.html

there is a schematic titled "A Practical Twin-T Oscillator".

In the text under the schematic we find:

"Now hold on a minute: an emitter follower has no voltage gain and surely
you've been taught that an R-C oscillator must have voltage gain? Well this
one works and has no voltage gain (of course it does have current gain)."

The schematic does show two emitter followers closing the loop. One would
think this couldn't work, but the poster says it does.

Is Barkhausen wrong?

 

[MooseFET]

At this site:

http://www.4qdtec.com/singen.html

there is a schematic titled "A Practical Twin-T Oscillator".

In the text under the schematic we find:

"Now hold on a minute: an emitter follower has no voltage gain and surely
you've been taught that an R-C oscillator must have voltage gain? Well this
one works and has no voltage gain (of course it does have current gain)."

The schematic does show two emitter followers closing the loop. One would
think this couldn't work, but the poster says it does.

Is Barkhausen wrong?

Look again at the schematic.

 

[The Phantom]

Look again at the schematic.

I've looked at it several times. What do you see there that's relevant?

 

[John Popelish]

I've looked at it several times. What do you see there that's relevant?

The signal input to the right transistor is not from an
emitter follower.

 

[John Larkin]

At this site:

http://www.4qdtec.com/singen.html

there is a schematic titled "A Practical Twin-T Oscillator".

In the text under the schematic we find:

"Now hold on a minute: an emitter follower has no voltage gain and surely
you've been taught that an R-C oscillator must have voltage gain? Well this
one works and has no voltage gain (of course it does have current gain)."

The schematic does show two emitter followers closing the loop. One would
think this couldn't work, but the poster says it does.

Is Barkhausen wrong?

You can make an oscillator with just an emitter follower and some
r-c's... I did it accidentally ca 1975, and was surprised. The
impedances are such that you usually need a darlington to pull it off.

There are a couple of simple 2-terminal-plus-ground RC networks that
can have an AC voltage gain above 1.

John

 

[The Phantom]

The signal input to the right transistor is not from an
emitter follower.

I'm not sure what you mean. Are you sure you're looking at the right
schematic, the 5th one down? The one titled "A Practical Twin-T
Oscillator"?

 

[The Phantom]

You can make an oscillator with just an emitter follower and some
r-c's... I did it accidentally ca 1975, and was surprised. The
impedances are such that you usually need a darlington to pull it off.

There are a couple of simple 2-terminal-plus-ground RC networks that
can have an AC voltage gain above 1.

But he says this one has no voltage gain. So how does this work?

 

[John Larkin]

But he says this one has no voltage gain. So how does this work?

I think he meant that the transistors have no voltage gain, and maybe
he's assuming that an RC network can't have gain. Dunno.

John

 

[Jim Thompson]

But he says this one has no voltage gain. So how does this work?

POWER gain IS the requirement.

...Jim Thompson

 

[John Popelish]

I'm not sure what you mean. Are you sure you're looking at the right
schematic, the 5th one down? The one titled "A Practical Twin-T
Oscillator"?

Sorry, I was looking at the second schematic.

The unusual part of this twin T is that the component values
are not the usual C and 2*C, 2*R and R. A ratio of about 5
is used, instead. And the way it is connected, instead of
the usual notch response, it has a broad bandpass filter
response with a peak gain of about 1.089 at about 130 Hz,
with the loading effect of the 1n cap but not counting the
darlington.

So it is possible to get a little voltage gain from an RC
circuit.

 

[John Popelish]

Sorry, I was looking at the second schematic.

The unusual part of this twin T is that the component values are not the
usual C and 2*C, 2*R and R. A ratio of about 5 is used, instead. And
the way it is connected, instead of the usual notch response, it has a
broad bandpass filter response with a peak gain of about 1.089 at about
130 Hz, with the loading effect of the 1n cap but not counting the
darlington.

So it is possible to get a little voltage gain from an RC circuit.

Simulating just the twin T, unloaded, the highest possible
gain seems to occur close to an R and C ratio of about 4.75,
and that voltage gain is about 1.094.

