Talk:Electronic oscillator

From Wikipedia, the free encyclopedia
Jump to navigation Jump to search
WikiProject Electronics (Rated C-class, Mid-importance)
WikiProject iconThis article is part of WikiProject Electronics, an attempt to provide a standard approach to writing articles about electronics on Wikipedia. If you would like to participate, you can choose to edit the article attached to this page, or visit the project page, where you can join the project and see a list of open tasks. Leave messages at the project talk page
C-Class article C  This article has been rated as C-Class on the project's quality scale.
 Mid  This article has been rated as Mid-importance on the project's importance scale.
 

Large image removed[edit]

A large image was added to this article, which was of such an enormous size that it caused serious browser lag and swapping. Removed, for the time being. The Anome

Harmonic oscillator[edit]

The article doesn't say if there is any neback the original signal delayed by one cycle" descriptions of harmonic oscillators really just the same thing said in two different ways? or are they two differnet types of harmonic oscillators? i guess if i thought about it i'd figure it out, but i'd rather just have someone tell me... Omegatron 16:01, Jun 10, 2004 (UTC)

Yes, 2 very different ways of describing the same thing. I think of it more as "feed back the original signal delayed by some fixed amount of time".
Does the main article need to explain this better ? --DavidCary 01:01, 20 Jul 2004 (UTC)

we also need to mention that it needs a nonlinearity to keep it from feeding back forever. it needs to be limited at a certain amplitude. - Omegatron 16:01, Jun 10, 2004 (UTC)

All amplifiers go nonlinear when the input gets "too large". That automaticially limits the output amplitude, but also introduces "nonlinear harmonics", spoiling the perfect sinewave that is the goal of harmonic oscillators. Many people just accept this slight imperfection (the difference from a perfect sinewave is often too small to notice). Others add components to reduce gain once the sinewave has reached adequate amplitude (I've been told that HP's original oscillator used a light bulb; so does http://www.4qdtec.com/singen.html )

--DavidCary 01:01, 20 Jul 2004 (UTC)

I'm sure some filter the harmonics as well. All of this stuff should be in the article. - Omegatron 22:22, Sep 25, 2004 (UTC)

Our picture of a Colpitts crystal oscialltor[edit]

Our picture of a Colpitts crystal oscillator is nice, but someone should probably edit it to provide a pull up resistor from the collector of the transistor to Vcc. :-)

Atlant

Generalized nonlinear oscillators[edit]

Could someone expand the entry to include this too, please ?

thanks a million

jake

Audio Frequency Oscillator /Audio Oscillator[edit]

The Term Audio Oscillator redirects to this article, but this article only makes one mention of Audio Frequency Oscillator. I don't know if these are one and the same or not, but either way it's not very clear to me what Audio Frequency Oscillators are. Is it just another name for Electronic Oscillators? I'm confused and would like some clarification. Thanks in advance. Lewiscode 20:16, 13 September 2006 (UTC)

An audio oscillator is an electronic oscillator whoss bandwidth ranges from 16 HZ to 20,000 HZ. Doktor Who 00:00, 4 September 2007 (UTC)
Or maybe 10Hz to 20kHz, or maybe 10 Hz to 100kHz. Some audio oscs go to 1 MHz i think--GreenSpigot (talk) 02:59, 31 January 2009 (UTC)

Time delay oscillator[edit]

Our ring oscillator article currently mentions

"The ring oscillator is a member of the class of time delay oscillators. A time delay oscillator consists of an inverting amplifier with a delay element between the amplifier output and its input."

Alas, this electronic oscillator article completely neglects that entire class of oscillators. Or is "time delay oscillator" a synonym for "relaxation oscillator"? --68.0.124.33 (talk) 03:38, 24 August 2009 (UTC)

I was under the impression that a ring oscillator is in fact an example of a harmonic oscillator (time delay oscillator, or feedback oscillator) and not a relaxation oscillator. A ring oscillator can be thought of as two inverting gates forming the amplifier part and a third inverter acting as the negative feedback. My personal feeling is that the ring oscillator entry should be moved to the harmonic oscillator section. --Hsauro (talk) 16:27, 29 October 2009 (UTC)

Relaxation versus LC oscillations[edit]

I have created a new section about the nature of relaxation and sinusoidal oscillations but it was removed without any explanations. To help understanding the basic ideas behind the nature of relaxation and sinusoidal phenomena, I have copied below a discussion extracted from the related page of Circuit idea wikibook. I have compressed it to not irritate non-thinking persons. If you do not belong to this category and you want to realize the truth about these circuits instead, click the show button below to see the extended content. Circuit dreamer (talk, contribs, email) 20:11, 1 January 2011 (UTC)

Extended content

Relaxation oscillations[edit]

(...We might explain what an LC oscillation is, if we say what it is not. Here, we might say that an LC oscillation is not a relaxation one. Only, we would explain what a "relaxation oscillation" is. This phenomenon is no less vague and mystic than the LC phenomenon (I have never met good resources about it). Now, we have opportunity to reveal the secret behind relaxation phenomenon by comparing it with LC one. After that, we may move these thoughts to the future story about relaxation oscillations...)

A relaxation oscillator has only one accumulating element; the energy moves only in one direction.

The basic idea. According to the powerful idea, we might first build the simpler and more intuitive relaxation oscillator by using three elements: a flow (current) source, an accumulator of potential energy (a capacitor) and a load (a resistor). Such an arrangement operates in two stages as follows: in the first stage, we connect the current source to the accumulator and begin charging it; in the second stage, we connect the accumulator to the load (turn on the switch SW) and discharge it. Let's repeat it in other words: in a period of a relaxation oscillation, we first draw a kind of energy (usually, potential) from the flow source to fill the accumulator, then "suck dry" the energy from the accumulator and just "throw out" it to the environment. Thus, in the first stage of every period the source has to restore the energy in the accumulator.

Features. Note that the relaxation arrangement needs only one accumulator. It lies on the way of energy and serves as a buffer that interrupts temporarily the flow (only, the average flow remains constant). In this way, it intermits, tears to "pieces" the energy flow. It is interesting fact that a continuous current enters the accumulator while an intermittent current leaves it; so, we might say that a relaxation oscillator "discretisizes", "portens", "doses" the unceasing input flow of energy. At the same time, the accumulator's contents increases and decreases periodically. The shape of the curve is not sinusoidal because at the points of switching (at the peaks) the "flow-creating" quantity has a maximum magnitude. Note also that increase and decrease can have different durations. Examples of this possibility (see the analogies below) are a photoflash (slow charging, fast discharging), a toilet tank, a husband that fills slowly and continuously a family money deposit while his wife empties rapidly and regularly it:)

What does "relaxation oscillation" mean? I have been asking myself many times what the word "relaxation" means in this context. What does "relax" here? Maybe, the capacitor? If yes, when does the capacitor relax? Maybe, when we charge it (in this case, "relaxation" would mean restoring, retrieving, recreating the content of the accumulating element). Or maybe, the capacitor relaxes (returns to its initial "zero energy" state of equilibrium) when we discharge it...

Relaxation analogies. We can see many manifestations of this phenomenon in nature, routine, engineering and finally, in electronics. For example, I have a remembrance from my childhood how sometimes faulty toilet tanks began working intermittently (filling -> emptying -> filling -> emptying... up to infinity:) Similarly, imagine you were foolish enough to make a common money deposit with your wife:) Then, you begin working hard to feed up continuously the deposit. As a result, the amount of money increases continuously through time. Only, at some point your wife decides that's enough and draw out all the money accumulated. You begin working again... and she empties the deposit again... and this repeats again... and again... and again...:) Circuit-fantasist (talk) 20:06, 18 January 2008 (UTC)

Comment[edit]

The above does not qualify as WP:RS. Glrx (talk) 17:18, 1 January 2011 (UTC)

Simple, obvious and clear for everyone truths do not need sources... It would be better if you discuss here the contents of the new section about the nature of relaxation and sinusoidal oscillators instead to remove it mechanically... Circuit dreamer (talk, contribs, email) 17:36, 1 January 2011 (UTC)
Although it's interesting, I feel it's not obvious. Unless you can cite a source, it's WP:OR. --ChetvornoTALK 17:49, 1 January 2011 (UTC)
Please, do not waste my time; it is valuable. I have listed one similarity and eight differences between the two types of oscillators. I have been thinking about these phenomena a long time ago... If you can, help me with editing them (tell here what is "not obvious"); if you can't, please do not impede me. Circuit dreamer (talk, contribs, email) 17:55, 1 January 2011 (UTC)
I have removed C-F's WP:OR wiki-book content from the talk page. C-F has been told several times in the past that this material does not belong on talk pages or in articles. If C-F can supply citations for this new material it can be included. Zen-in (talk) 18:27, 1 January 2011 (UTC)
As I can see from your contributions, the only thing that you can do in Wikipedia is removing... Circuit dreamer (talk, contribs, email) 19:49, 1 January 2011 (UTC)

