Talk:RC oscillator

Inventor of the Wien bridge oscillator

Hasn't Mr. Hewlett actually INVENTED The Wien bridge oscillator?... I've heard so. -- NIC1138 03:01, 25 April 2007 (UTC)

No,he made the first reliable oscillator:The lamp as an AGC cancelled the need of a multi-turn potentiometer continiusly corrected

Ex-Giannizx1 —Preceding unsigned comment added by 82.55.254.42 (talk) 14:32, 3 May 2008 (UTC)

Twin-T oscillator

The "Twin-T" oscillator description is completely unclear. In fact, the idea is: two T-shaped circuits, one is (input)-R1-(center)-R2-(output), (center)-C3-(ground), second (input)-C1-(center)-C2-(output), (center)-R3-(ground) - a schematic would be useful; usually R1=R2=R, R3=x*R, C1=C2=C, C3=x*C, the x is around 2.

For x=2 a signal of a frequency $f={\frac {1}{2*\Pi *R*C}}$ does not get to the output, as it passes with equal amplitudes and reverse phases via both circuits - such a 2T circuit rejects just frequency 'f' - if put in negative feedback connection in an aplifier, the aplifier will amplify the frequency 'f' exactly as it contained a parallel tuned circuit on its output.

For x>2 the frequency 'f' passes, but its phase is reversed - the amplifier with the circuit behaves as it has a tuned circuit with a positive feedback, i.e. it becomes oscillator (since the aplification is limited, the 'x' must be > '2+something', depending on the amplification).

JerzyTarasiuk (talk) 19:28, 27 May 2009 (UTC)

Is there a way to put a source (to be used in some program) for an image (like schematics) I put here (in a way it is shown with the image)?

JerzyTarasiuk (talk) 20:29, 27 May 2009 (UTC)

Comments and questions

1. The section entitled "Sine wave oscillator" is somewhat inappropriate since RC oscillators are feedback oscillators, therefore falling under harmonic/linear oscillators which are sine-wave oscillators.

2. The Twin-T oscillator has a figure of a Twin-T filter which should be replaced by an image of a Twin-T oscillator.

3. The "Non-sine wave oscillators" section says "Many designs exist for RC oscillators that are not required to produce a sine wave. Square waves are the most common. Multivibrators are one approach."

RC is a category of harmonic/linear oscillators. The sentences above seem to contradict the classification of electronic oscillators. Although multivibrators have resistors and capacitors they are relaxation/nonlinear oscillators and they are not classified as RC oscillators because they produce square waves so, by definition, they are not RC oscillators.

ICE77 (talk) 04:05, 4 August 2015 (UTC)

I agree enthusiastically with all the above comments. The point which the "Sine wave oscillator" section is trying to make is that some low distortion RC oscillators such as the Wien bridge oscillator have nonlinear "gain control" feedback loops which prevent the active device from saturating, and without such a feature the plain RC oscillator produces a distorted ("clipped") sine wave output. These are called various names but not "sine wave" oscillators; one term is "low distortion oscillators", so I think the section could be renamed that. In addition, besides the Wien bridge, the section inexplicably also includes the twin-T oscillator and the phase shift oscillator, which are plain RC oscillators without gain control and should be moved out of this section. --Chetvorno^TALK 04:37, 4 August 2015 (UTC)

Twin-T and phase shift oscillators are sinewave oscillators just as LC oscillators are sinewave oscillators. If the gain is chosen properly, there will be a little compression and the result will be low distortion. If the gain is too large, then significant compression/clipping occurs. Even Hewlett's Wien bridge oscillator depends on nonlinear compression (in addition to the lamp filament) to make it work well (see Oliver). Glrx (talk) 19:44, 7 August 2015 (UTC)

Article is much improved by Glrx's rewrite. For the first time article has an adequate explanation of amplitude determination in oscillators. --Chetvorno^TALK 20:40, 10 August 2015 (UTC)

I'm thinking that we need a section on "Low distortion oscillators" to make clear that there are some RC oscillators which deliberately include a "gentle" nonlinear element to reduce distortion, such as the Wein bridge's light bulb or thermistor, to distinguish them from the rest of oscillators that just use the (nonlinear) saturation of the amplifier to limit amplitude. These have a much higher distortion than the oscillators with "gentle" nonlinear elements. The distortion of ordinary oscillators can be reduced somewhat as Glrx says by reducing the excess loop gain above one, but there is a limit to how much you can reduce it. If the oscillator does not have a small signal loop gain significantly above one it will not start quickly and reliably; a small change in gain due to component drift will cause it to fail to start. If low distortion could be achieved with ordinary saturation-limited oscillator circuits they would not have invented the ones with "gentle" nonlinearities like the Wien. --Chetvorno^TALK 00:52, 13 August 2015 (UTC)

Thanks for the feedback and for adding the Twin-T oscillator image. I have one more question and two more comments.

4. Who is the inventor of the Twin-T oscillator?

5. The phase-shift oscillator could be expanded. It really doesn't say much.

6. The Robinson oscillator could be included/mentioned in this article since it is another RC oscillator.

ICE77 (talk) 00:53, 12 August 2015 (UTC)

Is the Robinson oscillator a linear oscillator? It wasn't clear from the inadequate article, which should be improved. Another RC oscillator that probably belongs here is the quadrature oscillator. --Chetvorno^TALK 04:19, 12 August 2015 (UTC)

Chetvorno, my understanding is that the Robinson oscillator is an RC oscillator. Also, adding also the quadrature oscillator is a good idea. It's great you pointed it out because I had a sine/cosine oscillator simulation and I didn't know what the circuit was named!

