Talk:Negative resistance

From Wikipedia, the free encyclopedia
Jump to: navigation, search
WikiProject Physics (Rated GA-class, Mid-importance)
WikiProject icon This article is within the scope of WikiProject Physics, a collaborative effort to improve the coverage of Physics on Wikipedia. If you would like to participate, please visit the project page, where you can join the discussion and see a list of open tasks.
 GA  This article has been rated as GA-Class on the project's quality scale.
 Mid  This article has been rated as Mid-importance on the project's importance scale.

Neon lamps as negative resistance devices[edit]

Neon lamps and their cousins should not be mentioned as negative resistance devices.

The usual story is somebody collects some data points and draws a static IV curve. The problem is the IV curve is not time invariant. It gets modified by the changing concentration of ions throughout the tube. At low currents, there are few ions and the electric field is uniform; after breakdown there's a huge change: there's a high field at the cathode fall. It takes time for the heavy ions to migrate to the cathode to set up that field. The internal structure of the device is changing.

To put it another way, the neon lamp has internal state.

I don't think of an electromechanical buzzer as a negative resistance oscillator; it depends on a spring and a slowly moving inertial armature; I have the same view of neon lamp relaxation oscillators. In any event, a neon bulb is not a simple negative resistance device; there's a lot going on; the internal state of the device is important.

Also, some IV curves are not carefully drawn. Discontinuous jumps and unstable regions are ignored in order to get a conventional-looking curve. File:GlowDischargeVoltAmpere.jpg has a gap between D and E; the redrawn File:Glow discharge current-voltage curve English.svg omits the gap; File:Glow discharge current-voltage curve vertical.svg is based on same image but has sloping returns and adds hystersis. These graphs become so common place that they are accepted without references.

Compare the figure above to the figure here. They are much different. Which one is right? What's going on?

See also Reich (1944) Theory and Application of Electron Tubes, 2nd ed., McGraw-Hill, pp 417 ff. (compare dynatron, p. 380.)

IIRC, some other references describe a slight negative slope on the vertical portion of the second graph just below B, but that small curvature is not something that the conventional relaxation oscillator exploits.

Another reference maps instabilities in the IV plane for hydrogen discharge tubes. Sadly, I cannot find that ref right now.

Do those instabilites show a negative resistance curve or that the ions distribution cannot find a stable configuration? No matter what, it means the IV characteristic is changing; it's not a static thing; that makes neon lamps a lousy example of a negative resistance device.

There's also that thorny little issue about the neon lamp needing some background radiation to generate the initial ions. There's a lot of additional state for a complete description of the neon bulb.

Even the negative resistance portion of an arc discharge is more an example of state change than a fundamental effect of a gas discharge. In one view, local heating starts thermionic emission -- the high currents/ion impacts are heating the electrodes. (In the Poulsen arc, the anode is metal and cooled with water while the carbon cathode is not cooled but rather rotated as it burns up.) Another arc discharge view is field emission.

Yes, there are books that claim the neon lamp oscillator is a negative resistance oscillator, but many of those books start with the static IV curve viewpoint. It's not a good or a clear example of negative resistance. Glrx (talk) 23:45, 10 January 2015 (UTC)

