Baking or de-magnetizing toobs?

[GrimJosef]

Originally Posted by Dave1000

...
On modern machinery the best vac' you'll get in a jacket is about 10^-2 Torr, the getter takes it down one decade. That is with silicone oil diff' pumps ...

I guess that's a measure of the very limited time that lamps are pumped for. The twin oil diff pumps on the e-beam generator that I looked after for a few years would reach 10^-5 mbar (not so different from Torr) in 20-30 minutes if we hadn't had the diode chamber open to damp air for hours. Wikipedia reckons they can get down to 10^-9 mbar https://en.wikipedia.org/wiki/Diffusion_pump, presumably under ideal circumstances. Even a good rotary pump would get below 10^-1 mbar if the oil was nice and dry.

Cheers,

GJ

[julie_m]

That's still considerably hotter than a standard kitchen oven gets! Gas Mark 9 is only 250°C. And the temperature rise on the outside is still only about a tenth of the temperature rise on the inside. So, assuming that the transfer of heat works equally well (or equally badly) either way, then baking a valve at no. 9 is going to make it about 25°C hotter internally .....

Unless I'm missing something really obvious, there's no way it can make a blind bit of difference. Not that there's any way to tell whether or not the valve's characteristics have changed anyway; since it's the same one, so you can't do proper side-by-side comparison. And you can't even start with two "identical" valves and bake just one of them, since you have no way to know exactly how paribus your caeteris is (to murder a phrase).

[GrimJosef]

Originally Posted by julie_m

That's still considerably hotter than a standard kitchen oven gets! Gas Mark 9 is only 250°C. And the temperature rise on the outside is still only about a tenth of the temperature rise on the inside. So, assuming that the transfer of heat works equally well (or equally badly) either way, then baking a valve at no. 9 is going to make it about 25°C hotter internally ...

I don't think that's right is it ? Heat flow from the inside to the outside and from the outside to the inside aren't simple mirrors of one another. In the former case the outside of the valve has the rest of the universe to cool to. That's what stops it from getting nearly as hot as the inside. In contrast when we heat the outside of the valve heat will flow to the inside but it has nowhere to go from there. So the temperature of the inside will rise until, given long enough, it is at the same temperature as the outside. If there is any heat transfer at all then there is no steady-state solution with the outside hot but the inside still cold (which is why baked Alaska has to be made quickly).

Of course there's no way of getting the inside hotter than the outside if the external heating is just conductive. But operating a valve will raise the temperature of the electrode structure above the temperature of the envelope, and usually by a large amount.

Cheers,

GJ

[Dave1000]

Lamps and valves are made on rotary machines with an index time that varies a lot, but on older machines, such as would have been used back in the heyday of valves, around 5-10 seconds - it's a long time since I was lamp-making, but that should be about right.
On any rotary exhaust/evacuation machine, the assembly OUGHT to be heated to out-gas everything as much as possible - I did not notice such on the Mullard Blackburn valve manufacture film that is available online (which may well yield a figure for index time). Conventionally this would include some heating/use of the thing being bottled - an arctube, filament or valve assembly - as well as external heating. To reduce oxygen and water as far as possible, conventionally there would be purging with dry nitrogen at least once. So, on a machine with something like 20 heads...…….. do the sums. Maybe 30 seconds final pump across 3 positions on the turret.

You also do not need better than 10^-3 Torr for any lamp/valve, or if you do, it ain't achieved anyway, not on a mass-produced product at realistic cost.

[GrimJosef]

Originally Posted by Dave1000

... You also do not need better than 10^-3 Torr for any lamp/valve, or if you do, it ain't achieved anyway, not on a mass-produced product at realistic cost.

Do you mean that you don't need the pressure to be any lower than this before you flash the getter ? My copy of Vacuum Tubes by K.R.Spangenberg (McGraw Hill, 1948) says that the pressure in an old receiving valve (operating, so hot) will be at 10^-5 Torr and in a new one will be at 10^-6 Torr. Transmitting valves will have pressures a factor of 10 lower than these. When they're cold (room temperature) the pressure in the valves will be 10^-8 Torr. All these pressures will be after the getter has been flashed, of course.

The big difference between lamps and valves, which leads to the requirement for better vacuum in valves, is that valves rely for their operation on the generation and unimpeded flow of free electrons. Lamps, at least filament lamps, don't.

