"Less heat" from tube filiments... theory and reliabliity questions

[earthtonesaudio]

Different tubes have different power requirements for their filaments.  Common low-power tubes like 12A*7 require something like 300mA at 12V, while some submini tubes take 50mA at less than 2V.  Why can one get away with so much less power?  Can you "starve" a more "hungry" tube filament without risking damage to the rest of the tube?

[amptramp]

If you reduce the temperature of the heater, you will go from a space-charge limited to a temperature-limited state.  This can increase the noise, but many tubes operate just as well or better with the heater voltage reduced by 10% to 20% because they are well into space-charge limitation.  You will find some information in various sources in this page:

http://audiophool.cjb.net/Techno.html

particularly the Steve Bench page which was nuked by AOL in the general AOL implosion of 31 Oct 2008, but the alternate site may still operate (it doesn't today).  Warning: you may disappear into this site and not come up until the snow flies again.  Similarly:

http://www.tubecad.com/

has some advanced articles on tube circuit design.

The difference in heater power is the difference between directly and indirectly heated cathodes with indirect heat taking more power but allowing more versatility since the cathode can be used as a floating circuit element.  Note also that running filaments on DC will shorten life since there is a net migration of the tungsten.  Just look at  the light filaments on a car compared to a home incandescent lamp - ragged crystalline filaments on DC compared to more normal filaments on AC.

[PRR]

You can run 12AX7 at low cathode current with 11V across the "12.6V" heater. Gm will fall a bit, Rp rises a bit, Mu is unchanged.

12AX7 is, by design, "over-heated". It improves performance. You could build a high-gain low-current tube with less heater power, but then you lose the high-gain higher-current market, and honestly most tube workers don't mind the heater power.

You can run less heat, but as Ron says the performance declines. For small drop it is a lot like a smaller cathode or a larger cathode resistor. In fact THD may decline marginally, just as if you used a larger unbypassed cathode resistor. But at some point it stops acting like a tube and more like a cold empty bottle. That point may be 20% down on voltage which may be only 10% down on heater power. Not a terrific energy savings. And if you are also under-Volting the plate, it may punk-out sooner.

Damage? Only at high plate voltage and negligible plate circuit resistance. PULLING more electrons than the cathode wants to make will do harm. However the nominal emission on most oxide tubes is 10 times what they will ever be used at, often 50 times more than we need in thrifty audio. And that current is limited by large plate resistor, often low plate supply voltage. In this "stompbox" context I think under-heat damage is "impossible". However prolong trying may foul-up the oxides so it will need some hours at full normal heat to get back to full normal performance.

> 50mA at less than 2V

You can't (easily) build a useful unipotential AC-powered heater/cathode structure that small. That's a directly-heated naked oxide filament. Your heater circuit is also your cathode circuit. If you have more than one tube, this complicates circuit design: all your cathodes are connected together. These will also be LOW-current tubes, with generally lower gains. Battery-power designers DO mind heating power, because heater battery life is brutal.

Also note that most small naked-filament tubes "must" use DC heat: AC heater power would overwhelm the audio with hum.

> 12A*7 require something like 300mA at 12V

No. 6.3V at 300ma -OR- 12.6V at 150mA.

Also you may be comparing a twin-triode against a single unit. I can't offhand recall a 1.4V 50mA dual triode.

> filaments on DC will shorten life

Perhaps. But heaters run very much cooler than light bulbs, much slower migration. And heater life is NOT an issue on 12A_7-like tubes working anywhere near (or below) nominal voltage.

Also beware comparing 12V with 120V. The 12V has short fat filament, which means more thermal losses to supports. Efficiency is lower. Also car/home, power is tight in a car and hours are short so lamps are rated to run much hotter, better efficiency against shorter life.

[earthtonesaudio]

Thanks guys, very informative an interesting information.

With that in mind, I suppose I'll start starving some already starved-plate designs of their heater power as well, and see if anything interesting comes of it.

[Paul Marossy]

I've always heard that the heater voltage should be within +/- 10% of whatever it is supposed to be or it will shorten the life of the tube. And definitely not much higher than 10% or it will "burn up" the tube rather quickty. I don't know how true all of that is, but I've heard it at various tube amp forums over the years.

I don't worry so much about the heat from the tubes, I just put a little fan in my amps to keep them running a little cooler.

[earthtonesaudio]

The temperature doesn't actually bother me, it's the power consumption from the battery (or daisy chain, etc).  I like the idea of having tube based pedals, but I'd be more inclined to actually build one or two if the operating costs were a little more modest.

[merlinb]

I've always heard that the heater voltage should be within +/- 10% of whatever it is supposed to be or it will shorten the life of the tube.

If you running the anode starved then running the heater starved will have no effect on lifespan.

[R.G.]

Common low-power tubes like 12A*7 require something like 300mA at 12V, while some submini tubes take 50mA at less than 2V.  Why can one get away with so much less power?

At the risk of sounding ingenuous, subminis are little. Their filaments are little. They have small surface areas.

This is one of those things which should be a question on the final for every first semester physics courses. Why is a 10W light bulb filament hotter than a 1KW household clothes iron?

When you put power into things, they get hotter. But when do they stop getting hotter? Easy - it's when outgo equals incoming power. When incoming power is greater than the thing can get rid of, it gets hotter. There is *no* limit to the temperature something will achieve if you can keep pumping power into it that it can't get rid of. The temperature rises until energy out equals energy in.

