Not able to get the desired output frequency using a Wein Bridge Oscillator

• handyman123

handyman123

TL;DR Summary
I had to get a f of 1MHz but I am only getting the f as 1kHz, even though my calculations are for 1MHz.

Can anyone help me find what is wrong in this circuit? given, slew rate of the op amp is 400V/us and max output current for opamp is 40mA but the opamp is lm741.

Have you checked the datasheet to determine whether the gain and bandwidth of the op amp is sufficient to sustain the 1MHz oscillation?

In addition, the oscillation frequency formula is ##~ \frac 1 {2 \pi R C}~##, have you checked whether the values of the components are correct?

Last edited:
Merlin3189 and DaveE
Summary:: I had to get a f of 1MHz but I am only getting the f as 1kHz, even though my calculations are for 1MHz.

given, slew rate of the op amp is 400V/us and max output current for opamp is 40mA but the opamp is lm741.
A quick check of an LM741 datasheet at Digikey shows more like 0.4V/us. LM741 opamps are okay for audio frequencies, but generally are not fast enough for MHz tasks, IMO...

davenn and DaveE
A quick check of an LM741 datasheet at Digikey shows more like 0.4V/us
Good catch. 0.4V/us is 400 mv/us

When I saw 400 volts per microsecond and 741 in the same sentence I nearly choked.

DaveE, Tom.G and berkeman
A good rule of thumb is to never use a 741 .
It is an absolutely ancient design (~1965) and there are lots of other equally cheap, easy to get op-amps with much higher performance.

davenn, DaveE and Averagesupernova
When I searched megahertz wein bridge oscillator, I found a circuit for a one megahertz Wein bridge. The capacitors were 100 pf. That would be 0.1 nf, which is 500 times less than the 50 nf shown in the OP. Tweak the resistors appropriately, and your frequency will increase to what you want.

hutchphd and anorlunda
@handyman123
Welcome to PF.
The condition for oscillation to begin requires voltage gain greater than 2.00000; It will either run, or it will not run depending on the gain. By duplicated the 30k feedback resistor you appear to have set sufficient numerical gain with your resistor ratio. The Wien bridge requires careful control of gain when running. If the gain is too high, the simulation will produce distortion. In some respect, a model will differ from reality and there is no telling what a simulation might do.

Notice the spelling is Wien, not Wein.

PeroK
I think, the gain of the amplifier has to be slightly larger than "3". This is because the bridge attenuates the signal at the desired frequency with a factor of 1/3.

Merlin3189
I think, the gain of the amplifier has to be slightly larger than "3".
Sorry, not gain but the resistor ratio in the inverting feedback must be 1 : 2 when on the edge of pure sinewave oscillation. If the circuit losses and finite gain of the op-amp were included, then 30k : 60k would be insufficient to oscillate. So I don't know why the model actually oscillates, maybe the turn-on transient is sufficient.

The Wien bridge requires excess gain to start, then some form of gain control to limit the output amplitude. The form of gain control will determine the sinewave purity.

Sorry, not gain but the resistor ratio in the inverting feedback must be 1 : 2 when on the edge of pure sinewave oscillation. If the circuit losses and finite gain of the op-amp were included, then 30k : 60k would be insufficient to oscillate. So I don't know why the model actually oscillates, maybe the turn-on transient is sufficient.

The Wien bridge requires excess gain to start, then some form of gain control to limit the output amplitude. The form of gain control will determine the sinewave purity.
Yes - I could imagine what you mean.
In this specific case, I think the unavoidable tolerances of the two resistors did allow the start of oscillations (gain slighly >3) - together with the turn-on transients which always are sufficient for a safe start.

I think the unavoidable tolerances of the two resistors did allow the start of oscillations
That was lucky. I say that three real resistors from the same batch will have a less than 50% chance of the oscillator starting and running. As a numerical simulation the oscillation would not start.

The finite open-loop gain of the op-amp will require a slightly greater feedback ratio to compensate. If the three resistors used were identical, oscillation would not continue, even after a turn-on transient.

Without an excess feedback ratio, it can take several seconds for the oscillator to start.
The circuit capacitors take time to charge, so a turn-on transient may be insufficient.

Without feedback gain control, if it runs, the amplitude will be limited by clipping of the op-amp input and/or output voltages, so the sinewave will be distorted. Peak clipping also changes the frequency.