Hi this is my first post, sorry If I accidentally clobber anything, There are significant differences between "relaxation" and "resonant" oscillators , (personally I would not call the latter "Harmonic", this term is used by physicists, while EE's use crystals in 3rd overtone to make true harmonic oscillators), and is important to distinguish the two types, I think "resonant oscillators" is a better name, and it implies the resonant interchange of energy. Getting somewhat off track the article fails to mention "parametric oscillators" and "phase lock loops" too , some might argue relaxation oscillators are part of the parametric family; others may argue YIG oscillators are also part of the parametric family. Not mentioned anywhere is the order of an oscillator, relaxation oscillators are usually first order R + C (but can be second or third or delay line) critically a first order oscillator frequency is linearly dependent on any parameter value while a resonant oscillator the frequency is a square-root dependance. Phase shift oscillators of the third order type (3 x RC's at 60deg shift can be made as sinusoidal or relaxation types to add more confusion). A true relaxation oscillator has a Q less than 1, has a startup time down to zero, and can be modulated upto its fundamental frequency (e.g. most PWM's), they are almost always strongly voltage dependant, (I think an essential requirement is the existence of some (non linear) trigger mechanism) ; on the other other hand a harmonic/LC/sinusoid has a small signal linear representation so can be modelled accurately, an EE could carve out this type of oscillator by requiring a dominant complex pole pair (but that means nothing to the average wiki reader) . Sinusoidal oscillators require some form of amplitude control, and the article is really sloppy describing this, there are two ways to achieve this (a) faster or (b) slower than the oscillator frequency. Faster implies some type of clamping (e.g. third harmonic compression) this creates artifacts in the output spectrum at the harmonics , slower implies some form of automatic gain control (a.k.a. automatic amplitude control) e.g. with a light bulb, the cost of this approach is broadening of the phase noise, and some amplitude modulation when the frequency is rapidly varied. Also conspicuous by it's absence is any kind of formulae, there should be at a minimum , F=1/(2.pi) . sqrt(L/C) , and a passing reference at least to Q and damping and the effect of R in an LC circuit. In terms of presenting material to the reader, I would favour adding a table comparing various aspects of relaxation vs harmonic , which would cut through a lot of the existing waffle. Adding a few more circuit examples, e.g. the 74HC14 RC oscillator would help.

    Salbayeng (talk) 00:34, 14 February 2016 (UTC)salbayeng

Posing the removed section for a discussion[edit]

I have placed below the text of the removed section (in bold) for a discussion. I have accompanied it with explanatory text (in italic) to make obvious things more obvious. Please, insert your comments between the points. Circuit dreamer (talk, contribs, email) 14:26, 2 January 2011 (UTC)


Comparison between relaxation and LC oscillators

Although the two types of circuits above produce oscillations, there are significant differences between them; they are revealed in the text below.

Similarities:

  • Both the oscillating circuits contain (at least one) accumulating element that acts either as an integrator or as a source. If you prefer, we may use "storing" instead "accumulating" element. This element can store something (usually energy) acting as a load and then it can produce energy (taking it from the stored one) thus acting as a source. A capacitor is a typical example (when we charge a capacitor, it acts as a load; when we discharge a capacitor, it acts as a source).
A sine-shape oscillation.
A triangle-shape oscillation.

Differences:

  • A relaxation oscillator consists of only one accumulating element while an LC oscillator consists of two accumulating elements. Look at any relaxation oscillator (e.g. a neon lamp oscillator, 555 timer, op-amp relaxation oscillator, etc.) and you will see only one capacitor (an astable multivibrator contains two capacitors as it is built by two identical relaxation monovibrators).
  • The relaxation accumulator is either flow or pressure like; the LC accumulators are heterogeneous (the one is flow-like and the other is pressure-like). The two storing elements of an LC tank have to be heterogeneous in order to be fully discharged (e.g., if you connect a charged capacitor to an equal but empty one, the both capacitors will be charged to half the initial charge).
  • A relaxation oscillator stores only one kind of energy (usually potential) in the accumulator while an LC oscillator stores two opposite kinds of energy (kinetic and potential) in the two accumulators. A capacitor of a relaxation generator stores potential energy; the capacitor of an LC tank stores potential energy while the inductor of an LC tank stores kinetic energy.
  • In a relaxation oscillator the energy is wasted while in an LC oscillator it is treasured temporarily in an additional accumulator with the purpose of future usage. That is why LC oscillators are more economical than relaxation ones. Look at any relaxation oscillator, e.g. a neon lamp oscillator. The source charges the capacitor and then the energy of the capacitor is dissipated by the neon lamp.
  • Resonance phenomenon does not exist in a relaxation circuit; it can be observed only in an LC tank. The resonance uses the stored energy in the second accumulating element; the input sinusoidal source only adds (superimposes) a small portion of energy to this energy; so, it can be observed only in an LC tank.
  • In a relaxation oscillator the energy moves only in one direction (source -> accumulator -> load) while in an LC oscillator the energy changes periodically its direction (it circulates between the two accumulating elements). Look at some relaxation generator, e.g. 555 timer: the capacitor is first charged by the power supply; then, it is discharged by the resistor connected in series to the open collector transistor (i.e., the energy moves from the power supply through the capacitor to the discharging resistor). In an LC oscillator, the capacitor charges the inductor, then the inductor charges the capacitor (i.e., the energy circulates between the capacitor and the inductor).
  • The shape of a relaxation oscillation is peaked, angular while the shape of an LC oscillation is rounded (sinusoidal). The reason of that is that at the peaks of the wave the source of a relaxation oscillator changes sharply its output quantity while the "source" of an LC oscillator (charged accumulator) does not change its output quantity. Look at point b in the first figure on the right: the magnitudes of the quantity in the moments Tb-1, Tb and Tb+1 are almost equal. Now look at the second figure: the magnitudes of the quantity in these moments are quite different.
  • The shape of the relaxation oscillation can be asymmetrical (the increase and the decrease can have different durations) while the shape of the LC oscillation is precisely symmetrical. We can charge and discharge the capacitor of a relaxation oscillator through different resistors (to have different time constants); but the capacitor and the inductor of an LC circuit exchange the same energy.

Circuit dreamer (talk, contribs, email) 14:26, 2 January 2011 (UTC)

Comment[edit]

  • Oppose. There are no citations in the above material. It is WP:NOR, and the editor has not offered any WP:RS. Furthermore, the original wikibook material is the author's own work, so it has a self-citing issue. Glrx (talk) 16:29, 6 January 2011 (UTC)
As usual, the same idle talk again... Have you written my detailed explanations and examples in italic? Can you make (at least one) reasonable comment about the topic? Do you understand something from the written at all? And where have you seen some references to a wikibook material? Circuit dreamer (talk, contribs, email) 18:16, 6 January 2011 (UTC)
I understand what you are saying, but I don't think you understand what we have said. WP is not about WP:TRUTH. To include material you need WP:V and WP:RS. If you cannot provide citations, then the material should not come in. Glrx (talk) 19:16, 6 January 2011 (UTC)
Thank you for the link to the funny essay. I was amused with it and now I am inspired to continue exposing great circuit ideas on Wikipedia pages:) Now a bit more seriously... You are right about citations but only in the case when there is a need of citing. It is more than obvious for everyone (including you) that simple, trivial, banal, obvious, convincing and cogent assertions do not need any citing. Otherwise we should second with citations every assertion like "the black is not white" and v.v., "the white is not black", "the sun rises in east", "the capacitor stores energy", "a resistor dissipates energy", etc... Thus Wikipedia articles would be full with references diverting readers outside of articles; Wikipedia would serve just as a link portal...
I have taken the trouble to inspect your contributions since 2007 till now to see what type of person you are and I have ascertained that we are quite different. But we have to reach an agreement for the sake of Wikipedia. The main problem is that you have not made any comments about the contents of my concrete items above. Thus I am under the impression that you do not understand the topic; everything is new for you and everything written by me seems as original research for you. Please, deny my impression simply commenting my sentences (saying what is wrong and what is not...); I will answer to you and then we will decide if there is a need of citing. To make easier this process, I begin breaking up the sentences above, posing them step-by-step and waiting for your comments. Let's begin with the similarities between the two versions. Circuit dreamer (talk, contribs, email) 06:56, 7 January 2011 (UTC)


Comparison between relaxation and LC oscillators
Similarities:
  • Both the oscillating circuits contain storing elements acting either as integrators or sources.
Now look at any oscillator. Do you see an oscillator without a storing element as capacitor or inductor? Is it possible to obtain a signal that changes through time without using time-dependent (storing) elements? Should I second this fact with citations? Then, look again at these elements. Are you agree that they are reversible, i.e., they can store energy (acting as integrators) and then they can use the stored energy to produce energy (acting as sources)? I ask you again, "Should I look for sources seconding these basic truths from elementary physics and electricity?" I am waiting for your comments (not for common phrases!). Circuit dreamer (talk, contribs, email) 06:56, 7 January 2011 (UTC)

Negative resistance LC oscillator[edit]

The article explains LC oscillators with a feedback network. Then, after doing the amplified noise going round several times to build the oscillation, Circuit Dreamer invokes a negative resistance. He claims his view is justified because there are lots of Google hits. My complaint is the mixed view is not needed and is confusing. The notion of negative resistance is used in the design of some oscillators, but that is not the dominant view. Negative resistance design is common when the gain element is reduced to a one port (so there are no explicit feedback paths). With a one port, a negative resistance is required to make up the resonator losses. The technique is common in microwave designs that use arbitrary resonators (and where explicit feedback paths will have significant phase shift problems). LC oscillators are uncommon above 1 GHz. I intend to revert Circuit Dreamer's reinsertion of negative resistance in LC oscillators. Glrx (talk) 19:51, 25 July 2011 (UTC)