This has been discussed on Talk:Neon lamp. To make sure we're talking about the same things, the diagram at right shows what I think the curve is shaped like. There is a jump discontinuity (bg) as you say, but a portion of the "normal glow" curve (eg) (red) has negative resistance too. Solid black parts of the curve have positive resistance. The limit cycle (abcdea) of the simple RC neon lamp oscillator and its load line (blue) are shown, as given in many sources. This is just Neon bulb relaxation oscillator hysteresis curve.svg transposed. These are my concerns:
Neon bulb relaxation oscillator hysteresis curve 2.svg
  • The jump discontinuity or instability (bg) as shown in most curves, is in the wrong direction to be responsible for the limit cycle oscillation. It is a voltage jump, most curves show the voltage switching between Vb and Vg at nearly constant current. This makes sense physically (avalanche multiplication and appearance of the cathode fall allows the gas to conduct a given current at a lower electrode voltage). But in the relaxation oscillator it is a current jump (dotted lines). When the voltage across the bulb rises to breakdown b, it doesn't drop along the discontinuity (be) as it would in a simple lamp; because of the capacitor the current jumps at almost constant voltage to c. When the voltage across the neon lamp declines from c to e, it doesn't jump back up to b, the current drops at constant voltage to a. During the limit cycle the state of the lamp never traverses the discontinuity (bg) (the "nonstatic" portion of the I-V curve). So the operation of the oscillator would be the same even if there were no discontinuity, even if there were a smooth curve between b and e (as there is in other negative resistances that are used to make relaxation oscillators).
  • You say: "...some other references describe a slight negative slope on the vertical portion of the second graph just below B, but that small curvature is not something that the conventional relaxation oscillator exploits." It seems to me it is essential. The negative resistance (NR) section is (eg). The load line (blue) represents the linear portion of the circuit: . If the load line crosses the bulb's I-V curve at a point of positive resistance (positive slope, the solid black portion of the curve), that is a stable equilibrium point, and the circuit will not oscillate. You can see geometrically that, if (eg) has positive resistance, regardless of how wide in current the discontinuity (bg) is, the load line must intersect the curve at a point of positive resistance, either on the segment (cg) or (ab). The load line cannot reach the discontinuous region (bg) without also intersecting the stable curve. In that case there is no value of R for which the circuit will oscillate. As long as the discontinuity is a voltage (horizontal) jump, the segment (be) must have some region of negative resistance for the load line to pass through (it doesn't matter how "slight" the slope is), or the circuit cannot oscillate. This doesn't prove that neon lamps have negative resistance (the discontinuity can be some more complicated kind of jump) but it is certainly plausible that it has negative resistance before the jump, as shown in a number of the curves in the literature.
Every NR device works by a change in internal "state". A tunnel diode gets its NR from the change between ordinary majority carrier conduction and tunneling conduction. When an S-type (current controlled) NR like the neon lamp is driven by a voltage source, it will have hysteresis (sudden changes of internal state). That doesn't mean it doesn't have NR. The IMPATT diode, which has an I-V curve somewhat similar to a neon lamp with a discontinuity, works by a similar internal "breakdown" mechanism and increase in charge carriers. Going by the above arguments, a lot of other "negative resistance" devices besides the neon lamp, like the IMPATT, couldn't be called negative resistances. As far as I know, no one has ever suggested that.
Of course, what you or I think is immaterial. What matters is what WP:RSs say. There are plenty of reliable sources which say that gas discharges have NR, I gave 4 in Talk:Neon lamp and this article had 9. Here are some more: [1] (note graph 1.54, p. 63), [2], [3], [4], [5], [6], [7], [8] You haven't produced a single source that explicitly supports the above alternative theory. The difference between a jump discontinuity and a NR is pretty clear. If all the textbooks were wrong about gas discharges, I think someone would have pointed it out in the literature. So I don't see that WP:RSs support your position. Unless some pretty convincing contrary sources appear, it is obvious what the article must say about neon lamps. --ChetvornoTALK 00:17, 18 January 2015 (UTC)
In my opinion, the wholesale removal of gas discharge material from this article was ill-considered. Even if Glrx is correct and these are not genuine negative resistance devices, given the widespread coverage in sources we should still be covering it in our article. At worst we should be pointing out that this view is an error. But I am far from convinced that it is an error. Doubtless there are some circuit cases where the neon is being used as a simple switch and it would not be correct to view it as a negative resistor, but not in every case. I am seeing a lot of hits in IEEE Xplore from heavyweight papers backing up the idea of negative resistance of neons. Just to pick one, this paper is coauthored by Leon Chua of memristor fame, not someone likely to make errors of circuit theory. Sadly, my subscription to IEEE Xplore has expired, but I have this snippet from google "...with large source resistance R (approximately 1 MO) is attached to a shunt connection of neon bulb (ALCO-type BNE4R (with current-limiting resistor removed)) (acting as a current-controlled negative resistance), and capacitor C, forming the basic relaxation oscillator...". We can get hold of this paper through WP:LIBRARY if it is wanted for sourcing the article. Glrx, you need sources of at least equal stature before you can say what you want to say in the article. SpinningSpark 11:10, 18 January 2015 (UTC)
The discussion at Talk:Neon lamp#Why the neon lamp is a negative resistor and how it behaves when voltage driven went nowhere.
The IV diagram description above has a lot of handwaving in it. Is the IV characteristic static? Apparently not, because capacitance is invoked when one has to explain instantaneous jumps off the IV curve. What's happening in the tube? The heavy ions (slowly) reorganize rather than instantly jump. With insufficient voltage, new positive ions are not generated fast enough and the old ones neutralize or crash into the cathode. The description also starts from an assumption of a negative resistance region. I don’t dispute that there are self-consistent negative resistance explanations. The NR explanation is not the typical one used in RS.
Instead, let the supply voltage be above the breakdown voltage. Choose R so (1) the current supplied at the breakdown voltage is larger than the Townsend current (so we can charge to greater than the breakdown voltage) and (2) the current supplied at the sustaining voltage is less than the sustaining current (so the glow discharge is stopped). Such an explanation is not tied to a static IV curve; it is tied to the physics of the tube.
