Newbie - Twin-T oscillator : how to increase gain?

hugo23

Joined Mar 3, 2010
15
Hello,

I have simulated a Twin-T oscillator using LTSpice. As shown in the
attachment I have done this with a UA741 op-amp. I am however
having an issue with the gain. Assuming the op-amp is powered
on both rails (is that how we say this?) with 5V sources and knowing
that the UA741 will not reach the 5V maximum I was assuming that
I could get at least a sine wave with about 4Vpp.

Simulation shows a maximum of about 742mv pp with R5 set to 27K.
I have tried various values of R5 but cannot get anything better than that.
What is more increasing/decreasing R5 does not correspond to an
increase/decrease in Vout.

As I understand it, at the resonance frequency the negative gain
feedback loop attenuates all signals so the gain is simply determined
by the combination of R5 and R6. Can anyone confirm this?
What is the gain of this circuit? How could I maximize gain?

TIA.

Attachments

• 57.9 KB Views: 78

studiot

Joined Nov 9, 2007
4,998
Edit that was a silly question for a twin T, sorry. Please ignore it.

Why are you applying the frequency selective feedback to the inverting input and the gain setting feedback to the non inverting?

Surely it should be the other way round?

Last edited:

crutschow

Joined Mar 14, 2008
34,392
R5 and R6 have no effect on the gain since they are not in the feedback loop. R5 is just a load to ground and serves no purpose.

hugo23

Joined Mar 3, 2010
15
Edit that was a silly question for a twin T, sorry. Please ignore it.

Why are you applying the frequency selective feedback to the inverting input and the gain setting feedback to the non inverting?

Surely it should be the other way round?
No problem.

hugo23

Joined Mar 3, 2010
15
R5 and R6 have no effect on the gain since they are not in the feedback loop. R5 is just a load to ground and serves no purpose.
Ok. So I made some tests and this still seems to oscillate with no R5 and R6. Possibly stupid question: why use R6 then? Couldn't I just ground the op-amp's non-inverting input then? Simulation seems to work. BTW, some changes in R5 and R6 do change the output slightly. Is this expected? Why?

Theoretical question: my very limited knowledge assumes that to make a circuit oscillate I need positive feedback to compensate for the loss through the pass band filter network. Can I say that re-enforcing the resonant frequency is equivalent to removing all other frequencies via the notch filter?

TIA.

crutschow

Joined Mar 14, 2008
34,392
Ok. So I made some tests and this still seems to oscillate with no R5 and R6. Possibly stupid question: why use R6 then? Couldn't I just ground the op-amp's non-inverting input then? Simulation seems to work. BTW, some changes in R5 and R6 do change the output slightly. Is this expected? Why?
Yes, you don't need R5 and R6 so you can connect the (+) input to ground.

The low value of R5 is overloading the op amp output (R5 should be no lower that about 2kΩ) so that can have some effect on the output.
Theoretical question: my very limited knowledge assumes that to make a circuit oscillate I need positive feedback to compensate for the loss through the pass band filter network. Can I say that re-enforcing the resonant frequency is equivalent to removing all other frequencies via the notch filter?
The phase-shift through the notch-filter changes negative feedback to positive feedback at the notch-frequency. The notch-filter does attenuate all frequencies outside the notch.

Edit: To get a larger output voltage add some positive feedback by connecting a 100kΩ resistor from the output to the (+) input and a 1kΩ resistor from the (+) input to ground. This should give an output of over 4V peak, but note that at least one of the output peaks are now flattened by the op amp limiting. That's why the circuit posted by bertus has a gain control loop to prevent output limiting and distortion.

Last edited:

hugo23

Joined Mar 3, 2010
15
The phase-shift through the notch-filter changes negative feedback to positive feedback at the notch-frequency. The notch-filter does attenuate all frequencies outside the notch.
Ok. I did not consider phase shift. Still have to try and understand how a phase shift changes a negative feedback to a positive one. Doesn't the term negative/positive feedback refer to the op-amps input +/- input?

Edit: To get a larger output voltage add some positive feedback by connecting a 100kΩ resistor from the output to the (+) input and a 1kΩ resistor from the (+) input to ground. This should give an output of over 4V peak, but note that at least one of the output peaks are now flattened by the op amp limiting. That's why the circuit posted by bertus has a gain control loop to prevent output limiting and distortion.
While I was looking at the circuit suggested by Bertus I did realize that the circuit I showed may not be correct. In Bertus's suggested circuit it explicitly states that "positive feedback required for oscillation is provided by R4". I then changed my circuit (resistors R5 and R6) and got a squarish wave of about 3Vpp (will try with the values you suggested).

I am now trying to see how the AGC works. However I see that the wave is much more distorted that the Wien bridge case. Practical question: in such a circuit should one tweak R6 and R5 to get a sine-like wave (less gain) or should one just use these to set-up gain and let the AGC prevent the clipping? I ask because I cannot seem to reduce clipping with R5 and R6 alone.

Thanks for the feedback.

crutschow

Joined Mar 14, 2008
34,392
Ok. I did not consider phase shift. Still have to try and understand how a phase shift changes a negative feedback to a positive one. Doesn't the term negative/positive feedback refer to the op-amps input +/- input?
Only if the feedback network is purely resistive with no appreciable phase-shift. In general, whether the feedback is positive or negative depends upon both the polarity of the input it is fed to and any phase shift in the feedback network. Any combination of those two that gives a total zero (or 360) degree phase-shift from output to input is positive feedback.
...................................
I am now trying to see how the AGC works. However I see that the wave is much more distorted that the Wien bridge case. Practical question: in such a circuit should one tweak R6 and R5 to get a sine-like wave (less gain) or should one just use these to set-up gain and let the AGC prevent the clipping? I ask because I cannot seem to reduce clipping with R5 and R6 alone.
You can't avoid clipping by adjusting R5 and R6 since, if the feedback is sufficient to sustain the oscillations, then the output will continue to increase until the waveform starts to clip. That's why an AGC circuit is used to adjust the gain and keep the feedback below the clipping point.