Note I have said as though it is, not it is a negative resistance circuit. The compensation idea can be implemented either by positive feedback or negative resistance but the final result is the same - the internal resistive losses are canceled. But it is always useful for understanding to show a few different ways to the same truth. BTW the phase-shift oscillator is based on a negative feedback configuration. Circuit dreamer (talk, contribs, email) 20:26, 25 July 2011 (UTC)
If the analogy requires qualification ("as though it is"), then it adds confusion and should not be used. The amplitude control of oscillators is not simple. See, for example, Streeter's Strauss' book (fixed Glrx (talk) 15:48, 9 August 2011 (UTC).) (gain compression used for stabilization) or Oliver's astounding HP Journal article (non-linear amplifier gain is required for AGC stability). Your reference to phase-shift oscillator is irrelevant; it is not an LC oscillator. Even so, your premise is confusing because it suggests that negative feedback is magically transformed into positive feedback; at the design frequency, the phase shift network has negative gain; the feedback loop gain at the design frequency is positive, so it's positive feedback. Positive feedback is needed to support all nearby frequencies; higher gain near center frequency selects those freqs over others; no sharp selection at nearby freqs admits phase noise. Glrx (talk) 21:31, 25 July 2011 (UTC)
You have shown a good understanding of the topic in details. I have tried to use one of these sentences in Negative resistance but it was not so successful attempt...
This page is not solely about LC oscillators (it is named Electronic oscillator, not LC oscillator) and its LC oscillator section is not solely dedicated to positive feedback configurations. Phase-shift oscillator is a kind of oscillator and it should be considered as well. Negative resistance LC oscillator is a kind of LC oscillator and deserves attention as well. It seems that LC oscillator section should be divided in two subsections.
IMO this is a general article about electronic oscillators. So it has to be more philosophical, to show the concepts benihd oscillators than specific implementations. My suggestions above along these lines continue being open for discussions. Circuit dreamer (talk, contribs, email) 16:55, 26 July 2011 (UTC)


Yes, the article is about electronic oscillators, and the section was about harmonic oscillators, but the paragraph is about LC oscillators. Although there were tunnel diode LC oscillators in the 1960s, that was because transistors didn't have a lot of performance. That is not the case today.
The negative resistance comments remain a troubling tangent / confusing addition to the article.
Do you have a reliable source for the statement "The amplifier behaves like a true negative resistor that cancels the internal resistance of the LC circuit"? (The statement also has a slight factual error; it must cancel all losses -- not just the internal losses.)
Although you like the notion of a negative resistance canceling the coil losses, the argument is not that simple. An oscillator must have excess small signal gain to start up. From a negative resistance viewpoint, the neg R must more than compensate for the losses near the bias point. At larger amplitudes, the neg R must do less than compensate for the losses. The value of R changes throughout the cycle; it is not a simple resistance cancellation but rather a harmonic energy balance argument. The statements are more direct when they are made about amplifier gain rather than negative resistance.
I disagree with your philosophical comment. WP should be straightforward and clear. In many of your edits, you try to generalize an idea or to connect two ideas. Although I understand your philosophical interest in generalizations and parallels, that interest takes you into WP:OR and WP:SYNTHESIS -- things that do not belong in WP. It also means that your abstractions do not have a WP:RS. That is one reason why your generalized ideas about relaxation oscillators above were removed. It is also a reason your statements about LC and relaxation oscillators did not gain consensus. My guess is that same philosophical interest is why you want to bring up negative resistance and talk about cancellation.
FWIW, I'm not sure that a phase shift oscillator is properly a harmonic oscillator. It generates a sinewave, but I'm not sure that is enough. It does not have a typical resonator. I know it's a feedback circuit with three LHP real poles; the root locus creates a complex pair that migrates toward the RHP. Those ideas fit with one of your comments about relaxation oscillators not having a resonance. Does that mean a phase shift oscillator is a relaxation oscillator? It doesn't have a switch element, but it does have negative feedback. Generalizations can get muddy fast. When statements do not have reliable sources, then they don't belong in WP.
Glrx (talk) 19:22, 26 July 2011 (UTC)


Glrx, thank you for reply; I will comment it with pleasure. It turned out you are good company in this area of Wikipedia.
Negative resistance versus negative feedback. Obviously, negative resistors are not perfect resistance neutralizers. They are not "interested" in the final result of compensation; they compensate only positive resistance with the same value (a disadvantage). So, if a positive resistance R changes (e.g., because of temperature variations), the negative resistor will not compensate the change; it will compensate only an R part of the whole positive resistance. But negative resistors are 2-terminal and non-inertial devices (advantages). In contrast, circuits with negative feedback neutralize any changes of positive impedance elements as they "observe" the final result (virtual ground) of compensation (advantage). But they need a third wire to "sense" the result of comparison and they are slower because of the negative feedback (disadvantages). Both negative resistance and negative feedback circuits do the same (compensate resistive losses) but in different ways.
My motives. Now about my "philosophical interest" in generalizations and parallels... My pursuit is to reveal the fundamental ideas behind circuits - to show not only what circuits do but how and why they do it. Inventors, scientific researchers and producers do not like to disclose the secrets behind circuits as they want to benefit from them. They have no time to explain them to people; they do not see any reason to do it for people; they have to earn money (for themselves or for their employers). See for example Widlar's articles in National Semiconductor application notes. Bob Widlar was circuit genius but only try to understand something from his explanations there... I needed years to comprehend his clever circuit tricks in LM301 input stage, bilateral current source, capacitance multiplier, etc... Bob Peace is a remarkable circuit designer but only try to understand (from his reputable EDN article) what the basic idea behind the transimpedance amplifier is, what the op-amp does and why it does it in this 1-resistor circuit. Will you read there the simple truth that the op-amp just add voltage that is equal to the voltage drop across the resistor to compensate it? That it acts just as a small varying "battery" producing voltage VR? Or that the op-amp behaves just as a true negative resistor with resistance -R neutralizing the positive resistance R (R - R = 0)?
I have refused to benefit something from my insights about circuits; I am not obliged to generate them for someone; I am free to share them with people... I do not bury them in some pay sections, articles or books where they will die. I do it in Wikipedia because of its highest Google rank. Now, if some curious young visitor type "negative resistance" in Google and then click on the first Google suggestion; he/she will learn what differential and true negative resistors are, what the difference between them is, how to make them, etc. The only problem is how to bridge over WP:OR but you have to admit that I have made a sizable headway along these lines.
RC oscillators. Your notes about the nature of phase-shift oscillator are wonderful. I will add to this discussion all RC oscillators (e.g., Wien bridge) that are a big challenge for human imagination. Why? Just because it is too hard for a mere mortal:) to imagine how the humble RC circuit can produce sine wave, how it can act as a "resonator" at all. Three years ago I managed to reveal how the more sophisticated LC circuit does this magic. Then I began thinking about how the humbler RC circuit could do it... and this was a big challenge for my imagination. Here are my intuitive achievements about the most general (philosophical:) idea behind RC oscillators. I have used, as usual, a figurative and colorful language to picture the circuit operation.
RC oscillators stay between relaxation and harmonic (LC) oscillators; they possess properties from the both. Like relaxation oscillators, they have only one storing element (capacitor) that continuously charges and discharges; it stores only one kind of energy (electric) that is wasted. Like LC oscillators, the storing element is connected in a positive feedback loop to sustain the oscillations; they produce "rounded", "smooth", sine waves... Let's see why and how.
Simply speaking, both the relaxation and RC oscillator consist of a voltage source (an amplifier) driving a capacitor through a resistor. To make the voltage across the capacitor wiggle, this source has somehow to change its polarity at the peaks of the halfwaves.
  • In a relaxation oscillator, the amplifier output voltage stays constant (maximum or minimum, at one of the supply rails) until the capacitor charges/discharges. When voltage drop across the capacitor reaches the peak, the amplifier switches sharply (helped by the accelerating positive feedback) this voltage from the current to the other rail. As a result of this voltage jump, the shape of the relaxation oscillation is peaked, angular, not sine...
Classic Wien bridge oscillator
  • In an RC oscillator (Wien bridge oscillator is a good example), the storing element (the grounded capacitor in the figure on the right) is connected in the positive feedback loop. (IMO) the loop gain has to be close to but yet a little more than unity. At these conditions, the amplifier output voltage is constantly a little higher than the voltage drop across the capacitor and the latter continuously charges. The capacitor voltage tries to reach the amplifier voltage that continuously shuns up because of the positive feedback. Figuratively speaking, the capacitor is "self-charging"; it "pulls up" itself (with the help of the supplied amplifier) like Baron Münchhausen escaping from a swamp by pulling himself up by his own hair:) If the loop gain was exactly unity, the amplifier output voltage would be equal to the voltage drop across the capacitor... no current, no voltage change, no wave... Another impressive analogy is a cage equipped with "antiweight". Imagine you are in the cage but some "joker" has increased slightly the antiweight and, of course, loosed the brakes:) As a result, to your great surprise, you will begin lift up just like the voltage across the capacitor... If the antiweight was equal to the cage weight, you will stay immovable.
When the capacitor voltage approaches the positive supply rail, the amplifier begins saturating; the loop gain begins decreasing and the voltage change looses its nerve. Finally, at the top of the halfwave, the amplifier does not amplify at all (unity gain) and the voltage stops changing; thus the upper sine peak. Now the grounded capacitor begins discharging through the parallel connected resistor and its voltage goes back down (note at the peak there is no charging current from the amplifier output since the upper capacitor impedes it). The positive feedback helps this process as above (now the joker has decreased slightly the antiweight and you begin travelling down:) The voltage begins to decrease trying to reach the amplifier voltage that continuously "shuns" down. When the capacitor voltage approaches the negative rail, the amplifier begins saturating; the loop gain begins decreasing and the voltage slows its change. At the bottom of the halfwave, the amplifier does not amplify and the voltage stops changing; thus the bottom sine peak.
As a final conclusion, RC oscillations arise because of the slight positive feedback with dynamic loop gain (between peaks it is bigger than one; at the peaks it is exactly one). The shape of an RC oscillation is smooth (sinusoidal) since at the peaks of the halfwaves the amplifier output is saturated and does not change its voltage (just like an LC oscillator).
These were only my insights. I would be glad if you share and enrich them. Regards, Circuit dreamer (talk, contribs, email) 20:12, 28 July 2011 (UTC)
The continuing problem with your comments is the comments are your insights -- also known as your WP:OR. They may be true, but they do not belong in WP unless they have WP:RSs. Neither you nor I are reliable sources. There are reliable sources for harmonic balance and gain saturation.
Phase shift and Wien bridge oscillators are not mysterious. One can imagine setting up a second order differential equation on an analog computer. If the solution includes a sinewave, it's not too surprising that one would happen.
Glrx (talk) 21:08, 28 July 2011 (UTC)
I agree with Glrx's comments. Remember that this article will be read by nontechnical people, Circuit dreamer. The term "negative resistance" is usually used to describe individual one-port active elements with negative resistance such as tunnel diodes. By using it for feedback oscillators, the article risks confusing people, implying that the active element in the circuit, the amplifier, has negative resistance. The goal is to explain feedback oscillators in the simplest, most understandable way. Cheers. --ChetvornoTALK 16:14, 29 July 2011 (UTC)
I'll use Chetvorno's comments as a WP:3O to remove the negative resistance explanation. Glrx (talk) 18:50, 29 July 2011 (UTC)