The state description is (in a practical sense) equivalent to modeling that thorny/unstable transition region with a stable negative differential resistor. (Maybe that's why Kennedy/Chua did it.) If one assumes a static IV curve and imposes continuity and believes in the mean value theorem, then there has to be a negative resistance region. If you don't assume a static IV curve or continuity, then there is no compulsion.
In solid state devices, the massive atoms in the crystal are not moving around. Yes, some carriers are slower/less mobile than others and stored charge can stick around a long time — especially in a PIN diode's intrinsic region. Neon is 40,000 times more massive than the lowly electron.
I don't buy into the notion that the above talk-section sources are RS for the topic. They do not seem to be the best sources. From the &q arguments, they resulted from searches for references that held the negative resistance viewpoint; that is a biased search.
  1. (note graph 1.54, p. 63); Lasers by A. E. Siegman. The book is about lasers, and it has little detailed information about gaseous conduction. It is passing mention. Furthermore, the application is not exploiting but rather avoiding any negative resistance characteristic. I’m not sure how much I should make of the diagram; the negative slope is more pronounced than I would expect, but there could be a lot more going on in the diagram. The gas laser graph shows a 14 W light bulb; it is not a hyperbola; higher currents mean higher powers and higher temperatures. It would make sense in a laser application for the graph to be a final value (after temperature equalization) rather than a quasi-instantaneous characteristic. An IV curve for a transistor is usually done at a constant junction temperature. The result after thermal runaway is something different. (An arc discharge can be a thermal runaway.)
  2. University Physics by George Arfken; this is a basic physics text with no apparent authority in electron physics and no discussion of the neon lamp's physics; it is passing mention.
  3. Atom, Laser and Spectroscopy by S. N. Thakur, D. K. Rai; also passing mention. Furthermore, mentioning negative resistance here is not needed. One needs a high voltage to achieve breakdown; once the tube is conducting, it exhibits a low resistance, so a current limiting resistor is needed. We don't connect zener diodes to voltage sources.
  4. Handbook of Display Technology edited by Joseph A. Castellano; a handbook on display technology is hardly the place to get a definitive argument about gaseous conduction. It is also passing mention about negative resistance. The book does state the time to initiate the discharge (without priming) can take 100 μs. Further on, it makes the statement "The memory effect is realized by the fact that there are two stable states on the current–voltage characteristic." (By the way, GE's Glow Lamp Manual says it may take 300 μs to initiate a glow; up to 50 ms to deionize; and 6 μs to stabilize to a step change when conducting.)
  5. Vacuum Deposition onto Webs, Films and Foils by Charles Bishop. I'm not sure that a book on vacuum deposition is an appropriate venue to seek an expert on gaseous electronics, but the book does have a discussion of the physics. Figure 14.3 is promising (it may show carrier reorganization), but the linkage to the text is weak. The IV curve is static, there's no hysterisis, and the transition region is glossed over. The second paragraph of the transition region section seems confused about the cause. The arc discussion does point to a hotspot (drastic even destructive state change) going into other conduction modes.
  6. Industrial Plasma Engineering: Volume 2: Applications to ..., Volume 2 by J Reece Roth. Passing mention with difficult to interpret meaning. It also identifies a "mode transition" from an unstable arc to a stable thermal arc.
  7. Gaseous Electronics: Theory and Practice by Gorur Govinda Raju. Finally a source with a clearly significant title. The preview does not show figure 8.1, so I don't know B-C, but it sounds like the transition region. Section 8.2 is about conduction in a uniform field, but the transition region is all about the transition from the uniform field Townsend avalanche to the non-uniform field glow discharge. # Fundamentals of Light Sources and Lasers by Mark Csele. More passing mention.
A comment about passing mention. A serious secondary source will provide citations to primary sources.
It is not my burden to provide sources that say the negative resistance theory is discredited. To use the negative resistance claim, this article needs to cite to reliable sources that claim neon bulbs and gas discharge tubes are negative resistance devices. That view pops up in sources such as the above, but where does that view appear in reliable sources about applied electronics? Many reliable sources about physical electronics do not use the negative resistance explanation.
There are sources such as Reich that are RS about tube technology, use negative resistance as an explanation of tubes such as the dynatron but do not invoke negative resistance when discussing neon lamps or relaxation oscillators. Terman 1943 uses negative resistance for one flavor of magnetron (Magnetron Oscillators of the Negative Resistance or Dynatron Type, pp 527–528); Terman discusses a tetrode relaxation oscillator that uses "dynatron action" (so Terman is not afraid of using NR when all the carriers are fast electrons); Terman discusses a relaxation oscillator using a gas triode (thyratron) circuit (p 516) "commonly used to generate saw-tooth waves for sweep circuits of oscilloscopes"; that discussion does not mention negative resistance, but rather ionization that causes the grid "to lose control"; shut off happens when "there is not sufficient plate voltage to maintain ionization"; the "highest frequency obtainable in this way is limited by the deionization time of the gas in the tube" (ie, the slow carriers). Applied Electronics, McGraw Hill 1943, Chapter III is "Electrical Conduction through Vacuum, Gases, and Vapors". Its discussion of breakdown (cf 142) gives a carrier generation explanation of breakdown. It also covers glow discharges. A plot (fig. 13) on page 148 labels a breakdown, a transition region, and a flat-sloped normal glow region. Fig. 14 on the next page plots IV characteristics for 4 different glow-discharge tubes. The 2 neon sign tubes have a slight negative resistance over most of the graph; for example, one has 2 mA and 500 V going to 30 mA and 475 V (see heating possibility expressed above); the 2 W neon lamp near 0 mA is 70 V falling to and 0.25 mA and 55 V (this is in the first millimeter of a 100-mm wide graph; the graph does track the WP diagram above that has the negative resistance region outside the hysterisis portion). The book separates two classes of oscillators: negative resistance and feedback oscillators (p 597). It then offers another oscillator classification as sinusoidal or relaxation oscillators. Relaxation oscillators "are characterized by a sudden change, or relaxation, from one state of unstable equilibrium to another." These books are serious texts with footnotes to original sources.
I don’t know what Penning or Loeb say; I’m not near the library right now. Another vacuum tube book clearly states that an IV characteristic is not static, but I doubt I could locate that ref again.
See also
