cold heaters

[SISKO]

hi, im bulding a 12au7 amp. While searching for ways to supply the heaters i came with this contradictory things. On one hand in this link it says that running the heater cold makes it life shorter http://www.tone-lizard.com/Marshall_Myths.htm but im not sure wich tube mr Trimanie is talking about.
On the other hand, Merlyn says that by starving the heater voltage its lifespan can be increased http://www.valvewizard.co.uk/heater.html

Im looking to run the fiilaments at 4.85v. What do you guys say? Is it safe? My guts tells me to trust  merlyn as he knows his stuff very well

[Johan]

Marshall jcm2000 run its preamp tubes at 9 volt dc( series heaters, 4.5 volt/tube.) I've had mine for 6 years without replacing the tubes and it still sounds great. .. Don't worry about it. .. powertubes is probably a different matter. But you're talking about 12au7. ..

[merlinb]

Here is an excerpt from a book I am writing:

"The valve characteristics are normally specified for a heater voltage that is within ±10% of its nominal value, which accommodates normal mains voltage variations. Operating the heater above its allowed range will lead to premature ageing without bringing many significant advantages and is therefore to be avoided. Conversely, many designers prefer to operate at the lower end of the allowed voltage range as it encourages longevity while still meeting the data sheet specifications (voltage regulators also come in round numbers like 12V rather than 12.6V). The lower temperature reduces cathode evaporation and interface-resistance growth, and improves heater reliability. Potential lifetime therefore increases dramatically. Under certain conditions reduced heater power will also reduce noise and some users have also claimed improved linearity. It is sometimes argued that reduced heater power will lead to reduced lifetime. This is not true, or is at least only half true. When the heater power is reduced the cathode emission reduces and therefore the saturation current is reduced. It is therefore possible that an operating current that would normally be considered quite safe will result in saturation if the heater power is unusually low, and this will indeed lead to reduced lifetime. But this is only a general truth about saturation and is hardly the fault of the heater. If we are careful always to operate below saturation, as usual, then reduced heater power can only increase lifetime."

In other words, you can lower the heater voltage as much as you like and lifetime won't suffer, provided you operate at commensurately low anode current too. I'm guessin you're planning to run at a couple of milliamps or less, which I doubt would be a problem even with a 4.85V heater (funny number?).

[R.G.]

Merlin, I think you may want to change that a bit. The issue is keeping current under the level which removes the space charge, so the actual condition is that you you may lower heater temperature (by lowering heater voltage) as long as the current you use on the place doesn't use up all the cathode emission. What you need is not a proportional lowering in current, just a plate current that is low enough not to exhaust the space charge.

Unfortunately it is impossible for the casual user to know when plate current is below emission. To compound the issue, most receiving tubes are normally run far below their emission limit, so even normal currents work fine at lower heater voltages. Probably the only thing the casual user of tubes can do is to ensure that the plate current being actually used with lower heater voltage is to check that the plate current is lower than the specified tube maximum plate current from the datasheet in proportion to the heater temperature, not voltage. Emission is roughly proprtional to temperature for a small range above where emission starts in earnest.

And that's another gotcha. Heater temperature is not linear with heater voltage. It's proportional to about the fourth power of applied voltage, so strongly nonlinear.

Lowering plate current used in proportion to heater voltage reduction is OK, but probably excessive.

[merlinb]

Merlin, I think you may want to change that a bit. The issue is keeping current under the level which removes the space charge, so the actual condition is that you you may lower heater temperature (by lowering heater voltage) as long as the current you use on the place doesn't use up all the cathode emission.

Pretty sure that's exactly what I said. The current must be below the saturation current, since reaching saturation means there is no longer a space charge (by definition).

I might break out the curve tracer later, to see what a 12AU7 looks like with 4.85V on the heater  

[SISKO]

Great stuff!

