# Thread: Cooling fan and tube life

1. ## Cooling fan and tube life

Hi All,
Cruising the endless blogistan, I found a thread a while ago, a young player bought a fairly new Marshall tube amp. The guy said he used it in his room, no A/C so it was hot as heck in there, and he had big trouble with tube life. Needless to say, no idea about bias or anything like that, but he said he had gone through tubes every few months until he put a cooling fan on the amp.
I don't remember this coming up, but if tubes are biased right, or wrong, why would a cooling fan help tube life? Maybe it was a red herring and he got several sets of s**t tubes, and just lucked into a good set when he happened to put the fan in.
Anyway, would one of those whisper quiet fans help with tube life, if so, why?
Thanks,
Sparky.

0

2. Well... There are two directions to go with this answer. One is that there is a vacuum in a tube, so any heating of the elements inside must cool exclusively by radiation. There is no convection. The relatively large space between the glass and the elements means that the temperature of the glass, as it relates to being a mass that can absorb and radiate heat away from the tubes internal elements, is almost moot. So, WRT tube life as it relates to tube temperature, a fan does very little.

But... Then there's the many amps that actually DO benefit from having a fan. I've read here that the fan is there to cool everything else in the amp that is heated by the tubes. While this is also relevant it doesn't account for reports of extended tube life.

So on the one hand we have physics telling us it doesn't work. On the other hand we have builders and users doing it that tell us it does. It wouldn't be the first time real world experience and scientific explanation are at odds.

3

3. The guy with no AC and problems with tube life was probably a red herring, although one which was being dragged along the track in the right direction. This anecdote has the same problem as most anecdotes - it is a retelling of that person's recollection and evaluation of events that may or may not have happened as he remembers them or perceived them, and leaves out a lot of things that would be needed to figure out if it was useful or not. One biggie would be whether he was using a power soak/attenuator to run it at full blast in his room.

First, there is no question that keeping tubes cooler will lengthen their lives. The application where tube life really matters, the finals of radio station transmitters, are actively cooled with serious blower or even water cooling systems. The advice from everything I've read from the "Golden Age" says to keep them cool.

The problem is how to do this in a way that works with guitar amps. The quiet computer cooling fan is kind of ideal, excepting that you need to come up with 12Vdc at a modest current. The stuff I've read on using a fan to cool tubes says that you can make the glass outside of an output tube cool enough to touch even at full power by using even a little airflow across it. The standard naysaying response to this is "it's too loud!", which is nonsense with the fans you're thinking of. I have messed with the fans in my computers and I can't hear some of the fans from 3 feet even outside the cabinet. The second issue is dust. Fans will cause dust collection. A layer of dust on your tubes will insulate them, and make them even hotter inside. Sigh. Filters, cleaning, etc.

My best guess is - yes, they'll last longer if cooled, but that means you'll need to do some worrying about dust, and that the blog guy was possibly relating what actually did happen, but that may not mean much.

3

4. One is that there is a vacuum in a tube, so any heating of the elements inside must cool exclusively by radiation. There is no convection.
This is not completely true. Equilibrium electrode temperatures are the result of several heat flows, mainly
1) Radiation from plate to bulb and ambience (cooling)
2) Radiation from hot bulb to plate (heating)
3) Heat conduction via pins and socket (cooling).

A higher ambient temperature will increase bulb and socket temperatures and can be expected to have some effect on electrode temperatures. I don't think though, that an increase of electrode temperature by say 20° will noticeable affect tube lifetime by itself.

But tubes also have a limit for bulb temperature (e.g. 250°C - who knows if modern tubes can stand that much). Exceeding this limit may release absorbed gas molecules from the getter and other parts. And it may even cause the glass seals around the pins to get leaky, permanently degrading the vacuum inside. A "gassy" tube is defective.

Apart from exceeding temperature limits, the failure rate of electronic (and other) components increases with temperature. Often the rule applies, that the failure rate doubles with any additional 10°C. This means that an increase of ambient temperature by 20°C is likely to reduce lifetime by a factor of 4.

My advice to the OP: Cooler tube envelope=longer tube life. But I am not sure if it is a good idea to direct a strong airflow directly at power tubes, as this might cause non-uniform bulb temperature and stress in the glass.

5

5. Originally Posted by R.G.
The application where tube life really matters, the finals of radio station transmitters, are actively cooled with serious blower or even water cooling systems.
Sorry but you canīt compare them to regular tubes in Guitar amplifiers.

