No, that's a completely wrong picture. The electron cloud/space charge is, on the physical scale that vacuum tubes are made, is centred at considerable distance from the cathode surface. If it were not so, then the tube screen or anode mu would be a LOT less than it actually is. You can look at it this way: Mu (voltage gain with const current load) arises from the ratio of the anode distance from the vitual cathode to the distance from the control grid to the virtual cathode.
The fact is, neglecting the other factors I mentioned (which increase the ratio), the cathode emits around 2.5 times the rated anode(+screen) current, regardless of whether the anode and/or screen draw current or not. The excess electrons are pushed back by the space charge and re-neter the cathode some time after they left it. It is no good just repeatedly saying they don't spend time in the cloud (cf in the queue) - the fact is they do. Some leave the cathode near vertically, for them its a certain time "up" and "down". Some leave at a shallow angle - for then the flight time is longer. Steep or shallow angle, each leaves at a random velocity. Fast electrons emited on a shallow angle havean even longer flight time.
The process of thermionic emission is much the same as a liquid partly filling a sealed container from which everything else has been evacuated. At any non-zero absolute temperature molecules continually leave the liquid phase and become part of the gas phase. Molecules continually leave the gas phase and rejoin the liquid phase. If you increase the pressure a the rate at which molecules return to teh liquid phase goes up. It is much the same because it is really the same process - heating the cathode adds to the kinetic energy to the electrons in the 'electron gas' (old term valence electrons) in the metal so a faction of them have enough energy to cross the surface barrier into the vacuum. That's why the emitted lectrons have random direction and random velocity about a mean - it is a continuation of the random veclocity and direction they had as brownian motion within the metal.
It you have genuinely done a numeric calculation, then post it. Text discussion is subject to doubts over just what you mean. With a numeric claculation or an algebraic manipulation, it stands on its own feet. You either did a valid calc or you didn't.
If you don't post your calcs, we must assume you can't. There's no excuse.
Last edited:
Hmm, that's not a fact in my book - and plainly doesn't stand scrutiny unfortunately. An obvious and obvservable contradiction is that flicker noise shows up directly in Ik - yet flicker noise is a documented cathode emission variation phenomenum........? If there were an electron emission buffer 'cloud', 'queue', 'bucket' or whatever, flicker variation in emission would not show up in Ik, but it does. Plus there's fundamental reasons to doubt, some of which are set out in the early chapters of the 1951 paper Merlinb linked, BTW.
And if I understand it, you would really like me to set out arithmetic of a calculation which you think are based on flawed physics principles...... ??? 😉
Sorry if I am over-simplifying it, but if the number of thermionic electrons emitted per second is Ne and the average 'flight time' of an emitted electron is T. Then when there is no current to the grid, screen or anode, the average number of electrons in the cloud (or queue) must be Ne multiplied by T.
(Neglecting any electrons used up in neutralising ions, or decaying into a photon and a neutrino!)
I am following this thread with interest, but if the above is not what you are disagreeing about then please excuse my interruption and carry on! Or can you explain the part of the above where you differ?
(Neglecting any electrons used up in neutralising ions, or decaying into a photon and a neutrino!)
I am following this thread with interest, but if the above is not what you are disagreeing about then please excuse my interruption and carry on! Or can you explain the part of the above where you differ?
You need to read a few more tube books. The fact is, flicker noise is a very low level noise compared to shot noise, and it can only be detected if at all below a couple of hundred hertz under the most favourable circumstances.
Flicker noise only occurs in oxide coated filaments and arises in the transport of electrons in the oxide grain interfaces. It does not occur in tubes having pure metal filaments.
But shot noise, arising from the random absorption of electrons into the anode, is realily measurable at very high frequencies, right up to the point where stray capacitance shunts it and transit time effects occur - flat to hundreds of MHz.
You obviously haven't heard of what are called "noise diodes" or "temperature limitted diodes". Or you think I haven't. A typical example is the A2087.
These diodes have long been used as a self-calibrating source of white noise good to hundreds of MHz, for measuring the noise figure of low noise input stages of preamps and radio receivers. The reason they are used is they are made with a pure tungsten filament/cathode. Because there is no oxide cathode coating, the cathode cannot be damaged by operating in the temperature limitted region where the anode voltage is more than sufficient to sweep all electrons emitted into the. Unlike operation with a normal tube, the anode cuurent is sensibly constant regardless of the operating voltage.
The anode current is instead controlled by adjusting the filament current. The noise generated at the anode is then pure shot noise that is directly proportional to the square root of the anode current. The noise power is independent of the anode voltage provide the anode voltage is above the minimum value required to draw off all the cathode can emit.
No flicker noise is generated in pure metal cathode tubes, regardless of whether the anode voltage is sufficient for temperature limitted operation or not. Flicker noise shows up in the anode current of oxide cathode tubes for the same reason any voltage source in series with the cathode will.
