Thermionic Valve Emission.

[Radio Wrangler]

Emission happens at the surface of the cathode. It is a function of the materials and structure of the surface, of the temperature and of the strength of the electric field created by all the other electrodes.

In a valve suffering reduced emission, the heater still takes the cathode to the same temperature and the voltages on the electrodes and their positions haven't changed. The reduction in emission has to be in material and structural changes in the surface of the cathode.

Those 'orrible brown stains in EL84s and line output valves came from somewhere.

In a bright emitter valve with a pure tungsten filament, the filament can't become any less tungsteny, what can happen is changes to the surface finish and reduction in diameter as tungsten slowly evaporates off until the filament fails. Light bulbs have long employed a gas filling to increase life by re-depositing tungsten vapour on the filament, allowing higher temperatures and whiter light, but that can't be done in a valve where the vacuum needs to be good. Early sets often had filament rheostats so compensation could be made for thinning filaments.

Thoriated tungsten filaments benefited in two ways. The thorium acts to promote electron emission and it increase the melting point of pure tungsten. It is used for these purposes in TIG welding electrodes. Thorium has a half life for the most stable isotope (the one therefore most likely to be found) of 14 billion years, so it's going to not change much over the life of a valve. To influence this much would need bombardment of the nucleus, not just ripping away electrons from the outer shell. These valves ought to be rather long lived, apart from manufacturing defects, accidents and leaky seals.

Coated cathodes and filaments are where things get interesting. Those hard driven TV valves are driven to very high cathode current densities. Those big currents could have been easily handled by a big valve with a big cathode and a modest current density at its surface. But that would have cost space and money. Clever coatings have allowed small cathodes to pass impressive currents, but if you really use that capability, you run into a wear-out mechanism in the cathode coating material/structure. The initial activation process shows that changes can occur, as do rejuvenation processes, all bringing active material to the surface.

It's interesting to note that high power transmitting valves take the very high voltage, moderate current density route for long life and high power.

Missiles.... If you're flying along in time of conflict, then as the old saying goes, "if the enemy is in range, then so are you". You really, really want your weapons to work reliably or else you are toast. Your life depends on them. The valves in a guidance mechanism or fuze may not need a long emission life-expectancy, but they have to be superb at storage life and they may get run up from time to time for testing.

David

[peter_scott]

Valves used in under sea telephone repeaters were tested for their lifetimes.
See: http://lampes-et-tubes.info/sp/Elect...ble_System.pdf

Peter

[PJL]

There are other articles about TAT-1 which ran with the same valves for 22 years. Seems to me that testing of valves could take a lifetime - so any datasheet figures will be by cautious extrapolation.

Failures I have seen are mostly from abuse such as coupling capacitor failure or mechanical problems such as vacuum failure. The older 30's and octal valves are more reliable as they ran cooler which brings us to the ECC85 which was designed for low life expectancy!

[Peter.N.]

Now I think about it ECC85's did have a relatively short life, didn't know they were designed that way though. On the other hand ECL80's some of which had quite a hard life lasted well and EF80's last forever.

Peter

[PJL]

I was joking.

I suspect the main cause of failure is cathode contamination from the practice of switching off the HT in AM settings.

However they have small anodes, minimal electrode spacing, about 5W dissipation, and are covered in a metal RF can none of which will help. Not sure what the failure rate is like in FM only sets.

['LIVEWIRE?']

In a long thread in the 'Record Players, etc., section, I have been commenting on, and asking questions about, a Marshall JTM45 Guitar Amplifier. As some will know, this has a 'standby' switch which removes the HT whilst leaving all the heaters powered. During the course of my tests, I have, several times, put the amp into 'standby' mode for short periods, then switched it back on again, with no sign of any ill-effects. Presumably, though, leaving such an amp. in standby mode for any length of time could cause valve failure due to the lack of HT?

[GrimJosef]

Another factor which seems to affect emission lifetime is operating the valve under cut-off conditions. This was sufficiently well known that the Philips/Mullard datasheet for the E182CC for example specifically mentions that this valve has been designed to maintain its emission when run like this.

Cheers,

GJ

[emeritus]

Here's an ad from 1945 about the reliability of the standard Mazda SP41 valves that were used in a submarine repeater. No change in performance after 11,000 hours continuous use, well over a year.

[G8HQP Dave]

The usual guitar amp 'standby' switch does nothing useful (for the valves) but has to be present because guitarists expect to see it and use it as a 'mute' switch.

[Radio Wrangler]

It seems like the right solution would be to fit a 'mute' switch and label it 'standby' ;-)

If we labelled it 'cathode poisoning on' no-one would turn it on, but then they'd not have bought the amp.

David

[ms660]

Some PA amps had stand-by switches, one old veteran told me it was to avoid any filaments going pop at the crucial moment.

Lawrence.