For unimpeded flow it's necessary that the electrons don't collide with any gas atoms or molecules as they move from the cathode to the anode. I thought I'd be able to look the mean free path for, say, 100eV electrons in, say, air at a given pressure up on the internet. But when I tried I couldn't seem to find any straightforward answers. In practice though I don't think it matters because I think that the pressure constraints on the generation (emission from a hot cathode) of the electrons are more demanding.

Two problems arise with the emission if there's much gas in the valve. First the oxide material on the cathode which actually does the emission is sensitive to chemical 'poisoning' by many gases. If we don't reduce the gas pressure to a very low value then gradually it will react with the oxide layer and destroy its emissive properties. This will shorten the valve's life. Second any accelerated electrons which do collide with gas atoms/molecules will tend to ionise the latter. The resulting positive ions will then be accelerated towards the cathode and will bombard it, which will also destroy it. It's these two cathode destruction processes which, in practice, impose the low pressure requirements that Spangenberg lists.

Cheers,

GJ

[Dave1000]

I have worked on lamp-making machines IDENTICAL to the ones exhausting valves in the Blackburn video - what goes into the compression head matters not one jot, lamp or valve, it is just a glass tubulation that is inserted and the head closed.
I have measured and had measured more than numerous lamps for pressure after exhaust, and after firing the getter you are around 10^-3 Torr.
Barium will react with oxygen and water and a few other things like carbon dioxide, it does not react with nitrogen at temperatures below around 200C, which will be present for a VERY short period as the getter is fired.

Forget incandescent lamps - they are nitrogen-filled if they aren't halogen. I am talking primarily about discharge lamps when I am talking vacuum jackets.

On machines indexing every several seconds, with a turret with only around 20 positions, even if pumped at every position, which is nonsense, you get a VERY short pumping time and once you get to around a Torr, you have such low pressures that differential pressures, which drive the flow of gas, are TINY - that is why genuinely very low pressure systems have HUGE pipes. Exhaust is not helped by the 3mm or so tabulation that connects the jacket to the exhaust machine, but you could not possibly tip-off anything bigger than around a 3mm ID tubulation.

[Dave1000]

Let me describe a “typical” rotary exhaust machine for small (internal) volume valves and lamps from back prior to the 1970’s (which still run to this day, I am absolutely sure), bear in mind it is a while since I worked on one, and I did not take detailed notes when I did.
The machine itself is roughly mushroom-shaped. The stem/stalk/column of the mushroom is hollow and carries all the connections/services to the valve-plate or, in the case of electrical connections, to sprung electrical connections in principle, akin to a commutator/carbon brushes. The valve-plate is attached to the top of the column. (A valve-plate is actually two lapped plates, on a 20 position head around 30cm diameter, on a 50 head machine something like 70cm). On a traditional 20 head machine the column will have an ID probably of less than 30cm, on a 50 head something like 60cm or a bit more.
On top of the lower half of the valve-plate sits the upper half, which is attached to the indexing drive and the wheel-like turret assembly that carries the compression heads etc..
Pumps and all other ancillary kit generally have to be located remote to the machine, for what I hope are very obvious reasons, although I have seen some machines with a small final “polishing” diff’ pump sat on top of the machine, above the valve-plate, but which must still be connected via the valve-plate, and I have seen one comparatively modern, HUGE, rotary exhaust machine with diff’ pumps mounted in the centre of the turret. Typically though, on older and certainly small machines, pumps will be 2-3m from the base of the column. I have never seen any pump connected with anything significantly larger than 75mm s/s tube. To fit all of the pump lines etc. through the inside of the column, some have to be reasonably small – typically 10-20mm.
The valve-plate will be a total of something like 70-100mm thick and each position will have a port of around 10-15mm diameter, through the plate. The top of the plate will be connected to the heads via rigid stubs attached to the plate and flexi or rubber vacuum hose. The compression head will have a small internal chamber and this attaches to the 10cm or so long, 3-4mm ID tubulation.
So between pump and lamp/valve, ignoring all the connections, twists and turns, each of them producing a pressure drop, there will be (roughly, probably conservatively) something like 3-4m of 75mm pipe, 2-3m of 10-20mm pipe/hose, 70-100mm of 10-15mm “orifice”, and 10cm of 3-4mm tubulation.
People may be familiar with CONVERT.exe by Josh Maddison? There is or was a similar file available as a free download that calculates pressure drops for you – long since lost by me, but if anyone wants to try to find it and plumb that lot in and see what sort of pressure drop you get between lamp/valve jacket and pump with an index time of umpteen seconds…….
As to operation of the actual machine, conventionally there would be, given in no particular order here, rough vacuum (in my experience, from a factory-wide system – “factory vac’”), leak check, purge (with dry nitrogen in the case of vacuum and most other jackets), ideally a station where the lamp (or valve) is lit to some degree to outgas things, plus the actual pumping to achieve evacuation of the jacket. Ideally there would be heat applied to the outside of the jacket as well, to outgas the glass. (Without a leak check station to find gross leaks, you are taking serious risks – it will let down your high vacuum lines, at worst blowing the oil out of diff’ pumps, or at least ensuring that the next 2-3-more indexes are making scrap as it will take that long to get even close to recovering the vacuum. Any leak detected would lead to that head/position being shut (clipped-off in the terminology familiar to me), upstream of the compression head.)