With a submini filament, there is a very small surface area available to radiate heat away, and conduction is not very effective. Convection does not exist in a vacuum, so the filament can get up to effective electron emission temp with a small power input. The duotriode's heater is bigger; it has to have more power in to balance the ability of the bigger surface to send power out as both radiated light (including infrared) and any electron emissions that happen to come from the heater.

Can you "starve" a more "hungry" tube filament without risking damage to the rest of the tube?

Damage to oxide coated cathodes is dependent on the cathode putting out more electrons than the plate needs. This forms an electron cloud in front of the cathode which helps shield it from incoming accelerated gas ions. When you deplete the electron space charge, you can now start damaging the cathode. How fast this proceeds depends on the voltages to accelerate gas ions, the amount of leftover gas, and secondary effects inside the oxide layer on the cathode.

Most oxide cathodes are very efficient at emitting electrons, more than needed. Lowering the temp of the heater (and thereby cathode) is OK as long as you don't deplete the space charge in normal operation, and this has the side effect of lowering any aging processes on the heater. But don't deplete the space charge cloud.

I must have typed in about lowered heater current and tube life fifty times over the years.

[earthtonesaudio]

I must have typed in about lowered heater current and tube life fifty times over the years.

Well, I appreciate you taking the time to type #51.

[PRR]

> heard that the heater voltage should be within +/- 10% of whatever it is supposed to be or it will shorten the life of the tube.

That's not wrong.

However we could also advise against clipping amplifiers, over-powering speakers, or hanging around sound greater than 75dB SPL.

Stuff that musicians do ALL the time.

People who did not grow up with tubes seem to think tubes die a lot. The present tube makers may be taking advantage of this by foisting half-baked bottles on us, though the few new-tubes I have used did seem fine. In fact pots and jacks fail at least as often as tubes. Do we coddle our pots and jacks?

Short life: 10 hours a week for a year is 500 hours. At Rated Voltage, tube heaters do not fail in 5,000 hours. They often last 50,000 hours on uncontrolled wall voltage. The only heater failures I have EVER seen were in series-string sets and the HP 200AB which I recovered from a dumpster. Long series strings of different tubes have high starting surge inequalities (strings of similar tubes are fine). And if you drop a tube 8 feet onto steel, it may break. In my experience, they don't burn out.

HIGH voltage: an indirectly heated tube at 2X rated voltage lasts long enough to go out for lunch. It takes 3X rated voltage to burn out in less than a minute. Tubes normally run much cooler than lamps, take a lot longer to burn out with over-voltage.

Under-heating: if a servo-motor tube MUST deliver 400mA to turn its motor, and it is rated to do this with 5.7V (6.3V -10%), then at 5.5V it may not turn the motor at all, which might be a Problem. Likewise if a starved-plate design "works" at 6.3V, and barely-works at 5.7V, it may not-work at 5.5V. Not from damage, but just because the design did not allow for the shifts in parameters (Gm and contact potential). It could probably be re-designed to limp along at 5V heat.

So why this +/-10% guide? Think of your grandmother's kitchen radio. Or think of a radar set high in the Arctic. These chores may accumulate a LOT of hours. Radios and radars may be "critical": the radar must not miss "the attack", grandma may not want to miss her programs. And grandma can't change her own tube, and nobody wants a 3-day dog-sled trip to change a tube.

So the tube-maker's advice is correct but conservative. They have some customers who are very sensitive to tube troubles and replacement costs. They do not want the blame if a mis-heated tube fails to give ample service. By posting a note, they shift the blame to the designer and user.

You won't run-up the hours like a radar (on battery, you can't afford to!). You WILL breadboard and tweak to satisfaction. If the tube fails, your first fix is to bypass it or use another fuzz... as long as you make some music-like sound you will get paid. And your pay will easily buy another tube, and you won't have to pay a Tech for a house-call or dog-sled trip, you change it yourself.

Getting back to the point: you will not "improve battery life" by running 6.3V tubes on 5.x volts. OK, some percent better, but less than the difference from one brand of battery to another, or fresh batts versus ones that hung around the warehouse for a year.

If you must run tubes on batteries, use tubes made for the purpose. Hearing aid tubes have very teeny output but are very thrifty. Beach-radio tubes can exceed 1/4W output with "tolerable" battery life. Maybe. My father recalls that he was "always buying batteries" for the tube beach radio. And it should be noted that many of these radios were promoted by the battery companies! Not for our portable pleasure, but because they could sell the radio at-cost and grow rich on battery sales.

If have space for a top-cap Octal, and you can get one (tube vendors can be a little careless) the 1D8 is a dandy triode-pentode. Filament is 1.4V 100mA and will give about 100 hours of life from one D-cell. This is 2 cents an hour.... assuming you get over $2/hour to play this is under 1% overhead, not a major expense.

Look at the triode section. Rp is similar to 12AX7, but Mu and Gm are far lower. Even zero-biased it does not want to pass a whole mA. If you look at the physics of where Rp Mu and Gm come from, you will understand that this is a physically "small" triode, as R.G. pointed out.

The triode could be run with 150K plate load, 1Meg next stage load, give gain a bit over 15. The pentode could be triode-strapped into 22K for gain like 6. With 90V (2.2mA) supply it might output 20V peak, and the pair would take 200mV without overload. If B+ were reduced to 9V (at 0.2mA!) it might overload at 20mV input, a nice fuzzer.

OTOH, a 6AV6 heater-cathode triode could probably be made to fuzz guitar with a 12V plate supply, but heater battery costs would be SIX times higher (6.3V 150mA). Heater-cathode construction has many advantages, but battery life sucks.

[Paul Marossy]

Huh, interesting stuff PRR. And of course, RG, too.