[kalee20]

Originally Posted by Radio Wrangler

...Clever coatings have allowed small cathodes to pass impressive currents, but if you really use that capability, you run into a wear-out mechanism in the cathode coating material/structure. The initial activation process shows that changes can occur, as do rejuvenation processes, all bringing active material to the surface.

Good post from RW. But exactly WHAT is that wear-out process?

Slow diffusion of active material to the surface takes place according to temperature, which I would have expected was largely independent of cathode current and anode dissipation. How does 'using the capability' affect life - are you saying, that a PL509 might have a short life when worked as a line-output valve deflecting a 26" 110° tube at a peak current of 500mA, but a much longer life passing 40mA peak in, say, a power oscillator for a tape recorder erase head - 'life' being determined by the point at which I could take it out, plug it into my valve tester, and find that it no longer quite meets its published Ia / gm?

I'm not disagreeing with what you say, but I can't see the reason behind it!

Failing emission is just ONE reasons for writing-off a valve as having reached end-of-life. Cathode contamination, as Livewire and GrimJosef mention, is another. Then there's also cathode interface resistance, caused by silicon as an impurity, which adds a time-constant into the cathode circuit...

[Radio Wrangler]

I was told of various ageing mechanisms. Material is sputtered off the cathode, barium has a very low work function and is a nuisance if much condenses on the grid, leading to it emitting grid current electrons.

The forming, activation, process for the cathode produces a very thin metallic upper layer. I always assumed the oxygen from the oxide form the metals started from must wind up on the getter. Subliming metal would remove this layer and reduce emission. Then the oxide coating can leach silicon ifrom impure nickel cathode tubes, making insulating silicates. So the emissive surface gets progressively insulated from the cathode's metal tube.

Thermionics lectures were a long time ago

David

[kalee20]

I've known about grid emission too, which causes reverse grid current. And the other mechanisms you mention.

All these, however, it would seem to me, take place with the cathode up to temperature. Barium slowly is sputtered off; silicon is leached; more active material diffuses to the surface. But what effect does drawing cathode current have? I can't see this making any difference to the processes, unless electromigration takes place.

Lucky you, attending thermionics lectures! I missed out!

[Radio Wrangler]

Electron emission has sme funny (peculiar) effects.

If you do electric welding, TIG is the cleanest and most versatile process. You normally weld with the tungsten electrode negative and the work positive. This means that the bulk of the heat occurs in the work, not in your electrode. I can be emitting up to 200 Amps from a 3.2mm electrode. It's usually thoriated tungsten which toughens it and pushes up its melting point as well as acting as an electron emission promoter. The tungsten is still an awful lot hotter than a coated cathode in a valve and it's emitting into argon plasma rather than vacuum so the current density is rather extreme compared to the valve cathode.

The surface finish on the tungsten makes a big difference to the quality of the arc. The end of the tungsten is ground to a cone shape of height roughly 1.5 times the diameter. But the orientation of the grinding is important. You grind them so the grain runs tip to base on the cone. not in circles parallel to the base. The latter would be the obvious way to grind a cone. But it illustrates that emission is influenced by field concentration at the tight radii of prominence in surface roughness. The exact tip of the cone gets the brunt of the heat and soon melts to roundness leaving more influence to the cone a bit further from the tip. If you let the tungsten get overheated, the tip forms a ball and the emission gets fatter and less dense. It gets harder to control as the point of attachment wanders over the sphere. Even with a cone, loss of detail in the grinding marks makes a difference.

Now, it gets interesting when the situation is reversed.

Welding aluminium is difficult in several ways, the colour change on the parent metal melting is much slighter. If you solder a lot and can see melting in the change of surface finish rather than glow colour, you'll find welding aluminium easier than most people. The other big factor is that the aluminium, even though you wire brushed it or sanded it immediately before has an oxide layer coating it. It is a very reactive metal.

The solution is to weld aluminium with AC. During the electrode-negative half-cycle the work gets the lion's share of the heat which is good. You need this to make your weld penetrate deeply. During the electrode positive time the greatest share of the heat is wasted in overheating the tungsten, which promptly indicates its displeasure by forming into a balled-end. Your arc defocuses, but you have to live with this because the last bit makes it all worthwhile. The electrode positive half-cycle blows flakes of oxide, along with a little metallic aluminium off of the surface of the work, exposing naked aluminium. This is quite dramatic. Along the path you have welded is a wider trail left behind you where you can see that the surface of the parent metal has been etched. This extends well beyond the melted zone. The flow of current, the emission of electrons, is clearly linked to the emission of material. It's all in inert gas, so nothing chemical is going on, it's purely physics.

I'm not surprised that in valves, high current densities result in loss of active material from cathodes and that surface finish has a big effect.

There was a time not too long ago when it was suggested that cold cathodes with nano-engineered spiked surfaces could bring valves back. Didn't happen, though.