It is well known that when these noise diode tubes, and any tube for that matter, is operated progressively below the anode voltage required for temperature limitted operation, the noise power progressively falls away below the value that would correspond to the same anode current if it was temperature limitted.
This well known phenomenon is one of two main reasons why it has been known since the very earliest years of tube manufacture that there is a dense electron cloud centred some distance out from the cathode - the other being that mu is much higher than it would be if there was no electron cloud.
The noise power drops away at anode currents below the temperature limitted value precisely because the space charge acts to smooth out the random arrival of electrons in the electron cloud, and their departure for the anode. Just like the movement of customers after checking out their purchases is a lot less random than their arrival at the checkout, IF the supply of customers is greater than the checkout processing rate.
Go look up "noise diode", "pure tungsten filament", and "measure noise figure" Or look up the datasheets for noise diodes eg A2087, K81A.
Last edited:
You are correct for the case of no anode or screen current - Ne x T
Its difficult to know exactly what Lucky thinks is how it works, as he often writes imprecisely and does not confirm my interpretations of what I think he is trying to say. But it appears he thinks that either Ne is equal to the number required for Ik or that for some reason there is no electron cloud.
One suspects he is just stirring, but I've given him the benefit of the doubt by assuming he is actually trying to understand it.
I've set out that equating Ne and 2.5x Ik is just a vehicle to show an apparent contradiction in either the 2.5x rule or an apparent defect in likely density required to provide cathode protection from ion bombardment. I genuinely question whether an electron cloud exists per se, and if so whether it can be dense enough to afford the cathode protection from ion bombardment as per common wisdom. As already said I have an open mind but am really looking for reasons of substance one way or the other. Actually the 1951 paper Merlinb linked offers good insight, discusses emission and cathode operation in great depth and does so without mention of electron clouds nor ion bombardment per se at all AFAIK..........
How I think emission might work I already set out, that in grid valves AFAIK there is a retarding field near the cathode due to grid potential which effectively means that only electrons destined to become Ik actually leave the cathode surface in any meaningful way. In a diode, only those electrons destined for the anode leave the surface (or within a very close distance of it say 10-6 m) because of potential gradient. As already said, happy to be wrong and only interested to explore what might really be going on.
Last edited:
Although flicker noise does indeed exist on the anode current -often dominating shot noise in the audio band- the space charge still has a reducing effect on the noise (both shot and flicker). No space charge could 'suppress' the noise entirely, but thanks to the presence of the space charge the noise is less than for a saturated cathode. I like to think of the space charge smoothing factor as a reservoir of water with an almighty water wheel at one end generating noise, but only smaller waves are to be found on the far shore, though certainly not placid waters!
Last edited:
Merlimb has given three different references. The 1951 refrence is not a paper, its a book. Which of Merlimb's links do you mean?
No matter, none of them say what you claim they say.
You cannot go around claiming there is no dense electron cloud. The RCA internal training books state that there is a cloud, and it protects the cathode by dissipating ion kinetic energy, just as DF96 said. Are you saying that leading RCA engineers who spent their professional lives in tube R&D didn't know what they were doing?
In summary, here's the three main things that evidence that there is an electron cloud:-
1) The tube mu of triodes is much greater than the physical griod cathode spacing would suggest.
You need to put up a coherent explanation why this is so.
2) Shot noise noise, which arises from the random arrival and absorption of electrons at the anode, is proportional to the square root of the anode current, and independent of anode voltage, when the tube is operated with the anode above the saturation voltgae, but progessively falls further below this proportion as anode voltage is reduced below the point where it allows the space charge to form.
You need to put up a coherent alternative explanation of why this is so.
3) The retarding field current is sloped when plotted against the logarithm of anode voltage. This shows that electron density progressively falls off the further you get from the cathode.
You need to put up a coherent explanation for this.
4) The retarding field current increases the closer the anode is to the cathode, until it is so close it's within the electron cloud mean distance from the cathode. This demostrates that there is an electron cloud, and vacuum tubes with very small cathode/anode spacing have been used for unattended power generation.
You need to explain this.
As Keit has explained, the existence of the electron cloud around the cathode cannot reasonably be questioned (under normal valve conditions). Anyone who doubts it has simply to investigate the transition from normal operation to temperature-limited operation (as in a noise diode) - or read a textbook. The issue is whether this cloud has sufficient density to protect the cathode from ion bombardment. The fact that many noise diodes only have a few hundred hours life, whereas a normal valve lasts for many many thousands of hours, says to me that the cloud is protective. I would like to see a rough calculation to confirm this, but so far I have not been able to come up with one myself. An 'order of magnitude' estimate would do.luckthedog said:
Hmmm......