"Ours" have a basically cylindrical plate inside a glass bottle, with vacuum (perfect insulator) between them.

High power transmitter tubes are not built that way.

Rather the metallic plate *is* the vacuum bottle, you still have vacuum between it and screen/grid/cathode but the plate itself is freely accessible for cooling.

See the popular 4CX250 as an example: the plate (the metallic tube in the middle of the assembly) not only is visible and accessible, but it comes with built-in fins and an aerodynamic tube enveloping them, for maximum efficiency:

3

6. Originally Posted by J M Fahey
Sorry but you canīt compare them to regular tubes in Guitar amplifiers.
Sure I can - just not completely accurately.
And strictly speaking, I didn't really say that they were the same as our glass-envelope tubes, only that they were actively cooled.

You are correct on the differences in structure, and that it's much more difficult to get heat out of a plate inside a vacuum envelope than out of a plate directly exposed to cooling fluids.

Combining some of Helmholtz' observations, the main ways that plates are cooled in guitar amp outputs is by radiation from the plate to the glass envelope, with some minor conduction down the supporting pin structure. But cooling the glass envelope does help cool the plates, as glass re-radiates heat if it's hot, so there is a back radiation from the glass into the hot plates that reduces the net heat the plates get rid of. Taken to the extreme, you could heat an unpowered tube plate by just heating the exterior glass enough to have it re-radiate to the plate structure. So keeping the glass cool helps keep the plates cool.

And you can do that by blowing just a little air across the glass. Glass isn't too effective a heat conductor, so removing even a little heat per unit area from the outer surface cools the outer surface a lot. Probably more importantly, it increases the heat flow through the glass by forcing a big thermal gradient from inside the bulb to outside the bulb, so the glass does conduct more heat out. Yeah, glass isn't a good heat sink material, I know, but it's all we have to work with.

How much better things get are tricky to evaluate. The glass and plate exchange heat back and forth according to the Stefan-Boltzman Law as modified for gray bodies, emissivity, incidence angles, and strings of things that are why I took my one thermo course and walked away slowly...

But from a boundary-conditions viewpoint, every watt-second carried off the glass started out on the plate, screen, and cathode, less only the minor pin-conduction transfers.

I mess with this a bit back in my amp design days, and concluded that although fans help, they're a PITA. Better to design for lower tube dissipation or better natural convection. What any air movement at all is good for is safety. A hard-working 6L6 or EL34 will have envelope temps as high as 200C. Touch that and you get burned. Minimal airflow means you don't get flash burns.

The >>safety<< testers pointed this out to me when they noted that although I had put a perforated steel plate over the tube access at the rear, children could poke their hands through the speaker cone from the front and still get blistered that way. Sigh.

2

7. I went off to look up some numbers. Turns out that glass is dang near a perfect radiator in the IR. A perfect black body has a radiation constant of 5.7 (times a bunch of units and constants), lamp black is 5.16, and glass is 5.13. So glass is a lot better than many other materials. A high radiation constant also implies a lot of absorption as well. So glass both sucks in the radiated heat from the innards and radiates it away well.

What it sucks at is transferring the heat from the inside of the glass to the outside, and at transferring it to the atmosphere gasses outside. The thermal constant of glass is about 1 W/m*K, or one watt per meter (thickness) times the temperature difference. If I knew the thickness of a typical output tube glass envelope, I'd have a go at calculating the temperature difference across the glass in normal versus cooled situations.

2

8. I know for a fact that conduction cooling via the socket is considered essential for the lifetime of (low and high pressure) discharge lamps. There is a close relationship between lamp and tube construction and technology. No wonder that all major lamp manufacturers also made tubes in the past: GE, Sylvania, Philips, Osram, Tungsram, Mazda, Matsushita, Toshiba etc.
It's not about plate temperature. Its bulb and pin temperatures that matter.

1

9. I suspect that the pins are a significant cooling mechanism for discharge lamps. Different setup, dramatically higher temperatures, and (I think) much higher dissipated power.

I do know that all the tube data books I've seen talk about maximum bulb temperature.

Electron tubes die by several means. A big one is loss of emission from the oxide coated cathode. Emission testers were the commonest drugstore tube testers. Over time, the oxide layers get poisoned through several paths, but they all lead to the tube getting "weak". If the tube gets too hot, it can bake the trapped residual contaminant gasses that were adsorbed onto the interior surfaces off, and this poisons the cathodes quickly. Overheating can kill a tube by letting the fine grid wires sag and touch other elements, or by the repeated heat/cool cycles mechanically stressing a grid or heater wire until it opens or shorts to something else. Plates can simply melt. I've seen this.