ICE77 (talk) 07:13, 13 August 2015 (UTC)

Based on the above exchange on distortion between Chetvorno and Glrx my understanding is that all RC oscillators show some sort of output distortion due to high gain except for the Wien Bridge oscillator which keeps the gain low (3 when R_f is twice R_b). If that is the case, how can the gain be carefully selected when it comes to the Twin-T oscillator? What values and ratio would you use for R₃ and R₄?

ICE77 (talk) 22:58, 26 August 2015 (UTC)

No, its not the gain that makes the Wien Bridge circuit low distortion, it is the "gentle" nonlinearity of the incandescent bulb R_b in the circuit, as opposed to the abrupt nonlinear "clipping" action of the transistor or op amp in other oscillators. Only in ordinary RC oscillators, like the Phase Shift or Twin T, where the amplitude is limited by saturation of the amplifier clipping the sine wave, does the amount of open loop gain have a bearing on the distortion. In the Wien Bridge oscillator, the gain of 3 does not refer to the loop gain of the circuit, but just the gain of the op-amp R_f/R_b, which must be 3 in order to compensate for the attenuation of the frequency determining part: R₁ c₁ R₂ C₂, which is 1/3.

The way the Twin T oscillator works is (according to what I read on the web): The filter part, in the blue box, is a "notch" filter; its frequency response has a deep "notch" so that it passes all frequencies except those those at its "notch" frequency, $\scriptstyle f_{0}\;=\;1/(2\pi R_{1}C_{1})$; at this frequency essentially none of the input signal gets through. Since the output of the filter is applied to the inverting input of the op-amp, the filter provides high negative feedback, keeping the loop gain far below one at all frequencies except the notch frequency, so the oscillator oscillates at the notch frequency $\scriptstyle f_{0}$. At the notch frequency, the output of the filter is close to zero, so it is as if the inverting input of the amplifier was grounded. That leaves a positive feedback loop with a high gain op-amp and a voltage divider feedback network:

$V_{+}=R_{4}/(R_{3}+R_{4})V_{O}$

If $\scriptstyle A$ is the gain of the op-amp, the open loop gain of the oscillator is: $\scriptstyle G=\beta A=R_{4}A/(R_{3}+R_{4})$. The condition for oscillation is that this has to be one:

$G={R_{4} \over R_{3}+R_{4}}A=1$

and for startup it has to be a little greater than one. (I worked through the math and got the more precise condition

$G={R_{4} \over R_{3}+R_{4}}A={4\pi f_{0} \over 1+4\pi f_{0}}\,$

but the quantity on the right is approximately equal to one when the frequency is well above one hertz).

Since the op-amp has a high gain, the feedback voltage divider must have a high attenuation to cut back the loop gain to one, so $\scriptstyle R_{3}$ must be much greater than $\scriptstyle R_{4}$, so the $\scriptstyle R_{4}$ in the denominator can be neglected:

${R_{4} \over R_{3}}A=1\,$

$R_{3}=R_{4}A\,$

So the ratio of R₃ to R₄ that gives loop gain G = 1 depends on the gain A of the op-amp at the oscillation frequency $\scriptstyle f_{0}$. As long as R₃ is less than the above value the circuit will oscillate, although for lowest distortion R₃ should not be too low. Op-amps have high gain of around 10⁵ at low frequencies, but they have a dominant pole "roll-off" which reduces the gain by 20 dB per decade as the frequency increases above about 10 Hz, so at audio frequencies the op-amp may have a gain of less than 1000. Therefore in the Twin T circuits I looked at on the web, the value of R₃ used was typically around 1000R₄.

I know the last part sounds like hand-waving. The gain of the op-amp depends on frequency, so depending on the oscillation frequency the open loop gain could end up 10 or even 100, nowhere near one. But (as you probably know) that is typical of the cookbook designs of oscillators. As long as the gain is high enough that the oscillator starts reliably, not much attention is given to loop gain, and it may be far above one. According to the sources I read, the Twin T oscillator has low distortion for an RC oscillator, because the narrow "notch" filter removes the distortion frequency components leaving a purer sine wave, so maybe the value of the R₃/R₄ ratio doesn't make much difference to the distortion. --Chetvorno^TALK 16:34, 27 August 2015 (UTC)

Chetvorno, thanks again for the thorough explanation. I feel like I learn a lot from your feedback. I set up a simulation in PSpice with R₁=2kΩ and C₁=100nF for f=796Hz. If I use a value of R₃ that is 1000 times bigger than R₄, interestingly, I do not get any oscillation. I only get oscillation with a value 10000 times bigger with R₃=10MΩ and R₄=10kΩ. The output waveform is almost perfectly sinusoidal (there is a very small clipping which is barely pronounced). Why is that 10000 times will work but not 1000?

Also, I noticed you uploaded the circuit for the Twin-T filter. What kind of software do you use to generate the images of the circuits? They look nice and clean.

ICE77 (talk) 20:18, 28 August 2015 (UTC)

@ICE77, all this work you've done is fantastic; I'm getting an education about oscillators from you. Sorry I've been away so long, but I'm really curious about your results. If the circuit oscillates for large values of R₃/R₄ and not for smaller values that seems backwards; the circuit should oscillate for R₃/R₄ < A. Maybe I got it wrong. You say you are not getting oscillation at R₃ = 1000R₄, but it looks to me that if R₃ = 10MΩ and R₄ = 10kΩ, you are: R₃/R₄ = 10MΩ/10kΩ = 1000? What gain does your PSpice simulation use for the amplifier? I've got some commitments, but give me a day or two to look at it again.

I drew the Twin-T oscillator image with Inkscape, a free vector drawing program; guess it turned out pretty. --Chetvorno^TALK 05:19, 4 September 2015 (UTC)