  • Petrović, Z. Lj.; Phelps, A. V. (January 1992), Optical and Electrical Characteristics of the Cathode Fall, Technical Report, Wright-Patterson Air Force Base, OH: Wright Laboratory, Air Force Systems Command, WL-TR-91-2094 , DTIC AD-A249 010, which show oscillatory instability in several operating regions of glow discharges. The source does state that "This region is inaccessible because of the negative differential resistance behavior of the discharge in which the use of too small a series resistance results in a jump of the current over the region." It is also not lost on me that oscillation could imply negative resistance, but in this instance the oscillation is probably the failure to settle into a stable carrier configuration. Sometimes one gets striations in gas discharges.
The Kennedy/Chua reference is dubious. The topic is about Van der Pol oscillators, and the abstract (I don't have the paper either) states:
Experimental confirmation has been made on a driven relaxation oscillator circuit, first presented by Van der Pol, of the period adding route to chaos. The nonlinear element in the circuit is a neon bulb, modeled by a three-segment piecewise-linear current-controlled resistor. A simple nonlinear circuit model has been used to reproduce in simulations the experimentally-observed period-adding phenomenon.
The paper is not about the physics of a neon oscillator but rather a simple model using a "three-segment piecewise-linear current-controlled resistor" for simulations. Adopting a negative resistance oscillator model says nothing about the actual physics.
Discharge tubes do not need to be discussed in this article.
Glrx (talk) 23:34, 20 January 2015 (UTC)
Ok, you've got a bunch of criticisms of sources that say neons have negative resistance, but do you have any sources that actually say the negative resistance viewpoint is wrong. SpinningSpark 00:08, 21 January 2015 (UTC)
Yes, it seems to me that's the first prerequisite. The fact that an author doesn't happen to mention negative resistance in his analysis of gas discharge does not mean he disagrees with the term - inferring that is WP:SYNTHESIS. --ChetvornoTALK 10:44, 21 January 2015 (UTC)
@Spinningspark: Demanding a source that says negative resistance viewpoint is incorrect is asking for evidence to prove a negative. The appropriate viewpoint for me is to look at how reliable sources treat the topic. If the reliable sources don't raise negative resistance, then WP shouldn't either. My criticism of the sources above is primarily their tangential / passing mention. Laser physics might be interesting, but it doesn't mean the author is a reliable source or authority for gaseous conduction.
@Chetvorno: Your comment about synthesis is misplaced. I'm pointing out that serious reliable sources are not invoking negative resistance.
Herbert J. Reich, Theory and Application of Electron Tubes, second edition, McGraw-Hill 1944. I'll recommend chapter 11, "Electrical Conduction in Gases"
Reich is the reliable secondary source WP seeks. He collates secondary sources.
Reich, fig 11-1, page 417:
As the voltage is increase, a value is reached, as at b, at which the current again begins to rise. If the electrode spacing is very small and the pressure sufficiently low, the current can be increased beyond b only by increase of voltage, and the characteristic is of the form shown by the dashed curve bm. With electode spacing and pressure used in most glow-discharge tubes, on the other hand, a current is reached at c, called the threshold current, at which the current begins to rise abruptly without further increase of voltage. The threshold current is still of the order of one or two microamperes. If the external circuit resistance is low, the voltage of the discharge remains practically constant, and the current jumps to a high value (milliamperes or even amperes), corresponding to n. If the external circuit is such as to prevent the current from rising abruptly, then the voltage drops abruptly to some lower value, as at d. The value of the current at d, and the path alogn which the change takes place, appear to depend almost entirely upon the external circuit. Experimental difficulties, such as the occurrence of relaxation oscillations, have so far prevented a complete study of the characteristic immediately beyond point c, but those experiments which have been performed appear to indicate that if the terminal voltage could be reduced rapidly enough by increase of circuit resistance or decrease of applied voltage, the current at point d would be the same as that at point c.
Reich, page 418, explains voltage drop with increasing current (arc discharge) as a temperature change at the cathode (a state change):
In the vicinity of g, however, the current is so high that, if it is maintained for an appreciable time, the cathode becomes hot enough to emit electrons. The thermionic emission reduces the voltage drop through the tube in a manner that will be explained in Sec. 11-14, causing further increase of current and greater emission.
Reich, page 418:
The exact predetermination of the behavior of a particular glow-discharge tube is difficult, if not impossible, because a given tube does not have a single current-voltage characteristic, but an infinite number of characteristics. The shape of the characteristic depends upon gas pressure, electrode temperature, and age of the tube; upon the amount of ionization remaining from previous discharges; upon the initial cathode emission, which varies with the cathode illumination; and upon the strength of other ionizing agents in the gas or container. A characteristic obtained with steady applied voltages, of which the curve of Fig. 11-1 is a typical example, is called a static characteristic. Characteristics obtained with varying voltages and currents are called dynamic characteristics.
Reich is not explicitly stating that "negative resistance" is wrong, but he is stating that a single static IV curve is folly, and he gives state change explanations for IV characteristics that others might label "negative resistance".
Reich explains that the gas discharge takes times to set up. Page 427, "11-7. Time Required for Ignition".
The field distribution changes. Reich p 429:
11-9. Breakdown.—As soon as appreciable curret flows, formation of space charge in the vicinity of the cathode causes almost the entire applied voltage, which was initially distributed uniformly over the whole cathode-to-anode distance, to become concentrated in th cathode dark space.
Reich page 431 explains how the space charge affects characteristics.
Reich page 433–434 describes dynamic characteristics he was involved with taking.
Reich page 435 has some nice numbered points about dynamic characteristics. For example, point 3:
3. When the current is decreased abruptly, it does not follow the static characteristic but, instead, a slightly curved path to the origin. This is explained if it is assumed that the time required for deionization greatly exceeds the time taken for the current to fall to zero. The increase in tube resistance accompanying deionization is evidenced by the decrease in slope of the characteristic as the current falls.
Reich page 439:
High-current-density glows may be maintained for time intervals that are too short to allow the cathode to heat to the temperature required for emission....
Glrx (talk) 03:35, 8 December 2015 (UTC)
In answer to your criticism that I am calling for sources "to prove a negative"; that would be valid if there were not already sources asserting the positive. But there are, so sources saying they are wrong are needed before Wikipedia can say they are wrong. The best you could say is that some sources do not invoke negative resistance in their analysis of gas discharge tubes. But why should they? Negative resistance is not a fundamental property of the physics of the tube. It is a property relevant to the analysis of an external circuit. That is why the Chua paper, from an expert in circuit analysis, is relevant here as a reliable source.