There may be grounds for concern at the emphasis on positive feedback in this section. The defining feature of any oscillator is negative feedback. On its own, positive feedback would make the circuit into a switch - once fully on (or off) it would remain there. Negative feedback is essential to counteract this. Trevithj (talk) 00:37, 10 February 2012 (UTC)

Linear oscillators use positive feedback. The signal must be self-sustaining, and that requires positive reinforcement. In theory, the loop gain must be precisely +1 for a constant output amplitude. See Barkhausen stability criterion; there's a better statement about the phase angle of the gain. In practice, the loop gain averages to +1. If the loop gain is > 1, then the output amplitude is growing; if the loop gain is < 1, then the output amplitude is decaying.
The issue for WP always comes down to reliable sources. Do you have a source for the statement that defining feature of any oscillator is negative feedback?
Glrx (talk) 03:36, 10 February 2012 (UTC)
Agreed that reliable sources are necessary. John Sterman in Business dynamics describes the need for negative feedback in an oscillation.
Agreed also that a sustained oscillation needs to be driven. Self-sustaining would require enough amplification (positive gain) to overcome losses. By definition, a loop gain of 1 doesn't grow the amplitude, so isn't reinforcing. It would need to be greater than 1 to count as positive feedback. Anyway, a dampened oscillator is still an oscillator.
It isn't that your points are unimportant - but the use of "positive feedback" in this context is confusing, and maybe not appropriate in context.
Trevithj (talk) 23:51, 20 February 2012 (UTC)
Trevithj, you're confusing negative feedback with the restoring force in a harmonic oscillator (for example, gravity in a pendulum) which has to act in an opposite (negative) direction to the deviation to push the mass back toward the equilibrium position. Electronic oscillators and harmonic oscillators aren't analogous, they are totally different beasts. --ChetvornoTALK 04:26, 21 February 2012 (UTC)
I think the two are analogous beasts, but the story is more complicated. The restoring force is a negative function of position. The feedback system must compare the same dimensional quantity. Newton relates a force to acceleration. Differentiate position, and you get a velocity. Differentiate velocity, and you get an acceleration. Laplace tells us differentiation is multiplication by s = . The repeated differentiation turns imaginary j squared into -1, so that negative restoring force develops positive feedback for position at the resonant frequency. Glrx (talk) 20:01, 22 February 2012 (UTC)
No, a harmonic oscillator is not analogous to an electronic oscillator; it doesn't contain a source of energy so it can't produce continuous waves. A physical harmonic oscillator always has friction or other dissipative processes so it loses energy, producing damped oscillations which decline to zero; its poles are in the left half plane. An electronic oscillator consists of a harmonic oscillator, the LC circuit, with the addition of a positive feedback loop. An LC circuit alone can produce electrical oscillations, but because it has resistance the oscillations die out. The feedback loop applies pulses of electrical energy from an external source to the harmonic oscillator to replace the energy lost in the resistance. This moves the poles of the combined circuit to the jω axis. In mechanical terms it provides a "drive force" to the HO. The "positive" in positive feedback refers to the timing (phase) of the pulses of drive force; to add energy they must be "in phase" with the oscillations of the harmonic oscillator. This only occurs when the phase shift in the feedback loop is 0 or a multiple of 360° - positive feedback. (negative feedback is a phase shift of 180°) (see Barkhausen stability criterion).
It's like when you push someone in a swing. The swing loses energy to friction so to keep it oscillating you have to push it. But you have to apply the pushes at the proper times - when the swing is moving in the same direction as your push. That's positive feedback. --ChetvornoTALK 02:36, 23 February 2012 (UTC)
Please explain the difference between negative feedback and restoring force. They sound identical, although given the lack of citations around the second entry, it is hard to be sure.Trevithj (talk) 21:38, 27 February 2012 (UTC)
I think you are keying on the sign of the quantity and making the wrong leap. The restoring force has a negative sign on it; when one is displaced a distance x from the equilibrium, there's a restoring force −kx for positive k. Even though the restoring force is negative, that does not mean the feedback loop must be negative. My description above points out that at the resonant frequency, another factor of -1 slips into the loop gain to make it a positive feedback loop. Don't directly compare a position difference (something that the feedback loop sees) with a force (something created by the difference). Glrx (talk) 19:17, 1 March 2012 (UTC)
Maybe there is a difference in use of the terms "positive/negative feedback" in electronics. I haven't found clear evidence of this, but given the extra connotation of "polarity", I guess its possible. I notice that many authors (Sterman, Senge) avoid "positive/negative", and talk instead of "balancing" or "reinforcing" loops (respectively). In the swing analogy, if the strength of each push is related to the height of the back-swing, then it would be a reinforcing loop. This would make the swing go higher and higher. If however the push is constant, then there is no feedback. More likely, the push will reduce if the swing gets too high (which is a balancing loop). In other words, gain control. Trevithj (talk) 21:40, 27 February 2012 (UTC)
Your comments are confused. There's feedback for the oscillator. There may or may not be another feedback loop to control the amplitude of the oscillations. Glrx (talk) 19:51, 1 March 2012 (UTC)
I hate to be a killjoy, but this interesting discussion has gotten very off-topic for a talk page, which is supposed to be limited to improvement of the article. Maybe continue it on someone's personal talk page? --ChetvornoTALK 21:04, 1 March 2012 (UTC)

Classification[edit]

I do not have a good sense about oscillators are classified - or filters for that matter. There are certainly RC and LC and quartz oscillators, but there may be a more general classification. To me, an RC (and the odd RL) oscillator are oscillators where some magic has been done to create the imaginary poles; I don't know what they are called. Then there are oscillators built from resonators -- the complex poles are already there but the losses must be compensated. Consequently, I lump LC, crystal, YIG, dielectric, transmission line, and cavity oscillators together. I know my view has problems: all but the first have multiple modes, so the desired mode must be selected. For some (such as an SC cut crystal), the undesired modes are closeby.

So the question becomes is there a source that offers a classification? Or is there just this ad hoc naming convention that focuses on the dominant components?