One of the most unusual ones is where the heating from the plate was so concentrated on the area of glass nearest the plate that the glass softened and air pressure forced the glass into an inward-pointing cone toward the plate.

The picture that JM showed us shows how to deal with this: rearrange the elements in the tube so that the part that generates the most heat - the plate - gets the most cooling.

1

10. I can testify to tube glass getting hot and radiating heat away! I know that in one amp I built out of an old Bogen, I used the basic Bogen layout, but I added a backlit panel with LED's, and I located those too close to the power tubes. I can get the whole string working, and then at some point a couple will get too hot and the whole string goes out, so I'll have to add a fan to this amp, but I think I'll blow it right on the LED's. Should solve the problem.

Greg

1

11. "One of the most unusual ones is where the heating from the plate was so concentrated on the area of glass nearest the plate that the glass softened and air pressure forced the glass into an inward-pointing cone toward the plate. "

I have seen this more than once on 6L6 tubes.

It's pretty neat.

1

12. I have an EL34 on my shelf with the glass melted and pushed into a tiny cone towards the plate.

2

13. Originally Posted by Enzo
I have an EL34 on my shelf with the glass melted and pushed into a tiny cone towards the plate.

nosaj

1

14. In this image I found you can actually see the glass flattened against the plate on the left tube.

2

15. To summarize my thoughts on this:

1) (Glass envelope) power tube plates dissipate thermal power mainly via radiation.
2) Forced air cooling has no direct and only little indirect effect on plate temperature.
3) When saying "It's not about plate temperature" I meant that raising the ambient temperature by say 50°C will hardly affect the plate under normal operating conditions, but
4) It makes sense to decrease bulb temperature, as this reduces heating by re-radiation from the bulb, but most of all reduces the risk of outgassing and maybe leaking seals.
5) Once there are enough free gas molecules inside, the risk of catastrophic runaway dramatically increases.

For more detailed information:
http://www.tubebooks.org/Books/Atwoo...um%20Tubes.pdf

See especially pages 24 to 29.

5

16. Originally Posted by Chuck H
In this image I found you can actually see the glass flattened against the plate on the left tube.

I found a similar treated EL34 in a Matchless Chieftain several years ago, a good olˋ fetish for serious Voudou Biasing...

1

17. Traynor used a fan on it's YBA-1A, 85 watts clean out of a pair of EL34's.

2

18. Originally Posted by mozz
Traynor used a fan on it's YBA-1A, 85 watts clean out of a pair of EL34's.
Mesa Boogie also used a fan in there Mark series amps.

2

19. So, if I get the main points right, the airflow around the outside of the glass part of the tube probably won't help much, since the guts of the tube transfer by EM radiation (through the tube glass), but the airflow across the chassis, and so tube sockets and base of the tube, might help some, since some conduction does occur through the bottom of the tube, pins, base, etc?

Early on, after learning bits and pieces, I was very surprised at the Fender typical design, the chassis is closed, no air holes or vents, and the open side of the chassis is bolted up tight to the head or combo cab. And the tubes are upside down, so hot air rising would tend to heat the chassis rather than cool it.

Even though we have 35, 40, 50+ year old amps out there, still cranking away, tube construction (as well as anything else) ain't what it used to be. I had thought about having holes cut in the top of the combo cabinet and little vent screens put there, but since the chassis is all closed up, no air could get through anyway.

But might be worth it to put one or two of those little pancake fans on the sides of the combo and have air move across the bottom of the chassis, transformer, and tubes?

0

20. Originally Posted by Zouto
I found a similar treated EL34 in a Matchless Chieftain several years ago, a good olˋ fetish for serious Voudou Biasing...
WOW the tube heated up enough to melt glass?

0

21. Don't you need like 1600 to 1800 C to melt glass like that? What happened inside the tube to generate heat like that?

0

22. So, if I get the main points right, the airflow around the outside of the glass part of the tube probably won't help much, since the guts of the tube transfer by EM radiation (through the tube glass),
Nope, cooling the bulb doesn't cool the plate but has been shown to increase tube life. Please see the link I posted above.