1. mu in triodes should be (fairly) independent of grid-cathode spacing (Maxwell), but I guess you might mean triode models require adjustment in terms of 'virtual' cathode potential and 'virtual' position? (Langmuir, Dowley etc). I think such adjustment is mathematically similar to altered potential barrier at the cathode surface, and adjusted initial velocity - ie doesn't necessarily require a physical electron cloud per se, so don't think it tells us either way. Either model could be made to fit I think.
In any event, suppose there really were permanent thermionic cloud of charge carriers near the cathode. 'Virtual cathode' location and potential would depend critically upon thermionic emission then, of course. Which depends massively upon temperature of cathode, and between samples of valves of a type for example. Then I venture valve parameters/characteristics would vary significantly with heater, but they don't.............
2. Yes, shot noise should not vary with Va, other than via how Ia depends on Va. However, shot noise discussion in valves often seems confused IME, often ending up with a kT term which always rings alarm bells of confusion with thermal (J-N) noise. J-N noise does vary with Va, of course as (non-saturated)valve effective internal resistances depend on Va........so devil's in the careful detail and scutiny methinks. I have a philosophical issue with how shot noise as seen by the anode could be anything other than classically random - entropy applies to Ia charge carriers no matter where they came from............so would appreciate a reference or two to check out further since if valid this might be convincing IMO. This point offers the best 'shot' at being convincing I reckon 😉
3. There seems two variables mixed up here: Va and distance between a & k......so some logic seems missing to connect the two sentences.....? Yes, space charge is not uniformly distributed and depends on Va and distance, and yes carriers (ie space charge) interact.....but that does not say anything about carrier concentration near the cathode per se.......?
4. Presumably you mean a connected plate brought increasingly near a thermionically emitting cathode collects more electrons until at some point it stops increasing ? That's effectively saturation, all charge carriers collected. The issue is how emission behaves in presence of an external retarding field to the point where emission might be supressed.
The 1951 paper I refer to as linked by Merlin is indeed the one published in 1951, and yes it is a book not a paper 😉 And it is well worth a thorough read, IMO.
I think its reasonable to question both, not quite heresy! I prefer to verify existence first, otherwise I think there's more than one way to calculate and crosscheck, if it hasn't been done already.
Member
Joined 2009
Paid Member
Interesting option - is this sufficient to address all the common modes of degradation ?
Member
Joined 2009
Paid Member
There is no way that a cloud of electrons can be said to physically restrain an energetic ion or atom through individual (multiple or otherwise) collisions - the mass difference is too large for meaningful energy transfer. It's a Red Herring.
Also, I don't think of electrons as points as there is no physical basis for thinking of collisions between electrons and other electrons/ions/atoms as a collision between solid objects. The collisions are an interaction of their electrostatic fields (that surround electrons and other charged particles such as protons in the nuclei of atoms and ions); the effective size of the particles involved in such collisions are therefore much larger than their so-called physical sizes.
I expect that the comments made by experts long ago about the benefits of the electron cloud do not refer to physical collisions between particles but to the existence of the electric fields the cloud generates. This space charge significantly changes the electric fields around the cathode. Ions approaching the cathode are influenced by these electric fields not by direct interaction with a very sparse collection of electrons. Are these fields strong enough to protect the cathode ?
I would also imagine that a neutralized energetic ion is going to smash into the cathode - nothing to stop it.
Also, I don't think of electrons as points as there is no physical basis for thinking of collisions between electrons and other electrons/ions/atoms as a collision between solid objects. The collisions are an interaction of their electrostatic fields (that surround electrons and other charged particles such as protons in the nuclei of atoms and ions); the effective size of the particles involved in such collisions are therefore much larger than their so-called physical sizes.
I expect that the comments made by experts long ago about the benefits of the electron cloud do not refer to physical collisions between particles but to the existence of the electric fields the cloud generates. This space charge significantly changes the electric fields around the cathode. Ions approaching the cathode are influenced by these electric fields not by direct interaction with a very sparse collection of electrons. Are these fields strong enough to protect the cathode ?
I would also imagine that a neutralized energetic ion is going to smash into the cathode - nothing to stop it.
Last edited:
Your 1st para is just nonsense.
In your 2nd para, you have confused the mean position of electrons with the number of electrons. Or, if you like, it's the same thing, you are confusing position of space charge with intensity of space charge.
Mu does vary with with heater voltage, but not strongly. There are several reasons for this:-
1) Tube heaters are self regulating, as heater resistance is approx proportional to temperature; thermal radiation from the cathode is proportional to the 4th power of temperature, and, tubes are operated in the space charge limitted region, since there is a dense cloud of electrons, changing the cloud exchange rate diesn't shift it much.