Glrx (talk) 20:11, 12 February 2012 (UTC)

I wrote that section. The terms RC oscillator and LC oscillator are pretty widely used classifications in electronics 1, p. 2-7, and books on circuit analysis usually analyze RC and LC oscillators separately. I'm not suggesting this is the only classification, but the article should explain these terms.
Your intuitive analysis that there are two categories - RC oscillators where "some magic has been done to create imaginary poles" and resonator oscillators where " the complex poles are already there" but must be moved into the right half plane - is pretty close to my classification. It's just that your "resonator oscillators" are usually split into two categories: those based on mechanical resonators, that is crystal oscillators, SAW, etc. and those based on electrical resonators; LC oscillators (including cavities, etc.). --ChetvornoTALK 15:22, 13 February 2012 (UTC)


Glrx, do you seek a general classification for "oscillators", or for "electronic oscillators" specifically? One possibility for the former may be found here [1]. The System dynamics school of thought views oscillation as corrective overshoot due to time delays in a control loop. Presumably (in this view) in electronics, the RC/LC circuits are seen as introducing the necessary delay. Trevithj (talk) 21:36, 29 February 2012 (UTC)

Good addition to history[edit]

I just wanted to say I thought the Thomson arc oscillator was an important addition to the history section, Glrx. I read about Duddell's oscillator but I had no idea there was one before that. So I guess Thomson technically built the first electronic oscillator? --ChetvornoTALK 00:53, 26 August 2012 (UTC)

Thanks.
I'm not clear on it, but I think "singing arc" may have been applied to Thomson, too. I don't know if he was first (he's the first arc oscillator I'm aware of), but it also depends on the definition of "electronic". In 1890, there was a "howler" oscillator that used a carbon microphone as the gain element and an acoustic path. Early definitions of "electronics" involved electrons in flight. We might be more generous today.
The article needs work. I'd love to find a reference that sorts the invention mess out. I think lots of radio guys would have accidentally made oscillators, but they probably viewed it as a nuisance. Details about De Forest's work is minimal, and it is never clear how much he understood. Armstrong understood the gain issue and probably the oscillator, but it is not clear he appreciated the application of an oscillator. Meissner probably understood the feedback gain issue, but I think he disclaimed the oscillator. It sounds like Round understood everything. I think the bias at the time was transmitters had to have huge power outputs, and the low power output of a vacuum tube was inadequate.
Glrx (talk) 17:16, 29 August 2012 (UTC)
Yeah, you're point that the history is unclear because of the many people that observed oscillation but regarded it as a nuisance should probably be in the article, if you can find a source. Who was Round? Solid state oscillations were also observed in this period, in point contact crystal detectors, by William Eccles and G. W. Pickard. Feel free to edit it. I'm just going to add some more citations. --ChetvornoTALK 20:15, 29 August 2012 (UTC)
Thanks for your sourced additions! I've crawled through about half of the sources, and they offer a lot of insight. Previously, I'd come across a bare mention that de Forest initially had the grid outside of the tube (I think I saw a picture of an audion with an external solenoid), and later, when he put it inside, it was not between the cathode and plate. One of your refs goes into great detail about that. Another ref discusses the oscillator, and specifically identifies de Forest's as being in the audio range (which Morse 1925 implies). Lots of good material.
I collapsed the inline {{cite ...}} so they wouldn't confuse WP's text compare function. I also substituted {{harvnb ...}} links where there were repetitious citations.
Fabulous work. Glrx (talk) 21:41, 5 September 2012 (UTC)

Comments and questions[edit]

This is an interesting article. It has good content but it also has a few gaps and lacks a little bit of structure. I'd like to make a few comments and ask a few questions.

1. I moved the history section up. It makes sense to me rather at the top than at the bottom.

2. The circuit in the first image produces a square wave. Where does it fall under the 6 categories for relaxation oscillators?

3. The image on the right shows a schematic of a relaxation oscillator that produces a triangular wave.

Relaxation oscillator (triangular wave)

Out of the 6 categories of relaxation oscillators, which one does it belong to?

4. It would be nice if there was a summary table that lists all the varieties of oscillators, something like a tree that merges the two lists of "types" in the same place.

5. Are Armstrong/Meissner, Vackář and cross-coupled oscillators classified under LC oscillators?

6. Are harmonic VCOs types of RC or LC oscillators or both?

7. Are opto-electronic, Tri-tet and Robinson oscillators types of RC, LC or crystal oscillators?

8. Which one of the oscillators under "Types of harmonic oscillators" is a negative resistance oscillator?

ICE77 (talk) 03:51, 31 July 2015 (UTC)

2. (I listed my replies under each of your questions; hope you don't mind) The two lists Types of relaxation oscillators and Types of harmonic oscillators are inaccurately named and should probably be renamed. They are not categories, just lists of example oscillator circuits. They shouldn't be thought of as exhaustive. That said, I'm not sure what the top oscillator circuit is called. It is a very common circuit, the "integrator-hysteresis" oscillator, in all the handbooks, so it should probably be in the list if we can find its name. I'll look around. --ChetvornoTALK 07:21, 31 July 2015 (UTC)
3. That is also a very common circuit, and again I don't know if it has a name. It works similarly to the above oscillator, only the functions that are done by one op-amp in the top circuit are split between two op-amps in this circuit. The right op-amp is the integrator that produces a triangle wave, while the left op-amp acts as a schmidt trigger to switch the integrator direction; it produces a square wave.--ChetvornoTALK 07:21, 31 July 2015 (UTC)
4. That's an idea. However I'm not sure the classification can be broken down any further than "harmonic", "feedback", "RC", "LC", "crystal", "negative resistance" and "relaxation". As I mentioned above, the lists in the article are not really lists of "types" of circuit, but just lists of individual circuits. I'm sure there are many additional oscillator circuits that do not appear in the lists. --ChetvornoTALK 07:21, 31 July 2015 (UTC)
5. Yes. --ChetvornoTALK 07:21, 31 July 2015 (UTC)
7. The tri-tet is a mostly-obsolete vacuum tube LC oscillator that was also used with crystals (most LC oscillators can be). The Robinson is an RC oscillator. I don't know how the opto-electronic oscillator should be classified; it uses a laser so should probably be considered a one-of-a-kind. --ChetvornoTALK 07:21, 31 July 2015 (UTC)
8. The tri-tet is described as a negative resistance tube oscillator [2], but I think the rest of them are feedback oscillators. The negative resistance oscillators don't really have names; they pretty much use the same circuit. A Gunn diode negative resistance oscillator is just called a "Gunn diode oscillator". --ChetvornoTALK 07:21, 31 July 2015 (UTC)

Chetvorno, thanks for the valuable feedback. I separated your answers from the initial message because after replying I noticed that things started to get somewhat messy due to repeated nesting.

2. If the two lists of types are inaccurately named then how would you call them? I agree with your statement that "They are not categories". They are just types of oscillators. If you find out what it's called and if it has a specific category please let me know.

3. You seem to suggest that the circuit has a name but does not necessarily have a category.

4. You seem to suggest this classification:

  • Harmonic
    • Feedback
      • RC
      • LC
      • Crystal
    • Negative resistance
  • Relaxation

If this is the structure we could place the circuits listed under the two lists under the classification I just placed above.

5. Thanks.

7. The Tri-tet oscillator appears to be listed under "Quartz oscillators" (crystal oscillators) rather than under "LC oscillators" as shown at the bottom of https://en.wikipedia.org/wiki/Clapp_oscillator.

8. You previously suggested that the Tri-tet oscillator was an LC oscillator and now you say that it is also a negative resistance oscillator. Does it fall under two categories now?

ICE77 (talk) 16:26, 31 July 2015 (UTC)

2. My feeling is that the lists of oscillators could just be introduced with a sentence saying something like: "Some of the many feedback (or relaxation) oscillator circuits are listed below:"
3. Both the circuit at the top of the page and the one on the Talk page are relaxation oscillators.
4. That classification (pretty much the one in the article now) looks good to me. My feeling is that the individual circuits should not be listed as subsections, if that is what you had in mind. I think it might give general readers the idea that they represent categories of oscillators instead of individual circuits. Also listing those circuits as subsections might give readers the idea that those are the only oscillators in that category. That's kind of why the circuits were originally put in bulleted lists, to distinguish them from the categories. That seems to me to be a good way to present them.
7&8. As it says in the Negative resistance oscillator section, negative resistance oscillators also use LC circuits to determine the frequency. Although it is a negative resistance oscillator, it uses a vacuum tube so it is often described in textbooks with tube feedback oscillators; sort of borderline case. It is obsolete now and not very important. I'd suggest since there are no other NR oscillator circuits listed it could just be left in the feedback oscillator list, otherwise it's gonna look like it's the only NR oscillator.
--ChetvornoTALK 09:06, 1 August 2015 (UTC)

Chetvorno, thank you again for the feedback. I have a few more questions.

2. Do you have any input on the name of the relaxation oscillator at the top of the article?

3. Do you have any information on the name of the relaxation oscillator that I posted in this section?

7. Based on your feedback it sounds as if the Tri-tet oscillator could fall under either LC, crystal and negative-resistance.

9. What would be the category for Lampkin oscillators?

10. According to the article on Geoffrey George Gouriet, the Clapp oscillator is called Gouriet-Clapp oscillator to attribute the invention to both inventors. It would be nice to modify the information in this article.