1

23. I think it was concluded that cooling the glass helps too, just not as much we might hope. But you can use a really small, quiet fan. You only need to move a little new air in and take old air out. Still air is a pretty decent thermal insulator. As long as there's any air flow at all what little heat it can be carried away will be.

EDIT: The one time I installed a fan in a project I followed the advice of members here and set it up to suck air out of the space to be cooled so the drawn air across the circuit was a gentler, wider swath. Rather than a concentrated blast of air going into the amp. Further, if dust is a concern you can filter the venting so the air coming in will be dust free. It's much harder to keep effective air flow filtering right at the fan.

2

24. Originally Posted by mikepukmel
Don't you need like 1600 to 1800 C to melt glass like that? What happened inside the tube to generate heat like that?
My explanation: A hot bulb frees gas molecules absorbed in the getter metal layer on the inner bulb surface. The gas molecules get stripped from some of their outer electrons in collisions with electrons. The remaining positive gas ions get attracted by the negative grid and cause a positive grid leak current, which produces a more positive grid voltage across the grid leak resistor. The result is increasing idle current and (worst case) thermal runaway.

1

25. Awesome, thanks everyone, got it.

0

26. I am not a big fan of tube down designs (except my 50w JCM800 2204 build which runs really cool) , so I usually put a 12v computer fan in the cab base drawing cool air from under the base right under the PT area(supported on rubber feet of course!). I normally run from a Li-Ion 9v cell, and it will run for many hours, keeping the PT, chassis and to some extent tubes a lot cooler.

I have this in my tweaked 6L6 35w 5E3 which runs really hot (30w 250R cathode resistor(s)).

2

27. Originally Posted by Rog_B
I usually put a 12v computer fan in the cab base drawing cool air from under the base right under the PT area(supported on rubber feet of course!). I normally run from a Li-Ion 9v cell, and it will run for many hours, keeping the PT, chassis and to some extent tubes a lot cooler.
Why use batteries? Why not run it from the heater supply?

1

28. There are a few misconceptions about heat transfer in this thread

It is better to think of heat transfer within the valve as just based on the temperature difference between plate structure and whatever it can radiate to - the plate is hotter than glass envelope or anything else nearby and so heat transfers by radiation away from the plate. Trying to think of it as transferring in both directions, but one direction dominating, does not help imho.

Most of the plate radiation does get absorbed by the glass, but there is a significant percentage transmitted through the glass and landing on anything nearby - that is because the heat transferred has a wide spread of wavelengths, and glass is transparent at the shorter wavelengths but has a rather sharp transition to opaque (that depends on glass thickness) and so absorbs a high percentage. So keeping the glass clean does have some minor benefit.

By far the most heat is transferred away from valve glass by convection to surrounding air - any improvement in air flow, whether by aiding the chimney effect or by movement from a fan, is going to lower glass temp.

Probably the main cause of lifetime depletion for tubes operating at or near rated dissipation is outgassing of glass and internal parts, whereby the getter finally gets consumed and can't then maintain a low enough gas pressure for normal tube operation. A bogey tube glass temp level at rated dissipation is the premise for the typical 2,000hr lifetime rating - there will on average be a plus/minus lifetime change for lower/higher glass temp. It's probably a law of diminishing returns for lower glass temp, as other lifetime issues start to come in to play, but certainly worth keeping in mind if your amp has a poor natural ventilation path to and from the output stage tubes, or your room is stinking hot.

There was obviously a lot of technical effort spent by large tube makers to confirm bogey lifetime ratings and what had to be done to achieve those levels - probably the best reference for a quiet Sunday afternoons reading is the 1962 RCA book on all matters technical within a valve.

2

29. Originally Posted by trobbins
Most of the plate radiation does get absorbed by the glass, but there is a significant percentage transmitted through the glass and landing on anything nearby - that is because the heat transferred has a wide spread of wavelengths, and glass is transparent at the shorter wavelengths but has a rather sharp transition to opaque (that depends on glass thickness) and so absorbs a high percentage. So keeping the glass clean does have some minor benefit.
I'm no physics major, but...