I am not seeing in your cites that Reich is saying "a single static IV curve is folly". Indeed, he seems to be saying just the opposite, that a static IV plot is possible. If he thinks it is impossible, why did he publish such a plot? What he does say is that there is more than a single dynamic IV plot because there are many dynamic variables. It is nevertheless still possible to make a plot for a given set of conditions. SpinningSpark 10:36, 8 December 2015 (UTC)
Glrx, re: "prove a negative"; citing a single 70 year old tube book while rejecting the 8 contemporary sources I gave above seems like pretty extreme WP:CHERRYPICKING. But the crucial point is that even your chosen source does not support your position. Reich does not say that gas discharge tubes do not have negative resistance - he simply fails to mention it. Spinningspark's suggestion that this is because "negative resistance" is an external circuit description which is not relevant to Reich's analysis of the physics sounds plausible to me. Quoting large amounts of text in an effort to demonstrate something that is not stated in the text is not only WP:SYNTHESIS, it is a very tenuous synthesis. The term "negative resistance" is very widely used in electrical engineering sources to describe the IV curve of gas discharge tubes, whether or not you accept these as WP:RSs. If experts believed this was a misuse of the term, someone, somewhere in the literature, would have said so. You have not cited a single source that says that. So I don't see that there is any support for your position. --ChetvornoTALK 13:15, 8 December 2015 (UTC)
Reich publishes a static IV curve, but he tells us that it is a static curve and that the dynamic story is different. He offers plots showing that the static IV characteristics don't apply. Compare to the IV characteristic of a transistor; the usual case (ignoring transit time) allows us to use the static IV characteristic as a model; dynamics are covered by some small signal linear capacitors. Gas tubes, especially ones where some authors will invoke negative resistance, are not following the static IV characteristic and are not using small-signal operation.
The negative resistance viewpoint of gaseous conduction is naive. I'm not saying WP should make that statement because sources don't make that assessment; they have no reason to make such a statement. I'm saying WP should be silent on the negative resistance explanation; if it does invoke negative resistance, then it should balance that with authors that do not invoke negative resistance. If we go to the sources that actually address the physics, they either offer explanations with clear mechanisms (e.g., transition to thermionic or field emission in an arc discharge) or claim the mechanism is not clear. They do not use a static IV curve and the mean value theorem to prove negative resistance. BTW, Reich's static IV is careful in a lot of areas, and includes discontinuities in the plot and axes.
In addition, there are many sources that describe dark current conduction with no appreciable space charge, the velocity differences between electrons and positive ions, and the glow discharge with its cathode fall. It takes time to reorganize the heavy positive ions.
Chua may be a circuit theory expert, but is he an expert on gaseous conduction? As stated above, the abstract suggests that the paper is about Van der Pol oscillators rather than gaseous conduction and that Chua isn't using an actual neon lamp but rather "a three-segment piecewise-linear current-controlled resistor". I don't have the paper, but does it delve into the neon lamp at all? The Van der Pol issue could just as easily be addressed with "a three-segment piecewise-linear current-controlled resistor" model of a tunnel diode -- something that none of us would dispute is a negative resistance device. Chua doesn't rise or fall on a bad model of a neon lamp. BTW, I'd be curious about where he set the break points -- the neon lamp breakdown is at a very low current. File:Doutnavka.svg suggests that breakdown is at 10 ma.;
I am confused by your comment that negative resistance is not a property of the tube but may be used as a model for circuit design. If negative resistance is not a propery of the tube, then why should it be invoked at all?
@Chetvorno: I don't know what to say. Gaseous conduction was a research topic in the early 20th century. Plasma physics for a few decades after that. Gaseous conduction is not a big research topic today because it has been mined out. Similarly, Newtonian physics is not a significant research topic today even if Feynman had some fun in the 1970s. The books that discussed gaseous conduction will be old. There's little reason for modern sources to address vacuum tube technology because vacuum tubes have mostly been supplanted by semiconductors. IIRC, you use a lot of old sources for old-technology radio detectors. Neon lamps are old technology. The physics hasn't changed, so the sources are still valid.
I don't see quoting Reich as WP:CHERRYPICKING. I am not selecting portions of Reich that favor me and ignoring sections that do not. Maybe I'm over-the-top with the static IV "folly" statement, but Reich is clearly pointing out that a neon lamp does not have a single IV curve.
As an editor, I believe I am entitled to comment on weight. Authors such as Reich and Terman have substantial weight.
You gave a list of references above, and I commented that the list was biased because it was the result of a search for negative resistance. Consequently, it would exclude references that did not mention negative resistance. Beyond that issue, most of the references were passing mention or only tangentially concerned with the topic; they were not surveys of the field. Reich is a survey of the field. The references in your list have little weight, and a lot of references with little weight do not add up to a lot of weight. Where the references had more weight, I took more time to address their comments. Laser tubes, for example, can reasonably present a static IV plot with a negative slope, but that does not mean they are negative resistance devices. I don't consider a negative-temp-coef thermistor a negative resistance device; it is a device that has more state than just current or voltage.
Your RS reasoning is backwards. If a source is not an RS, then WP does not care what it says. A published book does not automatically have the rank of RS for anything that it says. I've criticized several claimed RSs above; many titles were not on point; many comments were passing mention. I don't see a defense of those sources. More importantly, I don't see citations to sources that address the field of gaseous conduction and invoke negative resistance. The best source I've seen for negative resistance has been the GE manual, and even that book waffles about the transition region (Figure 1.1, page 1, EF; page 2, an "unstable region", "this region is often referred to as the negative resistance region").
My purpose in quoting long sections of Reich is to offer a serious source's explanation of the physics.
Glrx (talk) 19:33, 8 December 2015 (UTC)
In their paper, Kennedy and Chua are not just constructing a piecewise linear model in an entirely theoretical environment. They actually built a neon bulb oscillator and observed its behaviour. They were interested in the chaotic transitions of the oscillator between sub-frequencies. Their model accurately followed the experimental results,