ICE77 (talk) 04:24, 5 August 2015 (UTC)

Tuning range of crystal oscillators[edit]

Glrx, it seems to me the article should say something about the small tuning range of crystal oscillators. As you say, the frequency of a crystal oscillator can be tweaked over a small range, but not the large range that RC or LC oscillators can; they're not normally considered "variable frequency" oscillators. To tune the output of a crystal oscillator over a wide range, as in crystal-locked local oscillators for superheterodyne receivers, the output must be applied to a phase locked loop frequency synthesizer. I'm not fussy about the wording, but something should be included. What do you think? --ChetvornoTALK 17:09, 16 August 2015 (UTC)

I'm worried about unfair comparisons and dubious statements. The article is about electronic oscillators in general, so it doesn't have to say much about crystal oscillators. I don't want to compare apples and oranges. I don't want to say things that I think are true but might not be.
If it were possible to make an LC (or other) resonator with a Q of 50,000, how much could it be tuned with an external component? It I tried to tune it over a large range with an external C (which will have a Q of 100 or so), don't I necessarily destroy the high Q of the resonator?
I own several mechanically tunable crystal oscillators; I own a couple voltage tuned crystal oscillators (VCXOs). If a crystal could not be tuned, then it would essentially be worthless as a frequency or time standard. When I'm using it as a frequency or time standard, then I (and others) are not interested in a wide tuning range.
IIRC, old narrowband FM transmitters would reactance modulate a crystal and send the output through some frequency multipliers. The desire there is narrowband. Such a crystal oscillator is not intended as a fixed-frequency device.
Synthesizers are not always needed with crystals. One of my books has a block diagram of a Collins aircraft band receiver that uses about 30 switched crystals to tune the channels; it does not use a PLL and does not have a crystal for each channel.
Is the comparison fair? How do I tune an RC or an LC oscillator? I change the value of some of the components. There are variable Rs, Ls, and Cs, so that is easy to do. Well, it's easy as long as I don't want a variable C of more than a few hundred pF; all components have practical ranges. LC audio oscillators are unusual. Crystals usually are selected for high Q. Maybe somebody can build a crystal resonator with a much lower Q that can be pulled a lot more; I don't know. Is there a variable X out there? Maybe somebody can build a variable crystal (with varying pressure or wedge-shaped piezoelectrics); I don't know. Come to think of it, side-scan sonars uses large piezoelectrics and would not want very high Qs because they need reasonable pulse fidelity.
I can change the frequency of a crystal oscillator over a wide range by changing the crystal. Inside a HP200A or its LC equivalent, there will be band switching. High-performance electrically tuned LC oscillators use a lot of bandswitching: large VCO range is the kiss of death for low phase noise.
The simple answer: is there an RS for any statement that we would make? Saying less is safe.
Glrx (talk) 17:15, 17 August 2015 (UTC)
You have interesting gear. I knew that crystal oscillators were tuned with trimmer capacitors; I didn't know about mechanically tuned ones. How do they work? Is there a knob on the front panel that puts varying pressure on the crystal plate?
However, those are certainly the exception. I think the article can safely say (as supported by the WP:RSs below) that the majority of crystal oscillators are used for fixed-frequency applications (clocks, watches, clock signal generators in virtually all digital devices, almost all non FM radio transmitters, frequency synthesized local oscillators in modern radio receivers, carrier signal generators in frequency division multiplex communications channels in telephone networks, cable boxes, etc.) In contrast, since crystal oscillators can give such superior stability, nowadays LC oscillators are mainly limited to applications where (wide range) variable frequency oscillators are needed. A typical variable capacitor with a 9:1 range can tune an RC oscillator over a 9:1 frequency range, and an LC oscillator over a 3:1 range. But a typical crystal's frequency can be "pulled" over a maximum range of 0.3%. [3] (that's the largest I found; other sources say 10-4 is typical). What constitutes a "variable frequency oscillator"? The frequency spectrum of a frequency-modulated crystal oscillator, just like the spectrum of an AM transmitter, consists of a constant "carrier" frequency component at the center frequency, surrounded by symmetrical sidebands. The whole point of using a crystal, just as in an AM transmitter, is to keep the "carrier" frequency constant. Switching crystals does not contradict, but emphasizes the fact that they are fixed frequency oscillators whose frequency cannot be tuned over a wide range.
I agree that your concerns above are important. How about some wording like: "Although crystal oscillators can be tuned over a small frequency range, the frequency is mainly determined by the dimensions of the crystal, so they are generally used as fixed-frequency oscillators." Here are WP:RSs that support that they are used as "fixed frequency" oscillators: [4], [5], [6], [7], [8], [9], [10],[11]. This encyclopedia is for general readers, that may know nothing about electronics. This is the page they will come to for information on the different types of oscillator. Is it fair for us to withhold this important fact, that appears prominently in virtually every introduction to crystal oscillators in electronic literature? --ChetvornoTALK 13:00, 19 August 2015 (UTC)
Your proposed wording says too much and sounds wrong. The frequency is not just a function of the dimensions of the crystal. A quartz crystal is not isotropic. The reasoning is also backwards; even your sources don't make the statement that way. The crystal's high Q, stability, and compact size make them attractive for stable frequency sources; a narrow tuning range is not the driving characteristic. If one could make high-Q, stable, and compact LC resonators, then crystals would not be needed.
I don't have a lot of time now, but I'm quickly going through your refs. Generally, google searches can be useful, but a search string that is too specific can introduce bias. Any search can collect cruft.
  1. [12] A dictionary is not a secondary source for technical opinion.
  2. [13] Shallow coverage, but good perspective as "virtually" a fixed frequency that mentions tuning. Book has narrow focus on communications.
  3. [14] book has better depth, but focus on instrumentation (specifically signal generators). p 251 "Crystal oscillators are usually fixed frequency oscillators, where the stability and accuracy are the primary considerations." [sic] See next.
  4. [15] p 521 "Crystal oscillators are usually, fixed frequency oscillators where stability and accuracy are the primary considerations." [sic] Phrasing here and previous ref triggers my copyvio detector.
  5. [16] Subject limited to radios. p 115 indicates sloppiness: "Most commercial radio sets using crystal oscillators have a limited number of crystal-controlled channels (typically 6-18) available to the user." There is no depth in the book's discussion.
  6. [17] This book is about radio, and it also has no depth. The late J. J. Carr was a popular writer, but he is not an authority, he did not understand material well, and I've crossed swords with him.
  7. [18] Basic electronics text with little depth.
  8. [19] Better depth, and finally some credit for mentioning the crystal cuts. Just a couple sentences on applications. Author's comments about use in VLSI triggered another issue: fixed freqs are a problem, and some clock oscillators are deliberately spread.[20]
Many of the references above lack depth; passing mention does not cut it. To me, a primary source in a technical field will be something such as a paper by Colpitts, Clapp, Butler, or Meacham. A reliable secondary source is where a competent author has surveyed the field and puts those primary sources in perspective. Terman (1943) is a dated example: not only does Terman cover a huge field, but he also gives references to the primary sources. I suspect the better texts above do not have such references; poor referencing suggests cribbing off other authors who had done the work and had the perspective. A secondary but dated source for crystal oscillators is
  • Matthys, Robert J. (1992), Crystal Oscillator Circuits (revised ed.), Malabar, FL: Krieger Publishing, ISBN 0-89464-552-8
Matthys (1992) does not, however, say much about applications; its emphasis is fixed-frequency designs with a discussion on trimming (rather than frequency modulation). He doesn't even compare crystal to other oscillators, so he does not make the generic comment that you seek.
Terman (1943, pp. 488–498) covers crystal oscillators but is even more dated and does not make a statement close to what you want.
  • Terman, Frederick Emmons (1943), Radio Engineers' Handbook, McGraw-Hill
I'd go for something like my statement above, "The crystal's high Q, stability, and compact size make them attractive for stable frequency sources," but my refs do not say that. TM 11-690 p 175 comes close when it says, "A quartz crystal ... is normally used in an oscillator circuit because of its extremely high Q ... and good frequency stability over a given temperature range".
BTW, my oscillators are mechanically tuned by turning the screw on a trimmer cap/piston trimmer.
Glrx (talk) 16:53, 19 August 2015 (UTC)

Request for Comment: Additional wording on crystal oscillators as fixed-frequency oscillators[edit]

From the previous section, there is a dispute about the wording of the crystal oscillator bullet point in the Feedback oscillator section:

  • One editor, Glrx, would like the section left as it is.
  • One editor, Chetvorno, would like to add the following sentence to the section:
"Although it can be tuned over a small frequency range, the crystal oscillator is mostly used in fixed-frequency applications."

Which wording should be used? --ChetvornoTALK 13:00, 21 August 2015 (UTC)