Assume that you have a 50ml jacket at 0.01 Torr, attached a compression head at 0.001, or 0.0001 Torr, through a 10cm, 3mm ID tubulation. That is a pressure differential of 0.009, or 0.0099 Torr between head and jacket - how long to get the jacket to 0.001 or 0.0001 Torr? Yes, I know, in theory it is for ever, but you get my drift.

[GrimJosef]

Originally Posted by Dave1000

... I have measured and had measured more than numerous lamps for pressure after exhaust, and after firing the getter you are around 10^-3 Torr ...

Well, there's a conundrum then. I thought I would simply illustrate it by going to a few published references from the history of valve manufacture, quoting their pressure measurements and showing that, like Spangenberg (whose book is a recognised classic in the field) they all say something different from you. I've spent more time than I should have trying to do this. To my surprise, while book after book after book states that 'hard vacuum' is absolutely critical to the reliable functioning of valves and often spends pages and pages (e.g. 62 pages in Spangenberg's case) explaining how they go about achieving this - with very careful material selection, heating to outgas those materials, external pumping during the outgassing and oxide emitter formation and then sealing and firing the, again carefully chosen, getter, they almost all fail to quantify the actual final pressure that they achieve. I have no idea why this is. Clearly it's the only number that matters and it would be pretty easy to measure. It's very frustrating. But for what it's worth the final pressures that I could find are:

Wikipedia, which says vacuum tubes are exhausted to 10^-6 Torr https://en.wikipedia.org/wiki/Vacuum_tube#Vacuum#.

Sylvania's Walter Kohl, at the start of his 'Introductory Review' here http://www.tubebooks.org/Books/Atwoo...on%20Tubes.pdf says that a 'well evacuated tube' will contain gas at a pressure of 10^-7 Torr.

Herbert Reich wrote this huge, and hugely authoritative, book on Theory and Applications of Electron Tubes http://www.tubebooks.org/Books/reich.pdf. Towards the end he has a chapter on glow discharge tubes and on p420 (!) he says that the lower end of the pressure range for these is 0.1 microns. At the start of that chapter he says that normal valves have internal pressure so low that there is no discharge. 0.1 microns is 10^-4 Torr, so normal valves must, at the least, be well below that.

You say that you've never seen the pressure in a lamp below 10^-3 Torr and so you're sure that valves can't be below 10^-3 Torr either. All the valve experts from the past say hard vacuum really matters and when they (very rarely) quote numbers they say this means orders of magnitude lower than 10^-3 Torr, achieved largely by hot outgassing and, after sealing, gettering I believe. I'm afraid someone has to be wrong.

Incidentally, you say you did not notice any hot outgassing in the Mullard Blackburn films. It's shown and described at 6:45 here https://www.youtube.com/watch?v=1qMvhxo6B84 and at 22:40 here https://www.youtube.com/watch?v=dErsA6wFmlM.

Cheers,

GJ

[Radio Wrangler]

We could leave the valves as they are and see what the effect of baking and demagnetising the listener has on the sound?

David

[peter_scott]

With one of these little Chinese induction heaters you can induce red heat in the internals of small valves.

https://www.ebay.co.uk/itm/5V-12V-Lo...oAAOSw6GJZ81wT

Peter

[Dave1000]

Out of curiosity I was looking at vacuum gauge development yesterday, and the industrially useful version of the Pirani was developed mostly in response to the need for "hard" vacuum for tantalum lamp manufacture (I am insure how long they lasted in production, but not long. They were developed because tantalum is "easy" to make into wire, whereas tungsten certainly is not.) That struck me as odd, but presumably tantalum has some odd metallurgical properties not shown by tungsten (tantalum burns ferociously once ignited - it has been used for armour-piercing shell tips - it just ignites on impact and burns clean through. (I have also worked in tantalum cap' R&D as well.)).