Thus, in our simulation, we have accurately modeled teh dynamics of the system under investigation and have consequently succeeded in reproducing those period-adding phenomena present in the original circuit

— Kennedy and Chua, p. 977
If the negative resistance model is not a valid circuit model of the neon lamp, then this would be an extraordinary result. The coincidence would be unbelievable. I contend that it is not a coincidence. SpinningSpark 21:46, 8 December 2015 (UTC)
I'll have to look at the paper. Glrx (talk) 00:30, 11 December 2015 (UTC)
@Spinningspark and Chetvorno: Sorry for taking so much time to pull the ref.
Kennedy and Chua's CAS vol 33 issue 10, October 1986, pages 974–980, paper is irrelevant. They do not mention gaseous conduction at all. There is no discussion of the bulb's physics. KC are interested in "period-adding phenomena" rather than negative resistance. The only significance of neon lamp is that Van der Pol and other authors used a sinewave driven neon relaxation oscillator. The paper is muddled in another sense. The interesting issue is supposedly chaos (the "noise" that Van Der Pol skipped over), but KC developed a deterministic model so they could simulate something that does not have the nuisance of comsic-ray coin flips. Yes, there's subharmonic locking, but that's not chaos; analog TVs used that trick twice over.
The article claims:
It should be noted that a neon bulb is a very complicated device, yet we have been able to duplicate its dynamic behavior over the frequency range of interest (below 1 kHz) using the simplest possible model[10]. This confirms our belief that a neon bulb may be realistically modeled at low frequencies by a series connection of inductance (which we have indicated is an essential component of the dynamic model) and current-controlled nonlinear resistor.
The goal is model simplicity and not model accuracy. It is sufficient to duplicate dynamic behavior. The paper does not attempt to show that neon lamps are negative resistance devices. It only claims a negative resistance model will duplicate low frequency dynamic behavior.
The model is not related to the physics. "The neon bulb has been modeled by a current-controlled resistor with three-segment piecewise linear I-V characteristic, ...." The model is just by fiat; there's no explanation of why it is chosen. They do give a measured static characteric in figure 7. They also plot their model in figure 7, and the model just copies some gross characteristics. The exponential shapes have been thrown out. The goal is not accuracy, but rather making the simplest model. Also, there is no discussion about how the neon bulb measurements were made — something that has frustrated people who try to make such measurements. Even the GE manual raises the instability issue. Not a problem for KC: their VIC is stable and continuous from the beginning.
The article has some horrible nonsense wrt the inductor:
A parasitic inductance L, has been included in series with the current-controlled resistor in the neon bulb model to account for the element's dynamic behavior. (The circuit model is very sensitive to the value of L, (chosen to be 1 pH); too high a value of parasitic inductance reduces the widths of the transition regions between periods (n - l)T and nT.) This inductance is an essential part of the neon bulb dynamic model since its I-V characteristic is nonmonotonic and current-controlled. Note that the state equation of the circuit does not exist if L, is not included[6].
At first blush, adding some inductance could make physical sense because the neon bulb's carriers do not instantly vanish. But KC do not tie anything to physical phenomena. They just had to insert the inductor to the simulation work. Physically, the smallness and the sensitivity of the inductance destroys the validity of the model. A one-pH inductance would be swamped by real-world wire inductances. There's a lot more than 1 pH in series with a neon bulb. A rule of thumb is wire inductance is 10 nH per inch; the model inductance suggests a length of 0.0001 inches (2.5 µm). KC tell us the simulation doesn't work if the value is significantly larger. An NE2's pigtails will have a significantly larger inductance.
The inductor raises the spectre of Bob Pease's distruct of SPICE in reverse. Pease didn't like adding parasitic components to make SPICE converge; apparently KC have no qualms about adding such components. Yes, KC make some justification about the inductor being needed to increase the system order, but that does not make their model a real-world one. They really don't care about the real world.
As to the "extraordinary result" comment, KC admit in the paper that a hysterisis model achieves the same result:
Newcomb [ll] has observed chaos in an apparently second-order autonomous circuit, where the nonlinear element is described not by a single-valued function, but rather by one which exhibits hysteresis. Such a hysteretic component is in fact a dynamic element requiring at least one additional state variable before the circuit can be modeled by a system of ordinary differential equations[12].
The paper is irrelevant wrt negative resistance and gaseous conduction. Gaseous conduction is not Chua's field, and the article is an unreliable source for our purposes.
Glrx (talk) 22:37, 8 January 2016 (UTC)