My reasons for supporting the above addition are detailed in the previous section, but briefly I think this elementary fact is necessary for general readers to understand crystal oscillators, and I can't see why there is any objection to including it. The fact that crystal oscillators are mainly fixed-frequency oscillators appears in almost every introduction to the devices in technical literature; here are some RSs: [21], [22], [23], [24], [25], [26], [27],[28]
Although a crystal's frequency can be tweaked over a range of a few tenths of a percent with a capacitor [29], [30], it is a mechanical resonator whose resonant frequency is constrained by its dimensions. Ordinary crystal oscillators cannot be tuned over the wide range that RC or LC oscillators can. It is necessary to understand this limitation to understand the difference between crystal oscillators and other types of oscillators. Note the proposed wording does not say that crystal oscillators' frequency cannot be varied. As Glrx says, there are a few applications that use them as narrow-range variable oscillators, such as crystal-controlled FM transmitters. But the vast majority of uses are fixed frequency: quartz clocks, watches, clock signal generators for virtually all computers and digital devices, oscillators for virtually all non-FM radio transmitters, local oscillators for modern quartz-locked frequency synthesized superheterodyne radio receivers, carrier generators for multichannel FDM communication systems such as telephone trunklines and modems, time bases for test equipment. It is POV to withhold this essential fact from introductory readers on the basis of undue weight placed on a few applications and devices. --ChetvornoTALK 13:26, 21 August 2015 (UTC)
  • The wording of the proposed change makes it hard to decide if it is correct, and the correctness depends on details that are not readily available: "the crystal oscillator is mostly used in fixed-frequency applications." So what is an application? Is "wristwatch" an application. Or is a certain model of a Timex wristwatch an application? Or is each quartz watch ever built an application. What about computers, smartphones, tablets, and the like? They actually have several clocks inside them. Some of these are purely fixed-frequency and all adjustments are made in software. Some of the computer clocks may have a provision to alter the frequency a bit in hardware so that time is corrected gradually instead of jumping from an incorrect to a correct time. Which type of clock is in the majority? Jc3s5h (talk) 14:46, 21 August 2015 (UTC)
In reality there is no such thing as a "fixed-frequency" oscillator or application. RC networks, ceramic resonators, crystals, and temperature-controlled crystals simply provide increasingly stable frequencies; all of these exhibit frequency variations and all are "tunable". Lambtron (talk) 15:57, 21 August 2015 (UTC)
I think you're right that the term "fixed-frequency" might need to be explained better. How about:"The frequency of a crystal oscillator can only be adjusted over a tiny range, a few tenths of a percent, so they are mostly used in applications in which the frequency of the oscillator doesn't need to be adjusted."? Although I'm flexible on the wording I think that's pretty self-explanatory. "Fixed frequency" crystal oscillators far outnumber the ones in which the frequency must be adjusted. Virtually every digital device on the planet; computers, smartphones, tablets, games, etc. is run by a fixed-frequency crystal oscillator clock generator. It has no way to adjust the frequency, because the frequency is not critical. Crystal oscillators that are used for keeping time, such as the real-time clock in computers and the oscillators in quartz watches, sometimes need to be adjustable, and a few of them still have an adjustment screw inside the case that turns a capacitor to change the frequency of the crystal, but in most the crystal now runs at a fixed frequency and adjustment is done in the digital logic, by "skipping" a count every N counts. The crystal oscillators used in most radio transmitters are not "tweakable", they run at the frequency stamped on the crystal. --ChetvornoTALK 16:34, 21 August 2015 (UTC)
I think it is understood that "fixed frequency" means "not designed to be adjustable". All oscillators, even atomic clocks, have some degree of undesired frequency variation. Jc3s5h (talk) 17:03, 21 August 2015 (UTC)
So what wording do you think should be in the article? The thing that bothers me is there is nothing in the section to indicate that crystal oscillators can't be tuned over a wide range like LC or RC oscillators. The article should explain the differences between the types of oscillator. --ChetvornoTALK 17:15, 21 August 2015 (UTC)
  • (e/c) Oppose. The proposed addition is unsourced, so it can be immediately removed. The implied logic in the sentence is backward and flawed. It suggests that "fixed-frequency applications" do not require tuning. Oscillators used as frequency references (about as fixed-frequency as you can get) require tuning to put them on frequency. If HP couldn't tune its 10 MHz 10811 reference oscillator, then it would have to throw them away. It's tunable by about 10 Hz and has a stability of 5×10−10 per day. Even Cs oscillators are tuned. Despite Chetvorno's proposed addition, a better statement would be "Because crystal oscillators can be tuned, they are used in high-stability frequency references." I wouldn't put that sentence in because it buries the issue of Q and stability, but it illustrates a problem with the proposed statement; the sentence implies an exclusion between tuning and fixed frequency, but there's no exclusion. Crystals are usually made to tight tolerances, but a typical tolerance of 100 ppm is not tight enough for FCC transmitter requirements; the oscillator frequency needs to be tweaked to remove the manufacturing variation; it may also have to be temperature compensated. For bare applications such as microcontrollers and microprocessors, there isn't really a "fixed-frequency" requirement; it's OK if the oscillator is overdriven, has crummy frequency stability, varies with temperature, and is even a few percent off nominal. In fact, it would be better from an EMI standpoint if the clock oscillator were spread spectrum rather than fixed frequency. A crystal is used in microprocessors because it is a better alternative than using a low-tolerance and bulky inductor. The purported list of RSs is not reliable; Chetvorno has not addressed my criticisms of those sources (one or more of which may even have copyright violations). If I look at reliable, serious, secondary sources about crystal oscillators, they don't make Chetvorno's proposed statement. Matthys has many crystal oscillator circuits, and some of those schematics include tuning capacitors for crystal. I pointed out above that Matthys has a chapter on trimming the crystal oscillator frequency; it explains the that "A certain amount of ± frequency tolerance must be allowed in manufacturing crystals to a specific frequency." Another serious secondary source, Frerking, Crystal Oscillator Design and Temperature Compensation, 1978, devotes chapter 10 to temperature compensation: a "fixed-frequency application" where temperature sensitive components electronically tune a crystal oscillator to get performance such as "0.5 ppm from −55°C to +85°C in production" (p. 131). I do not have an objection to inserting sourced factual statements. One could certainly say, "crystal oscillators are used as frequency references in electronic instruments such as counters and synthesizers," and give some primary (e.g., instrument manuals) or secondary (textbook) sources. The above proposed statement, however, implies that fixed-frequency applications are not tuned. IIRC, even the above crummy sources do not clearly make the statement that fixed-frequency oscillators are not tuned/trimmed ({{fails verification}}), and the actual circumstance is that precision reference oscillators are tuned/trimmed. The proposed statement is either confused about tuning or inappropriate WP:SYN. The crummy sources could be used to make the statement, "Crystal oscillators are used in fixed-frequency applications." That statement, however, does not convey the reason why crystals are used in those applications. As I said in the above section, I don't know of a source that clearly ties the usage to high-Q, stability, compactness, and cost. Glrx (talk) 18:18, 21 August 2015 (UTC)
  • Reformulate. I perceive heavy disagreement on the meaning of the words tuned/trimmed/compensated/fixed frequency with respect to one another. I listed them according to my understanding with decreasing intended changeability. The last two notions appear to me to express the desire for a lifelong constant frequency, which is of course not meaningful in the light of ever increasing precision.
Personally, I'm convinced that a vast, overwhelming lot size of crystals are dedicated to -as I allow myself to call it- fixed frequency applications, which would happily let go of the compensation/trimming efforts necessary to uphold the desired, fixed frequency. One could even look at the crystal itself not as the per se oscillating device, but just as a synchronizing component in some circuit, which generates the basic oscillations per se.
Formulating a sentence emphasizing the small(?) number of crystals in user-tunable applications and allowing for the vast number of crystals in intentionally-fixed frequency applications would suit my taste.
I admit counting each single watch, mobile phone, and computer, ... as one fixed frequency application. Purgy (talk) 07:33, 22 August 2015 (UTC)
What wording would you suggest? --ChetvornoTALK 10:00, 22 August 2015 (UTC)
Sorry, I neither feel sufficiently knowledgeable in the orchidaceous topic of crystal-synced, tunable oscillators, nor suficiently belonging to some corps diplomatique to satisfy the requirements of opponents to the remark under question, the content of which is immediately palpable to me, considering the numbers of respective lots. Purgy (talk) 07:04, 23 August 2015 (UTC)
  • Leave as is - There is no problem with the section as it currently exists. Chetvorno's proposal to discuss fixed frequency needs a reference. Fixed frequency is a relative term. There is no such thing as absolute fixed frequency. You risk creating a tangential discussion by going down that road. ~Kvng (talk) 15:17, 25 August 2015 (UTC)
I gave 8 references: [31], [32], [33], [34], [35], [36], [37],[38]. But I think you are right about the confusing term "fixed frequency"; in electronics that simply means the frequency is not adjustable. I think a better addition would be: "The frequency of a crystal oscillator can only be adjusted over a tiny range, a few tenths of a percent, so they are mostly used in applications in which the frequency of the oscillator doesn't need to be adjusted." What do you think? --ChetvornoTALK 15:40, 25 August 2015 (UTC)
"...doesn't need to be varied" seems more appropriate to me. There is a difference between something that is adjusted during manufacture/maintenance and something that can be changed by the user of the equipment. SpinningSpark 22:16, 5 September 2015 (UTC)
Sounds good to me. My main concern is that the article include a statement about the tiny tuning range of crystals compared to LC and RC oscillators. They can only be tuned over a tiny fraction of one percent as opposed to a 3:1 range for a typical LC. There is nothing in the article about this defining limitation, which readers need to know. I can't see why anyone would object to including it. --ChetvornoTALK 23:23, 5 September 2015 (UTC)
The statement is still confusing. The logic is still backward. When designers are looking to make stable, low phase noise oscillators, they may turn to use quartz crystal resonators because those resonators have high Q. The statement is also contorted. There is no clear source. A long list of poor sources does not improve reliability. The statement attempts to make a conclusion from two separate ideas, and that is WP:SYN. Yes, quartz crystal resonators can only be pulled a small amount. Yes, quartz crystal oscillators are thus constrained by the resonator to a narrow frequency band that is less than one percent. A single quartz crystal is not a broadband resonator. The notion of fixed vs. adjustable vs. varied is not immediately clear to the reader and is not a fundamental distinction about oscillators. Maybe I want a VFO over a broadband (a LO for a radio) or maybe I want a VFO over a narrowband (a frequency reference) or maybe I don't need to adjust the frequency at all. The statement is also false when one looks at other materials. YIG crystals can be tuned over a frequency decade, and they are used as variable frequency oscillators and filters. The variability sacrifices the Q, so such an oscillator will not have the same normalized low phase noise performance of a quartz crystal. Make simple statements that have sources. Glrx (talk) 04:56, 6 September 2015 (UTC)
So you wouldn't object to a statement that (quartz) crystal oscillators can only be "pulled" (tuned) over a tiny range? That statement should be sourcible to everyone's satisfaction. --ChetvornoTALK 05:57, 6 September 2015 (UTC)
And why not adding a subchapter "Tuneability" containing the statement:
"YIG crystals can be tuned over a frequency decade, and they are used as variable frequency oscillators and filters. The variability sacrifices the Q, so such an oscillator will not have the same normalized low phase noise performance of a quartz crystal, whereas the latter can only be "pulled" (tuned) over a tiny range."
In this subchapter one could also get clear about the distinctions of fixed vs. adjustable vs. varied, which I do not consider as "not fundamental", especially when looking the lot sizes belonging to these three groups, even disregarding their current lack of being well defined.
Sorry, but from my PPOV readers are currently deprived of this economically relevant information by the reluctance of Glrx. Purgy (talk) 10:59, 6 September 2015 (UTC)
Thanks. My preference would be to keep the tunability information with the individual oscillators, but I'm okay with the above, too. Looks like we're approaching consensus. Anyone else have an opinion? --ChetvornoTALK 15:03, 6 September 2015 (UTC)
Kvng's worry was "creating a tangential discussion". In other words, introducing terse statements here and then trying to explain them here does not serve the audience; readers don't need all the details. The object of this RFC should be clear; it should not be a moving target that changes with each reader's comment. The original statement is not clear and has a bad implication. Subsequent variations are dubious or tangential. I don't think this article should be a compare and contrast of quartz and YIG, get lost in Q and phase-noise, or introduce a vocabulary lesson. It's a one paragraph basic intro. The OP apparently wants to make a general statement that crystal oscillators could not be tuned. To me, the OP's underlying observation is that fixed value crystals are available but variable crystals are not. We don't have a source for that observation, and the YIG crystal is a counterexample. Glrx (talk) 20:03, 7 September 2015 (UTC)
Glrx, would you be willing to accept a simple statement that (quartz) crystal oscillators can only be tuned over a narrow range, of a fraction of a percent? That would seem to answer your objection above. --ChetvornoTALK 22:15, 7 September 2015 (UTC)
This bullet point should be a brief introduction. It is already too long. How about improving Crystal oscillator? ~Kvng (talk) 14:00, 11 September 2015 (UTC)
In my humble opinion, I think it should be in this article also. I think this is an absolutely essential point that general readers need to understand the difference between crystal oscillators and other oscillators. Comparing the properties of the different oscillators is what this article is all about. --ChetvornoTALK 15:44, 11 September 2015 (UTC)