You can picture it this way, seeing that the flight of each electron not captured by the anode is sort of parabolic. In parabolic flight, velocity is max at take off and landing, zero just at the peak altitude. This measn most of the flight time is spent near the maximum altitude. Image ten 10 year old schoolboys standing in a 200 m2 area continually throwing equally weighted small rocks in the air - you get a "cloud" of rocks centred on the height that they are strong enough to throw them. Now imagine 100 schoolboys doing the same thing in the same area. The height of the rock "cloud" doesn't change, but the density does. Assuming not too many heads get cracked....
When you increase the temperaure of a cathode, the number of electrons changes dramatically. A 1% change in absolute temperature at typical cathode temperature changes emission quantity by about 12%. But the mean electron launch velocity changes only 0.5%.
Incidentally, as new electrons continually leave cathode, others continually return, and the electrons are accellerated by any applied electic field, the electron cloud not only acts as one plate of a capacitor, and the grid the other, the capacitance so formed appears to have a resistance in series with it. The capacitance increases with heater voltage, considerably at first, then slighty as cathode design temperaure is approached. The resistance decreases with heater voltage, and the change is quite marked. The capacitance and series reistance is easily measured by the Q-meter method, though not with any great accuracy.
This electron cloud impedance, a small capacitance in series with a resistance, appears in parallel with the static grid cathode capacitance (the capacitance measured with heater unenergised) Thus with the standard equation for the capacitance between two plates, one can in theory calculate the position of the electron cloud. Unfortunately it has little practical value as the cloud capacitance is typically smallish compared to the measured static value (which inevitably is not just the true grid-cathode capacitance, but includes the capacitance of the tube pins, the socket, and wiring strays), and the effective series resistance is quite high. Therefore the cloud centre calculated in this way is inaccurate. Calculating it from triode mu is much better.
Further evidence of the electron cloud, and evidence that the position changes only slightly with cathode temperaure, but its density change is quite marked.
The standard expression for shot noise in vacuum tubes when operated in temperature limitted conditions is:
in = (2 q I df)^0.5
where in is the RMS noise current amplitude; q is the electron charge, dF is the measurement bandwidth.
There is no kT term in it.
See any good book on measuring noise factor or look in the manual for any tube based noise generator.
Glad you agree there is a non-uniform space charge then.
That's exactly what I mean. In theory, (for tubes with a fixed anode position) the retarding field current decreases with increasing retarding field (negative anode voltage) and does not reach zero until infinite anode voltage is applied.
In practice with typical receiving tubes, the retarding field current is so low with only a few volts applied, its difficult to measure, and in any case leakage across the micas by then swamps the retarding field current.
However, ingenious methods had been devised to measure it (especially the method of Suketoshi Ikehara), and plotting it on a graph vs log anode voltage shows a straight line slope as I said before.
You had best read it it then. Read it carefully and with pencil and paper work diligently through all the math. Can you do that?
Last edited:
I suspect from literature that ion bombardment was historically a known hazard to the cathode in operation, mitigated by minimising the number of ions and their energy ie keeping vaccum good, impurities low and limiting Va. Perhaps there emerged some advice about re-heating getter material (inc cathode) after long term storage without B+, that might make sense and is what I alluded to in the OP, sure I read that somewhere along the line in vintage material - operate the valve without B+ to allow getter material to do its stuff? Maybe that somehow got lost in translation to become delay B+ to avoid ion bombardment, but explanation got confused somehow? Don't know, just speculation if it turns out that little protects the cathode in normal operation after all. In practice it all works fine, after all.
@bigun I don't know it makes much difference whether one calls it space charge, electron cloud or whatever - if there were a dense bunch of free charge hanging out between k and g/a the associated field and potential seems highly likely to be disruptive to operation IMO. If potential/field was large enough to deflect an ion seems certain to prevent operation I think.
@bigun I don't know it makes much difference whether one calls it space charge, electron cloud or whatever - if there were a dense bunch of free charge hanging out between k and g/a the associated field and potential seems highly likely to be disruptive to operation IMO. If potential/field was large enough to deflect an ion seems certain to prevent operation I think.
Last edited:
Exactly. But you say that mu depends strongly upon a thermionic 'electron cloud', and thermionic emission varies massively with temperature of the cathode we agree. And we agree one can vary heater temperature visibly, and observe only very small effect on mu. Surely that suggests mu is not dependant on thermionic emission then, just as one might predict if there were no cloud, but rather mu dependence on cathode barrier potential and (weakly) on mean escape velocity ?
Yes, a switch that puts enough negative voltage on the grid to keep Ia in cut-off (when heated) will suffice to maintain full lifetime and performance of an expensive DHT.
No need to regulate the voltage either. but try to have it appear near to the same time as the HT.
Hey, something we agree about. It's not hard to find expressions elsewhere in terms of kT though, which always rings alarm bells in my book. Do you have a source to confirm your post that shot noise does not follow root Ia below saturation ? Thx. The more I think about this the less likely it seems......
- Status
- Not open for further replies.