 

[The Phantom]

Sorry, I was looking at the second schematic.

The unusual part of this twin T is that the component values
are not the usual C and 2*C, 2*R and R. A ratio of about 5
is used, instead. And the way it is connected, instead of
the usual notch response, it has a broad bandpass filter
response with a peak gain of about 1.089 at about 130 Hz,
with the loading effect of the 1n cap but not counting the
darlington.

You get an A+. My mathematical assistant says the max gain of 1.08893
occurs with a ratio of 4.83152 at a frequency of 129.128034 Hz

So it is possible to get a little voltage gain from an RC
circuit.

See: "Synthesis of Passive RC Networks with Gains Greater than Unity",
Herman Epstein, Proceedings of the IRE, July, 1951, p. 833.

 

[The Phantom]

Simulating just the twin T, unloaded, the highest possible
gain seems to occur close to an R and C ratio of about 4.75,
and that voltage gain is about 1.094.

My assistant says that under these circumstances the maximum gain of
1.09384 occurs at a frequency of 132.629114 Hz with a ratio of 4.82843.

 

[John Popelish]

You get an A+. My mathematical assistant says the max gain of 1.08893
occurs with a ratio of 4.83152 at a frequency of 129.128034 Hz

See: "Synthesis of Passive RC Networks with Gains Greater than Unity",
Herman Epstein, Proceedings of the IRE, July, 1951, p. 833.

I don't have access to this paper.

 

[Joel Kolstad]

I don't have access to this paper.

In Vorperian's book, "Fast Analytical Techniques for Electrical and Electronic
Circuits" there's a 7-page discussion of it as well (under the heading, "RC
Filters With Gain" in the chapter, "Passive Filters: Where Inductors and
Transformers Still Get Respect," which I rather like ). That's probably
short enough to fall under "fair use" if you'd like me to scan and post it.

---Joel

 

[Rich Grise]

But he says this one has no voltage gain. So how does this work?

I looked at it as, the Twin-T, because of the component values, is
an impedance transformer. It has "voltage gain" in that the output
volts are greater than the input volts; but it has a "current loss"
which is why you need the darlington - for the light loading on the
output end.

The thing is, I don't know if an impedance transformer counts as
"gain" - does an ordinary step-up transformer have "gain"?

Thanks,
Rich

 

[The Phantom]

I don't have access to this paper.

You will very shortly.

 

[The Phantom]

In Vorperian's book, "Fast Analytical Techniques for Electrical and Electronic
Circuits" there's a 7-page discussion of it as well (under the heading, "RC
Filters With Gain" in the chapter, "Passive Filters: Where Inductors and
Transformers Still Get Respect," which I rather like ). That's probably
short enough to fall under "fair use" if you'd like me to scan and post it.

Post away!

 

[The Phantom]

I looked at it as, the Twin-T, because of the component values, is
an impedance transformer. It has "voltage gain" in that the output
volts are greater than the input volts; but it has a "current loss"
which is why you need the darlington - for the light loading on the
output end.

The thing is, I don't know if an impedance transformer counts as
"gain" - does an ordinary step-up transformer have "gain"?

If one's definition of "voltage gain" is "ABS(Vo/Vi) > 1", then, yes.
This is the sense in which Epstein used the word "gain". See my responses
to John Popelish.

And, of course, one could get even more particular and speak of "loaded
gain" and "unloaded gain".

I imagine that an oscillator could be built which used a small transformer
on the output of a network giving phase shift, followed by an emitter (or
source) follower for the necessary "current gain". The phase shifting
network could even be an all-pass network.

 

[Tim Williams]

Take a (triode, pentode, transistor, FET) and put the common
(cathode/emitter/source) to the CT on a tapped tank coil. Ground one end
of the coil. Cap couple the other end to the input (grid/gate/base),
biasing as appropriate. Connect supply voltage to other terminal
(plate/collector/drain).

Voila.

Tim