Posh TIG machines don't just let you switch between DC and AC output to handle steels and aluminium alloys, fancy ones can synthesise sinewave output and also adjust the mark-space ratio of squarewave, and can distort sines so that one polarity is made shorter than the other. The sinewave loses you some heating power (handy on thin stuff) but is kinder on your ears. Distorting the waveform allows smooth control of the ratio of heating current to cleaning current and you can trim the width of your etch zone. Adjusting the frequency of the AC also changes the weld and affects the depth melting penetrated down to.

Taking things to these current densities makes some effects of electron emission and bombardment happen a lot faster. Temperatures reach 25,000 Kelvin in the plasma. you need to keep the arc attachment point moving under control to avoid boiling the parent metal. Some metal alloy components boil more easily than others and leave a changed alloy behind where you've overheated things.

Oxy acetylene welding has its own trick. You can use a flame with surplus acetylene as a reducing agent to turn rust back to iron, to convert the grade of steel behind where you've been, and to reduce aluminium oxide coating slightly. Good gas welding of aluminium hinges on using your rod to stir oxide platelets out of the way and let metal fuse with metal. You mustn't dshove oxide below the surface, and you must fish out oxide from the depth you are trying to fuse. Definitely a black art! After a lot of accidents hit the news and even BOC's acetylene filling plant in Bristol blew up, it seemed a good idea to go electric and leave acetylene behind, I still miss it when there's a seized bolt to be freed or I want to shape something.

That wandered a bit, but electron emission-on-steroids seems to show up some relevant effects.

David

[ukcol]

Déjà vu.

I've asked simplistic questions before expecting a simple answer and of course these things aren't simple.

Once again an exceptionally interesting discussion has resulted and i am very grateful to all of you for your contributions. I would like to thank David in particular for sharing his wide knowledge of this subject.

[kalee20]

I second that! Thank you David!

[Radio Wrangler]

I've just found a paper abstrct that tells me that white Barium Oxide crystals have a conductivity of 10^-4 mho/cm at 1000K which is a handy temperature to find a figure for.

So if a cathode has a coating 0.1mm thick and has 1 square centimetre of area, then 10^-4/10^-2 = 0.01 Mho or 100 Ohs in real money. This would have a serious effect on circuit behaviour, giving 1 volt per 10mA of degeneration.

But this presupposes that all the coating is BaO, but they were mixtures. I suppose cathode coating recipes were guarded like secret rocket fuel formulae. I think some of the cathode conditioning process was to produce metallic bits in the ix which were a lot more conductive, but I'm mostly guessing.

I do know that overrunning thoriated tungsten filaments was supposed to boil off a bit more of the lower melting poin tungsten and reveal some fresh bits of (higher meltng point) thorium.

Certainly, material moved around in hot cathodes, probably impurity mobility due to electromigration.

These processes act to limit the useful lives of high power semiconductors. Get the hot and under heavy current densities, the carefully positioned and profiled doping goes walkies. This can spoil the transistor, and so solid state stuff isn't quite as solid as the name promises and has a wearout mechanism at high currents at high temperatures. You can also get surface metallisation diffusing in, so posh RF power devices boast of gold metallisation because that is better at staying put.

I wonder if the rustling noisy valves described by G-J are due to resistance being formed inthe cathode oxide by silicate formation, if the silicate forms a resistance suffering from 'excess noise'.

David

[woodchips]

It was some years ago when I first looked through the pile of valves left me by my dad, and significantly added to my me over the years. Up until then valve amplifiers were if no real interest, tinnitus has that effect. But there is any amount of stuff, I won't call it information, on the web about valve amplifiers. It became apparent that the modern, Chinese etc, valves aren't the same as the 50's Brimar ones. In particular the spitting and sparking at high voltages. Why?

I wouldn't pretend to know, but the investigation was interesting resulting in the purchase of many old books on valves.

I would point anyone interested to:

Electron Tubes published in two volumes by RCA around the 1940's. These books are very well known for the article by Scharde about the 6L6, but there are other equally intersting articles.

Effects of Space Charge in the Grid-Anode Region of Vacuum Tubes.
Development and Production of the New Miniature Battery Tubes, this seems to be the introduction of the B7G all glass valve.
Analysis of Rectifier Operation, by Scharde, and will tell you why valves spit and spark.
An Infrared Image Tube and its Military Application, more for me, if you have a Philips XX1080 image intensifier as used in the CVRT light tank then this is how it works.

Another book is volume 17 of the Radiation Laboratory Series, Components. Around page 500 it covers valves, but on page 563 it looks at the 7F8 with scatter plots of the variation in anode current and transconductance with fixed and variable bias.

The Rad Lab series are really worth buying, some more than others, but for a few £ per volume money well spent.

[woodchips]

Forgot to say, the RCA books are available for download from

tubebooks.org

plus many others including the 1953 version of Langford-Smith!