Anyway, I did that to try to understand what sort of process pressure measurement was available, when, and certainly valves were being made after (and also before) the Pirani was available.

Actually measuring the pressure inside a vacuum jacket, accurately, of a valve or lamp is not as easy as you might think, but it has to be done on occasion, so there are methods and no doubt the various manufacturers use variations upon a theme. Measuring WHAT is in low pressure gas mixtures is even more tricky as you are relying largely on diffusion to get the gas mix into your measuring device, aided by a VERY small pressure gradient, and different gases diffuse at rates dictated by molecular weight. For sure, 10^-6 Torr in a commercially made lamp/valve jacket is pure and unadulterated fantasy - I only wish I could remember what the vac' achieved on/at the actual pumps on the exhaust machines was, but I can't.

To be honest, you do not need measurements to realise that the pressure inside a "vacuum" jacket cannot be anywhere near close to what pumps can achieve, because of the reasons included in my post about how exhaust machines are made, the fact that nitrogen will be the major gas left, and the speed of manufacture/index. The major headache is the simple and unavoidable fact that once you are at low pressure, it is impossible to generate a significant pressure gradient that would cause gas to flow at any significant rate.

I had not refreshed my memory about the Mullard film/video, but going back now, the index speed is around 3-4 seconds. It is interesting that they use induction heating to warm and outgas the metalwork, but apply no heat to the glass, although how much the heat might transfer from metalwork to glass is difficult to guess. It looks to be a 20-24 head exhaust machine though, so outgas, flush, pump etc. etc. and tip-off in a minute to a minute and a half.

Just an off-the-wall thought - I have never heard of it being done, but if you purged with dry oxygen, rather than nitrogen, the getter would absorb that and you WOULD get a very good vacuum...………….. There are lots of knock-on problems to solve with that, but...….. and I cannot remember the Ba yield of a getter ring to be able to relate that to amounts of oxygen...……………………….. Saes data needed...……..

[Dave1000]

https://books.google.co.uk/books?id=...vacuum&f=false

[GrimJosef]

I tended to think of the Pirani gauge as the thing we used for softer vacuum really. It wasn't a lot of use below 10^-3 to 10^-4 Torr. When we were getting down to operating pressures on the electron-beam diodes (10^-6 Torr or better, for preference) we used commercial Penning gauges or hot filament ionisation gauges. The readings on these do depend strongly on the species being ionised. But since this whole debate is just about orders of magnitude that doesn't matter so much here.

The easiest way to measure the pressure inside a valve has to be by using ionisation, since you've almost always got all the electrodes you need already in situ. The issue of course is calibration. I guess you have to make up a dozen special valves, identical to the normal ones but with a side-tube on the envelope. Then you could cross-calibrate the ionisation current in the valve itself against an externally attached gauge. Once you know how the current in the valve's electrode structure varies with independently measured pressure you can apply the same calibration factor to valves without the side-tube.

I take your point about the pressure that can be achieved during pumping on the exhaust machine. But that's not the final pressure in the valve of course. The open envelope is pumped when everything inside it is extremely hot - much hotter than in normal valve operation. So whatever pressure is achieved at seal-off will drop very substantially as the metalwork cools and gas adsorption back onto the metal and, more importantly, gas reaction with the getter starts to work.

The pressure reduction due to gettering is what gets (no pun intended) us down to the vacuum we need. I see that in slide no 16 here http://www-project.slac.stanford.edu...pec-060204.pdf, magnesium is ruled out for gettering in small vacuum tubes because its vapour pressure of 10^-5 Torr is too high. Barium (often as part of an alloy) is used instead - one more indication of the very low ultimate pressures which have to be achieved if the valve is going to work well for an acceptably long time.

If it's any consolation I worked for a few years on this project https://en.wikipedia.org/wiki/ALICE_(accelerator). The whole machine was under high vacuum, but the electron gun was a particular challenge. The electron emitter (photocathode) was GaAs activated with essentially a monolayer of Cs atoms on the surface. To give this a reasonable lifetime the gun had to run below 10^-10 Torr (ideally around 10^-11) https://accelconf.web.cern.ch/accelc...rs/mopc062.pdf. That took about a month to pump down, using some pretty serious pumps. But that sort of vacuum is not a job for part-timers. The project had a group of vacuum specialists who did that and only that.

VB

[Dave1000]

Apologies - Penning/Pirani - I was the practical one, not the technical one...……..