To sum up, it looks to me like:

  • There is no consensus for including the phrase "crystal oscillators... are mostly used in fixed-frequency applications." or some variation of it. Purgy, I (Chetvorno), and Spinningspark are in favor of some variation of that phrase, while, Glrx, Kvng, Jc3s5h, and Lambtron are against it.
  • There may be a consensus for including "Quartz crystal oscillators can only be tuned over a small frequency range, of a fraction of a percent." or some variation of this. Purgy and I are in favor, Kvng is against, while, Glrx (?), Spinningspark, Jc3s5h, and Lambtron have not given an opinion.

Is this correct? Any additional opinions? --ChetvornoTALK 04:48, 14 September 2015 (UTC)

Do you have any proposed references for either of these statements? I ask not because I question their accuracy but because looking at the treatment of this topic by other sources would give us some reading on the importance of this point. ~Kvng (talk) 14:12, 14 September 2015 (UTC)
The ARRL Handbook for Radio Communications (H Ward Silver, ed., Newington CT: American Radio Relay League,2014, pp. 9.22 to 9.26) indicates the tuning range for fundamental mode crystal oscillators that are meant to be tuned only during manufacture and maintenance is a few hundred ppm. Oscillators designed as "VXOs", that is, variable crystal oscillators tuned by the operator from the front panel of the radio as the main means of varying the frequency of operation, can be tuned up to 1000 ppm. So I think the second statement is confirmed. It's hard to get a handle on how many crystal-dependent devices are adjusted during manufacture and how many are not, so in the narrowest sense of "fixed-frequency" it's hard to say if most devices are fixed-frequency or not. But I think it's pretty obvious that the vast majority of crystal-controlled oscillators are not adjusted after they leave the factory. Jc3s5h (talk) 14:24, 14 September 2015 (UTC)
[39], [40], p.497, [41], [42], [43], how easily a crystal can be "pulled" is called its "stiffness": [44], [45], [46]. Virtually every technical source on crystal oscillators mentions the narrowness of their tuning range, Kvng. If they don't explicitly discuss how to "pull" the frequency and how many parts per million the frequency can be "pulled", they simply say the crystal oscillator is a "fixed-frequency" oscillator, as shown by the eight references [47], [48], [49], [50], [51], [52], [53],[54] that I gave you the last time you asked. --ChetvornoTALK 19:53, 14 September 2015 (UTC)
I apologize if I'm having you repeat yourself. This has been a long conversation and I have other things going on. I don't find the treatment in most of your second set of sources does not put the fixed frequency aspect front and center in the introduction to the topic. The pullable nature of crystal oscillators is an important topic for Crystal oscillator but not for here. This is evidenced by many of your first group of sources treating pullability as a distinct topic and by the fact that pullability is not mentioned in the lead of Crystal oscillator. The correct thing to do here is to leave the text here as it stands. In any case, we clearly don't have a consensus to change anything. Tweaking your proposal doesn't look like it is going to get us there. It is time to move on. ~Kvng (talk) 20:10, 14 September 2015 (UTC)
I agree. I'm not suggesting discussing pullability, just mentioning the basic difference between crystal oscillators and LC and RC oscillators; if you want to tune it over a wide range, a crystal oscillator isn't suitable. And if there isn't support for including this, of course I won't push it. --ChetvornoTALK 21:06, 14 September 2015 (UTC)
I continue to be uneasy about this topic, but I'll flop to limited tunability should not be mentioned here. The one paragraph summary already states that the reason for choosing a quartz crystal is its high Q and stability. Those are the important characteristics of quartz crystal resonators. Practical LC resonators are not close to a quartz' Q of 50,000. The article states that quartz Q and stability are better than RC or LC (for LC, that statement is not so clear to me re stability; Vackář oscillators are VFOs with 50 ppm stability; stability can be different from phase noise). The limited quartz tuning range is secondary in the existing applications of quartz oscillators. A question in my mind is that high Q and stability may imply a limited tuning range. With YIG crystals, we lose the high Q and stability to get the tuning range. The current text emphasizes what is important; it need not include a secondary detail. Glrx (talk) 23:15, 14 September 2015 (UTC)

I had to edit (addition in italics) the following paragraph, because it seems to have led to misunderstandings:

Meanwhile I got tired reading that it were time to move on, that argumentation were tangential, arguments (argueing for change!) were sufficiently often repeated, information were too long, but should not be expressed in an extra paragraph either, given sources were not explicit enough, leaving things as they are were no problem, being more precise about "fixed frequency" were non fundamental, suggesting a compromise were introducing moving targets, ... and what phrases more were brought up by Kvng and Glrx against any suggested improvement, which would change the current text and would possibly, imho highly probably, improve the value of this article for any unbiased reader.
My interest in improving this article does not anymore surmount this reluctance to any change. For some unfathomable reason, some think that "high quality" and "stability" are properties worth mentioning, but mere "tuneability" is not. So be it. Purgy (talk) 10:53, 15 September 2015 (UTC)

My takeaway from this thread: "fixed-frequency" is a problematic phrase, but stability is relevant to the article and so is variability, as evidenced by the VCO section. Lambtron (talk) 19:10, 15 September 2015 (UTC)

Location of history section[edit]

Placing the history section at the bottom of the article rather than at the top does not make sense to me.

ICE77 (talk) 17:53, 7 September 2015 (UTC)

There should be an explanation of what the subject is about before offering the history of a subject that the reader, potentially, knows nothing about. SpinningSpark 20:03, 7 September 2015 (UTC)
I agree, the reader is most likely to just want to know about oscillators, not their history. --ChetvornoTALK 21:54, 7 September 2015 (UTC)
They might wat to know about history, but at the very least, any terms and ideas that are used in the history need to be presented in the article in their proper context first, otherwise the history will not be intelligible. SpinningSpark 22:00, 7 September 2015 (UTC)
  • Leave as is. Considering that this is a technical article about a small, though important topic on circuits, I think that the prime interest of most visitors is the technical side. Together with the reasons given above I vote for leaving the history section where it is. Purgy (talk) 06:08, 8 September 2015 (UTC)
  • I support Spinningspark's recent move of this section so I guess that's a Leave as is. History is often the first section in an article but I think it makes sense to learn about the topic first in this case. ~Kvng (talk) 14:16, 14 September 2015 (UTC)

Link anchor broken[edit]

The first image on the page has a description saying: "A popular op-amp relaxation oscillator." Here, the words "op-amp relaxation oscillator" link to this wikipedia article: https://en.wikipedia.org/wiki/Relaxation_oscillator#Comparator-based_electronic_relaxation_oscillator

Here, the site linked to is: https://en.wikipedia.org/wiki/Relaxation_oscillator

And it anchors to the line labelled: Comparator-based_electronic_relaxation_oscillator

It seems the relaxation oscillator article was edited. Someone please re-link it — Preceding unsigned comment added by 128.6.36.173 (talk) 15:59, 12 September 2019 (UTC)