Without feedback gain control, if it runs, the amplitude will be limited by clipping of the op-amp input and/or output voltages, so the sinewave will be distorted. Peak clipping also changes the frequency.
Yes - that is always the case (for the majority of oscillator circuits), with only some exceptions.
One of these exceptions is the "double-integrator oscillator" which can deliver a rather good signal - even without a separate amplitude regulation device.

One of these exceptions is the "double-integrator oscillator" which can deliver a rather good signal - even without a separate amplitude regulation device.
The Double Integrator Oscillator, (with a loop amplifier gain of –1), is amplitude controlled by the signal clipping the supply rails, so it generates a sinewave with about 1% odd harmonic distortion.

The Double Integrator Oscillator, (with a loop amplifier gain of –1), is amplitude controlled by the signal clipping the supply rails, so it generates a sinewave with about 1% odd harmonic distortion.
I suppose you assume two identical integrator blocks.
However, because of clipping effects it is advisable to use always two different time constants for both integrators. In this case, only the amplifier with the lower time constant (highest gain) will go into saturation (clipping) and the (filtered) output signal of the other opamp will be sinusoidal with much better signal quality.
(Simulation result: THD 0.02%)

Any oscillator can be optimised in some dimensions by sacrificing others. There are plenty of other problems with the double integrator oscillator.

Any oscillator can be optimised in some dimensions by sacrificing others. There are plenty of other problems with the double integrator oscillator.
Which "problems" you are referring to?
To me, this oscillator type is the most interesting one because it has (at least) three specific features:
* Why does it start and at which frequency?
* Why does it not work for ideal opamp models?
* Why does only slight clipping occur ? How much excessive gain?

To me, this oscillator type is the most interesting one because it has (at least) three specific features:
Do you find the double integrator oscillator interesting because you know the answers to your three questions, or because you do not know the answers?

Do you find the double integrator oscillator interesting because you know the answers to your three questions, or because you do not know the answers?
I know the answers... I was just trying to explain why I find this type especially interesting - in contrast to the other well-known oscillator principles.

FYI - the OP seems to be a hit and run. He connected last about 30 minutes after making the post. Never saw any answers/comments.

The classical Wien bridge oscillator incorporated an automatic gain control using a small light bulb. As these are increasingly difficult to come across, I looked through several examples and found this (on https://www.electronicsinfoline.com/pin/13626/):

The classical Wien bridge oscillator incorporated an automatic gain control using a small light bulb. As these are increasingly difficult to come across, I looked through several examples and found this (on https://www.electronicsinfoline.com/pin/13626/):
View attachment 264315
Yes - that is one of the classical solutions for gain control.
Even better (from the THD point of view) is a real AGC loop using - for example - a voltage controlled resistance (FET).

Any amplitude regulation technique that uses a fixed gain and then clips the excess amplitude, will generate some harmonic distortion. That goes for back-to-back diodes and for supply rail limiting.

To control oscillation amplitude without changing the frequency or introducing distortion, it is necessary to precisely trim the loop gain over many cycles. An absolute amplitude detector or true RMS detector with a relatively long time constant, controlling the gain of a transconductance amplifier, or an analog multiplier, can achieve that control.

A base metal filament has a resistance proportional to absolute temperature. The HP-200 oscillator used a warm lamp filament with a long thermal time constant to achieve gain control over the full cycle without clipping. The filament can be replaced with a bead thermistor to achieve the same amplitude regulation with low distortion. It is quite a challenge with any thermal amplitude integrator to avoid some amplitude dependence on the environmental temperature.

A Direct Digital Synthesizer chip can now generate higher quality amplitude controlled quadrature sinewaves with precise frequency and amplitude control. The amplitude profile and frequency sweep can be selected to modulate the output. It seems that the Wien bridge has played it's part in history.

Starting with post #13 some features of the Double Integrator Oscillator (DIO) were was discussed.
In the meantime I have prepared a short paper with some (partly new) background information about the DIO.
The paper is attached as a pdf file.

Attachments

• DIO_why_does_it_oscillate.pdf
415.5 KB · Views: 179
FYI - the OP seems to be a hit and run. He connected last about 30 minutes after making the post. Never saw any answers/comments.
Lucky chap. He dodged a bullet there then!

But he got his answer in post 2, so no need to linger. I expect he corrected the error Alan mentioned, then it worked.