PV=kT(=nRT). I doubt the temperature is much more, possibly less than, +300C at its hottest, so pressure on cooling would only around halve, I doubt it would drop by 60%. Don't forget that even red heat is only around 700-800C.

The "problem" with barium getters is that they do not react with nitrogen - the major gas in the jacket, even if it isn't purged with nitrogen. In fact, reaction with hot valve/lamp components apart, you'd be a little better-off using dry air rather than dry nitrogen for any purge. I suppose that I ought to say that the only pressures that I had measured had been achieved using barium getter. Some lamps certainly use zirconium alloy? (which is used as sintered pellets rather than being evaporated).

If you take a look on the link, that refers to how low gas pressures were measured in my time in lamp making - smashing lamps inside or attached to sealed systems, attached to either a mass spec' and/or high vac' system. It also references papers on measuring very low lamp and valve jacket pressures, although I failed to find the relevant ones. (One of the names at least, worked or works for Saes, who had/have a near western (whole?) world monopoly on getters, but it was back in the 70's).

[GrimJosef]

Originally Posted by Dave1000

... PV=kT(=nRT). I doubt the temperature is much more, possibly less than, +300C at its hottest, so pressure on cooling would only around halve, I doubt it would drop by 60%. Don't forget that even red heat is only around 700-800C ...

That's true only for gases which are not subject to adsorption/desorption on the metal surfaces or to chemical getter pumping. I suspect that once the valve is sealed those two processes dominate how much gas is present at (relatively) low temperatures and the simple Gay-Lussac dependence of pressure on temperature doesn't.

The adsorption/desorption process is a threshold one, so it's usually characterised using an Arrhenius equation. The evolution/disappearance of a great deal of gas over a narrow temperature range is so striking that it's actually used as a chemical analysis technique https://en.wikipedia.org/wiki/Therma...n_spectroscopy. I don't know the numbers, and everything will depend on them. But I've watched grid current rise as a gassy valve warms up (I mean on thermal timescale for the whole structure, not just the cathode warming up and the electron emission starting) and a two order-of-magnitude change is common.

Everything depends on the numbers of course, and we don't know them for the ultimate pressure so we really aren't going to know them for the effective pumping rates of these two processes. But as far as gettering goes, they put getters in for a reason. If they weren't having a major impact on the pressure then they wouldn't bother using them.

Cheers,

GJ

[Dave1000]

In the case of lamps, barium getters give you around one order of magnitude drop in pressure. Having measured lamps with the difference, the watts are substantially different.
I did not ask about the reason why this is so (unusually for me), but was assured that between 10^-2 and 10^-3, "something happened".

I have also pumped a lit (35W? 55W? SOX) lamp for days, on a system that I was advised was perfectly capable, with a manifold pressure substantially lower, only ever to achieve 10^-2 Torr. The lamp stubbornly refused to drop in watts.

Whatever the final pressure achieved in a valve, one thing is absolutely for certain in commercially made ones, the final pressure is at least one, perhaps as many as three orders of magnitude lower than results from simple pump evacuation.

[MotorBikeLes]

Maybe a bit off topic until near the end, but here is Tektronix making Ceramic tubes (as in CRTs). (7000 series, 454 and many others). More interesting to me as it took me back to my time as a ceramic technologist.
https://www.youtube.com/watch?v=G0Dci5RPe94
Les.

[Dave1000]

Well, that music certainly fits the date well enough...….groovy, man.....

Presumably the same ceramic used for lamp arctube manufacture? PCA - poly-crystalline alumina?
Interesting that it was machined accurately way back then - engineering ceramics, so-called, are a relatively recent thing to be generally available and in use. For lamp arctubes, essentially no machining was/is used - parts are manufactured (usually by pressing, but sometimes by injection moulding or extrusion, then sintering) to finished size.

Sadly, no mention of sagger makers' bottom-knockers :-(

[MotorBikeLes]

Just about everything "ceramic" was well established by then. I had met it all on my course, and that was finished five or more years before that publication date. all pretty well done though.
There was a reference to the pressure vessels for the isostatic pressing being made from the same material as gun barrels.
in about 1963 we visited either(or both) Lodge plugs and S.Smith (Champion or KLG, I forget) and their isostatic presses were actually modified naval gun barrels. Some of the practices there were rather outdated, but they had various ministry contracts, and they could not alter anything without prolonged consultation with the appropriate department. Easier to just stick with "outdated".
The use of spray drying was fairly new, eventually arriving in the tableware industry 15 years later, just in time to offshore everything to the far east and "shutting down" Stoke on Trent.
Les.