Negative resistance used for measurement[edit]

I have removed this from the article,

Also, a multiport oscillators, based on the negative resistance, becoming widely used for the precision measurements. They utilize a low- to a middle- frequency range (30 kHz - 30 MHz), and use multi-transistor circuits with the negative resistance instead of the "pure" negatrons. They called "frequency components" and there are a lot of different measurement devices based on this method (for example, to measure electrical parameters, environment parameters, geometrical sizes and so on).[1]


  1. ^ Krynochkin, R; V (2015). "Frequency components with negative resistance for intellectual measurement systems". PeerJ PrePrints. 3. doi:10.7287/peerj.preprints.1506v1. Retrieved 14 November 2015. 

I can't make head nor tail of it, and the (unpublished) paper it refers to is just as obscure. SpinningSpark 15:36, 27 November 2015 (UTC)

1. I was made that additional for the next reasons: a) this WP article is really greate! It looks better than some papers about the negative resistance. But I noticed something missed in it and feel that can add some important info; b) negative resistance is in the area of my scientific interests c) measurements is also in it. So, what is the thing that I feel was missed? It's a using of a negatron-based devices in the measurement of the environmental/circuit parameters. It's a few scientific scholars in the world that had a deal with this subject (I know 2: in Ukraine and in Russian).
2. That preprint is a short review/classification of this question so I referenced to it. If you think it will better, I'll add some additional references about this (I've found more than 50 papers about this on Google Scholar and trust that it's even much more of it). Jettec1 (talk) 16:04, 27 November 2015 (UTC)
The English is completely mangled in that addition. If I understood it, I might try to clean it up, but I don't - much of the terminology is obscure. For a start, what is meant by a multiport oscillator and how does that relate to negative resistance or measurement? And what on earth is a frequency component. You seem to be saying that's what a negative resistance multiport oscillator is called, but that doesn't make it any clearer. @Chetvorno: do you understand this? SpinningSpark 16:43, 27 November 2015 (UTC)
Ok, it's true - my English is really bad. Sorry. So seems your help will be very useful. Well: 1. As "multiport oscillator" I meant a device that can be described in the form of the four-terminal network ( instead of the "pure" negatrons (as a Gann diode) that seems to be a two-terminal. Please, refer for example to a papers (they are in Ukrainian peer-review journals, but you can see circuits with several transistors, that has a negative resistance on its external ports) the first, the second. 2. "Frequency component" is a collective name for current and prospective devices, where the negative resistance effect used for measuring/processing signals (for example we can build an RF filter or synthesiser based on the negative resistance effect). Hope this clarify something. Thank you for your help! Jettec1 (talk) 17:37, 27 November 2015 (UTC)
I'm not seeing any transistors in the first link. There are symbols with two chevrons in a circle that I don't recognise, but they can't be transistors as they are two-terminal. I don't want to put words in your mouth, but are you talking about an oscillator where the oscillator output is controlled by another port? Like a Voltage-controlled oscillator? Does frequency component mean "frequency-dependent component"? SpinningSpark 18:31, 27 November 2015 (UTC)
1. Ok, here I've made a screenshot for your convenience screen_link (It's from the previous link, but need to scroll one page up, I apologize). Two chevrons in a circle it's a (current source). 2. "...oscillator output is controlled by another port..." Yes and no. Yes - the oscillator output frequency is controlled by another port; No - because there are circuits that can be controlled by several another ports (so it may be a many port device (multipole)) 3. No, I mean a component, based on a negative resistance, that can change it output frequency according to a measurement parameter. Jettec1 (talk) 20:45, 27 November 2015 (UTC)
@Chetvorno, Glrx, and Catslash: Pinging more users for opinions. SpinningSpark 09:36, 28 November 2015 (UTC)
I don't understand what is going on, but WP:UNDUE is the hurdle here. As editors, it is not our function to decide what is important. No matter how good or important the work is, it needs to have more visibility before it goes into WP. "It's a few scientific scholars in the world that had a deal with this subject (I know 2: in Ukraine and in Russian)." That suggests the viewpoint is not widely held.
I don't get much looking at the sources.
  1. The PeerJ Comp Sci top line is "NOT PEER-REVIEWED. This is a rapid communication before peer review".
  2. The page 115 link would not work for me; the connection would time out.
  3. Технічні науки seems to be about the value of using simulators and modeling in circuit design. The circuit is a one-port oscillator that is intent on measuring the capacitance Cw of a capacitive sensor. The goal appears to use capacitance to imply the "thickness" / measure distance. For some reason, either the capacitance or circuitry (or maybe both) is nonlinear:
    Modeling capacitive transducers to determine the thickness
    A well-known method of creating models and calculation using frequency converters application of Kirchhoff's equations for the equivalent capacity of the circuit, followed by substitution Thomson in the formula for finding frequency generation. However, this approach has some shortcomings, in particular inability to simulate and observe the shape obtained and calculate nonlinear oscillations distortions associated with it, the complexity of modeling bahatoharmonikovyh generators and more.
    (BTW, "frequency converter" apparently means capacitance-to-frequency converter / aka oscillator)
    The article's focus is about using symbolic algebra in nonlinear modeling. It discusses edge effects of the parallel plate capacitor to point out that the capacitance is not simply the parallel plates dominating the fringing capacitance.
  4. Evernote would not load for me.
My sense is negative resistance is a minor issue here; it's used to make an oscillator so the capacitance and thickness can be determined. It makes sense to be a one-port oscillator, and that leads naturally to a negative resistance viewpoint. There's talk of the time domain and frequency components, but I don't get a clear sense why; the oscillator is LC; compare RC oscillator which would be close to linear with C in period. In any event, the goal is high accuracy measurements that account for sensor or other nonlinearites.
Glrx (talk) 16:58, 28 November 2015 (UTC)
Yes you are right about nearly all (especially about frequency convertor), but the next is wrong: "My sense is negative resistance is a minor issue here..." No, negative resistance is a thing that makes this method work. You are right that we have LC oscillator, but, in fact, C = Cvt + Cnr, where Cvt - is an equivalent capacity of the transistors junctions, Cnr - is a capacitive reactance of the negative resistance (impedance). Why it is important? Because Cnr is really changes great according to the voltage on the input ports. So, as the result, we obtain a VERY sensitive device (for particular I was obtaining ~ 60 kHz/pF on a 1 MHz center frequency). It gives a great possibility to use it in measurements of the small quantities. And we can, of course, measure a lot of different values: pressure, thickness, temperature and so on.
From the other side, as far as Cnr changes great according to a port voltage, ve can build an effective tune RF filters, synthesizers and so on. I was planning to add this information later.But now I'm completely not sure: if you think that this information is not good for your article, maybe it will better to create a separate article, or maybe just don't touch on this question. Waiting for you opinion. Thanks.
Jettec1 (talk) 19:59, 28 November 2015 (UTC)
Sorry, this is outside my area of expertise. However, for inclusion in the article, the paragraph should give a clear explanation of the principle of operation and some indication as to why negative resistance is central to it. --catslash (talk) 16:21, 29 November 2015 (UTC)
I have to agree with Glrx and Spinningspark. I also found the Krynochkin paper incomprehensible, and I wasn't able to access the other two papers you linked: [9], [10]. Capacitance-to-frequency or other sensors based on negative resistance oscillators (if that is the subject of the papers) sounds like an interesting application, but without any sources I can read there is no way to determine if it is WP:NOTABLE enough to include. It seems only tangentially related to negative resistance, maybe better articles for this content would be Voltage controlled oscillator or Signal conditioning. But to include it on Wikipedia at all, better sources are needed, to meet both WP:VERIFIABILITY and WP:NOTABILITY. As Glrx said, an unpublished research paper on a preprint archive is not a WP:RS. And research papers are not really enough; primary sources should be backed up by secondary sources (WP:PSTS). Does this subject appear in any textbooks or survey articles, Jettec1? --ChetvornoTALK 19:19, 29 November 2015 (UTC)
Chetvorno, thank you for you opinion. As far as I see, I've made a conclusion from our conversation, that this info (sensors based on negative resistance oscillators) is not widely famous on the English-speaking part of the scientific society. But I completely sure, that there are a lot of papers/research concern this, in Russian and Ukrainian languages. BTW, I've found an international organization dealt with this: International Frequency Sensor Association. They also publish a journal Sensors & Transducers Journal, in wich "...the Sensors & Transducers journal significantly contributes in areas, which are not adequately addressed in other journals, namely: frequency (period), duty-cycle, time-interval, PWM, phase-shift, pulse number output sensors and transducers...". About the survey articles, that I know - I know only that preprint and several books, but all of them are non-English. Jettec1 (talk) 21:33, 29 November 2015 (UTC)
Also, want to ask all of you (@Chetvorno, Glrx, and Spinningspark:), about smthng. I have planned: a) collect all English-written info about this question, that I have; b) place it to the Dropbox folder; c) shared it for you (because, you reported, that you cannot open my links - I don't know why). So, I want to know: will this plan helpfull to resolve your doubts and will you help write an appropriate info to improve this WP article? Because, it was really glad to hear your opinions, but I'm no the kind of the guy who wants just add something that he know, in every place where he is. I just want to hear a clear answer (if it's possible): if adding info about sensors based on negative resistance oscillators will improve this article - I will work, if not - (useless/not interesting/doesn't fit some rules) I will give up. Thanks. Jettec1 (talk) 22:05, 29 November 2015 (UTC)
I don't have a problem opening your links, but a drop box is likely to be helpful to those who can't, especially the circuit diagrams. You won't get a clearer answer than the one you got from Chetvorno on the "rules" you need to meet: WP:V and WP:N. If your sources meet them then there is no problem with including the material. It does not matter that the source material is in Russian, that is perfectly ok. But the article needs to be in English, and the problem we are all having is understanding it. What I have got so far is that these devices consist of a VCO that works by the input voltage altering the negative resistance at the input to produce a large change in frequency at the output. What would be really helpful here is a circuit diagram of a real application with a clear description in English of what it is measuring and how it works. Figure 1 in the Технічні науки article would probably make a good example, it seems particulary simple. By the way, is your term frequency component a translation of частотним виходом? If so, I suspect that that is not a good translation, I cannot find виходом in any online Russian dictionary so it is improbable that the meaning is something as simple as component. SpinningSpark 14:01, 30 November 2015 (UTC)
Misspelling of выход (output) perhaps? --catslash (talk) 15:12, 30 November 2015 (UTC)
виходом is the instrumental of вихід which is Ukrainian for выход (I reckon) - i.e. it's saying that the output of the sensor is a frequency. --catslash (talk) 15:29, 30 November 2015 (UTC)
And here is a Ukrainian patent for a Сенсор магнітного поля Magnetic field sensor з частотним виходом with frequency output. --catslash (talk) 15:40, 30 November 2015 (UTC)
Ahh! it's Ukrainian, that probably means I have mortally offended someone by calling it Russian. Apologies. Now we have the right language, Google translate tells me the paper was published by Khmelnytskyi National University. So, for what the OP calls a "frequency component", we would say "frequency output device" or somesuch, no? This is beginning to make sense now. SpinningSpark 16:17, 30 November 2015 (UTC)