It's my understanding that heat moves in the infrared spectrum of wavelengths (300GHz to 400THz). I'm open to learn about what the opacity of glass is at wavelengths in that range. What really caught my attention though was the implication that if more spectrum moves right past the glass that heat transfer would be improved. I would think that if a fan is in use, and the glass is subject to cooling via the convection of circulated air AND the glass is absorbing a maximum amount of heat from the plate then that would be the most efficient means of removing heat from the plate.?. Any spectrum outside of infrared that the tube is emitting may have consequences, but I don't know that heat is one of them. Further, as far as I know convection is a more effective heat transfer than radiation alone. So I would have thought that heating the glass, which is in close proximity to the plate, and keeping it cool so it can absorb MORE heat from the plate would be about as efficient as we could want. The implication is that any heat that might be radiating through the glass is dissipating more slowly that that which the the glass can absorb and then be relieved of by a cooling fan.

Like I said, not a physics major. Just thinking out loud.

1

30. Originally Posted by Chuck H

Like I said, not a physics major. Just thinking out loud.
I could hear the gears grinding and creaking.

Greg

2

31. Originally Posted by soundmasterg
I could hear the gears grinding and creaking.

Greg
Well I'd hoped to come across better than that. But it's alright I guess

2

32. Good reference information on opacity of valve envelope glass to plate thermal radiation is in this application note from a temperature sensor company:
http://support.fluke.com/ircon-sales...62_ENG_A_W.PDF

A pictorial of the emission spectrum of a blackbody down at plate temperature is shown at:
https://www.hgh-infrared.com/FAQ/Blackbody

Glass is pretty much opaque at wavelengths longer than 5 micron, and a 230C (440F) blackbody has a peak response at 6 micron. So I'd be estimating at least 70-80% of the plates net energy transfer that goes via radiation to the glass envelope gets absorbed as heat.

2

33. Originally Posted by Chuck H
I'm no physics major, but...

It's my understanding that heat moves in the infrared spectrum of wavelengths (300GHz to 400THz). I'm open to learn about what the opacity of glass is at wavelengths in that range. What really caught my attention though was the implication that if more spectrum moves right past the glass that heat transfer would be improved. I would think that if a fan is in use, and the glass is subject to cooling via the convection of circulated air AND the glass is absorbing a maximum amount of heat from the plate then that would be the most efficient means of removing heat from the plate.?. Any spectrum outside of infrared that the tube is emitting may have consequences, but I don't know that heat is one of them. Further, as far as I know convection is a more effective heat transfer than radiation alone. So I would have thought that heating the glass, which is in close proximity to the plate, and keeping it cool so it can absorb MORE heat from the plate would be about as efficient as we could want. The implication is that any heat that might be radiating through the glass is dissipating more slowly that that which the the glass can absorb and then be relieved of by a cooling fan.

Like I said, not a physics major. Just thinking out loud.
Well, I'm a physicist, but...

I have a language barrier and don't seem to get your point.

Considering envelope material properties and physical laws involved, I have come to the conclusion that envelope temperature has little effect on plate temperature. There is some re-emission from envelope to plate but as the plate is typically much hotter, this effect will be rather small. Heat radiation is governed by the Stefan-Boltzmann law of heat radiation, which shows that radiated heat power drops with absolute temperature to the fourth power. So if we manage to lower envelope temperature by say 50°C, the cooling effect on the plate will be MUCH less.

There has been a lot of discussion about plate temperature and how to bring it down. I think this misses the point. How many tube failures are actually caused by a somewhat increased average plate temperature as the primary cause?
The plate is the most rugged and massive part inside a tube and can take a lot. It seems much more important to avoid out-gassing and thermal run-away caused by excessive envelope temperatures (and of course to avoid over-dissipation of the delicate screens).

Forced air cooling helps to reduce envelope temperatures and prolongs the life of electrolytics, so it makes sense in particularly hot amps.

2

34. Originally Posted by Dave H
Why use batteries? Why not run it from the heater supply?
It was easy to supply clean 9v DC this way, no other reason and it has worked well for a couple of years now. I have used heater 6.3VAC rectified on other amps for various tasks, but I like the way this works.

1

35. Fan will help, I've done it often - you typically don't need to blow massive amounts of air - just put a air pressure bias in to assist convection.

Rectify the 6.3V heater to get around 7VDC and run a 12V DC fan from that. It runs slower but quiet with enough oomph to make a genuine difference.

Fans are better at blowing into a chassis (blowing against a higher pressure) than at extracting air from a chassis (sucking from a lower pressure). So blow air in, this is also always the quietest option as the chassis acts as a mechanical noise baffle.

Cheers,
Ian

2

There are currently 1 users browsing this thread. (0 members and 1 guests)

#### Posting Permissions

• You may not post new threads
• You may not post replies
• You may not post attachments
• You may not edit your posts
•