I propose to put this back in the article as follows. Any comment before I do? SpinningSpark 11:52, 8 December 2015 (UTC)

The negative resistance property of a feedback oscillator can be used to construct a very sensitive measuring instrument with a frequency output. In this technique the negative resistance is varied at the input of a voltage controlled oscillator causing the output frequency to change in response. These devices work within the range 30 kHz to 30 MHz. Many parameters, both electrical and mechanical, can be measured by this method.[1] For instance, a flowmeter has been constructed with a sensitivity of 500-1300 Hz per litre/hour[2]


  1. ^ Krynochkin, R; V (2015). "Frequency components with negative resistance for intellectual measurement systems". PeerJ PrePrints. 3. doi:10.7287/peerj.preprints.1506v1. Retrieved 14 November 2015. 
  2. ^ V.S. Osadchyk, A.V. Osadchyk, Y.A. Yushchenko, A.A. Yaroslavl, "МІКРОЕЛЕКТРОННИЙ ВИТРАТОМІР З ЧАСТОТНИМ ВИХОДОМ" ("Microelectronic flowmeter with frequency output"), ВІСНИК (Herald), vol. 6, pp. 113-116, Khmelnitsky National University, 2008.(Ukrainian)
SpinningSpark, sounds great. I apologize, that there is not too much help from my side - it's the end of the semester in the University and I have a lot of the work with my students. But I hope that I'll have a more time after the Winter holidays, and may return to this article. Jettec1 (talk) 14:56, 8 December 2015 (UTC)
  • Oppose addition. That is not how it works. The papers are not using the contorted Negative resistance#Feedback oscillators can be viewed as having a negative resistance, the papers are starting with an actual negative resistance. One can make an oscillator with a negative resistance and a tank. If you know L, then you can compute C from f. The basic notion of measuring C is simple. If an instrument is must use a single port for the measurement, then the designer is driven to the negative resistance model. All one needs is enough negative R to counteract Rloss.
The device is not varying the negative resistance to cause a change in the response. Ideally, everything is linear, so f is invariant to the amount of negative R. The real world has nonlinearities, so a linear model does not offer a good explanation of the oscillator's frequency or waveshape. Better (nonlinear) models would give a better correlation to actual C.
The varying is not an intentional feature of the instrument. In fact, it seems to be a weakness in the measurement technique. The classic approach (early 1930s electronics) would be not to model the nonlinearities, but rather to reduce them until a linear model is adequate. The waveshapes are overdriven; amplitude control (an intentional varying of the negative R) could keep things more linear.
They are not secondary sources.
There is still a matter of WP:DUE for these papers. I don't get the sense that the cited papers are a typical or even common approach to measuring capacitance. Laboratory instruments usually measure capacitance with a bridge. The bridge has its own low-distortion oscillator; null measurements are used, so the measurement is sensitive and use relatively small measurement amplitudes. There are instruments that measure C by incorporating it into an integrator or RC oscillators. Still others use LC oscillators. IIRC, some high-dielectric fluid level meters and some moisture sensors use oscillators to measure capacitance.
There is something deeper going on here, but I don't recall the papers coming out and saying it. I believe the capacitive sensors are physically very small, so the measuring port needs to be small and nearby. While that is good for the research issue, it takes its relevance further away from this article. That belief is tempered by the low frequencies; 30 MHz is 10 meters, so there are few transmission line effects such as those that drive the use negative resistance oscillators at microwave frequencies.
We just don't know enough about the approach.
Glrx (talk) 17:09, 8 December 2015 (UTC)
> That is not how it works...
1. Thank you for you opinion, but seems that I have to disagree with you. Yes, every oscillator should have negative resistance (either by nature, either by schematic) and so? If we use the schematic analogs of the negatrons we obtain the way to impact this negative resistance value by the measurement parameter (C, L or R). And so, increase the change of the generated frequency. And so, increase the sensitivity of the measurement device.
> The device is not varying the negative resistance to cause a change in the response....
2. Yes, it does. When the measurement parameter is changed - the equivalent C (in the oscillator LC circuit) is also changed. This is because:
- the great part of the equivalent C is a 'Cnr'-capacitive reactance of the negative resistance
- the capacitance of the transistor junctions has a dependence from the supply voltage of the transistors and this supply voltage was changed
> They are not secondary sources.
I'm unfamiliar with the Wiki source's classification (yes I read it, but feel that I don't sure how this system works). So, seems we need additional opinions about either they are secondary sources or not.
> I don't get the sense that the cited papers are a typical or even common approach to measuring capacitance.
Yes, you are right: you cannot see it in every lab. But it's a lot of different real devices that use this approach and papers describing it.
> We just don't know enough about the approach.
Is this a reason to not include this approach to the Wiki? And also note: we do not talk about the capacitance measurement itself, but about using a negative resistance-driven oscillators in measurements.
> That belief is tempered by the low frequencies; 30 MHz is 10 meters, so there are few transmission line effects such as those that drive the use negative resistance oscillators at microwave frequencies.
Don't forget that we use a schematics 'analogs' of the "nature" negative resistance devices. So, it's no problem with the low frequencies. While "nature" negatrons oscillator has an internal feedback, parameters of which determined during it production (semiconductor material, size). In the schematics 'analogs' we can easily adjust the central frequency (and the negative resistance value) by the supply voltage, or by changing the schematic connections, or by changing the transistors. I've seen 'negatron analogs' devices working on the sub-GHz. Jettec1 (talk) 19:29, 8 December 2015 (UTC)