Marshall jcm2000 run its preamp tubes at 9 volt dc( series heaters, 4.5 volt/tube.) I've had mine for 6 years without replacing the tubes and it still sounds great. .. Don't worry about it. .. powertubes is probably a different matter. But you're talking about 12au7. ..

Cool! This is what i intend to do, except that i have 9.7v (7809 + diode on the "gnd" pin)

.. This is not true, or is at least only half true. When the heater power is reduced the cathode emission reduces and therefore the saturation current is reduced. It is therefore possible that an operating current that would normally be considered quite safe will result in saturation if the heater power is unusually low, and this will indeed lead to reduced lifetime. But this is only a general truth about saturation and is hardly the fault of the heater. If we are careful always to operate below saturation, as usual, then reduced heater power can only increase lifetime.[/i]"

In other words, you can lower the heater voltage as much as you like and lifetime won't suffer, provided you operate at commensurately low anode current too. I'm guessin you're planning to run at a couple of milliamps or less, which I doubt would be a problem even with a 4.85V heater (funny number?).

Nice to see you around merlin as i cited your work. Is the author speaking of saturation as in clipping? In that case, i never knew that overdrinving a tube could reduce its life time.

A funny number indeed. And yes, just a couple of milliamps (havent measured it yet). Its just a class A 12au7 amp @ ~180v.
Ill give it a go. Thanks

Probably the only thing the casual user of tubes can do is to ensure that the plate current being actually used with lower heater voltage is to check that the plate current is lower than the specified tube maximum plate current from the datasheet in proportion to the heater temperature, not voltage.

How can i check this? Ive not seen a tube datasheet with a Ia vs heater temperature graph. Is there an equation for this?

[merlinb]

Nice to see you around merlin as i cited your work. Is the author speaking of saturation as in clipping?

No, saturation is not the same thing as clipping. This is a confusement of terms that some authors fall into after reading about clipping in transistors. In the tube world, saturation is when the cathode is passing the maximum amount of current it possibly can. Clipping is a separate thing, and has more than one cause (one of which could be saturation, much like in a transistor, but this doesn't normally happen in any convenitonal tube cuircuit you're likely to see).

How can i check this? Ive not seen a tube datasheet with a Ia vs heater temperature graph. Is there an equation for this?

I'll throw one on the curve tracer later and show you the results. I suspect a couple of milliamps won't be a problem.

[blackieNYC]

I think I was once told (or misinformed) that lower heater voltage is like putting audio thru a cold power tube, and that positive ions contaminate the cathode, or they are ripped out of the cathode. Something like that?  Been a long time.  Set me straight on that.

[merlinb]

I think I was once told (or misinformed) that lower heater voltage is like putting audio thru a cold power tube, and that positive ions contaminate the cathode, or they are ripped out of the cathode. Something like that?  Been a long time.  Set me straight on that.

Positive ions (created by collision with electrons in the anode current) are only a problem if there is no space charge protecting the cathode (it deflects the positive ions). That's why you need always to run the tube below the saturation current. And the saturation current gets exponentially less as you reduce cathode temp (i.e. heater voltage).

(Incidentally, at 4.85V the cathode will probably be running around 900 kelvin, some 100 kelvin below normal. Some quick estimates suggest the saturation current might be anywhere between 5mA and 50mA at such a low temp; it will be interesting to find out if I'm even vaguely right!).

[R.G.]

Pretty sure that's exactly what I said. The current must be below the saturation current, since reaching saturation means there is no longer a space charge (by definition).

It is now. Apparently you were editing while I was typing. 

[samhay]

might break out the curve tracer later, to see what a 12AU7 looks like with 4.85V on the heater  

If you wouldn't mind, it would be great to see what it looks like with 4.5V too - helpful for those of use playing with a 9V supply.

[PRR]

> running the heater cold makes it life shorter
> by starving the heater voltage its lifespan can be increased

So?

I can't see why we have *ANY* interest in heater life.

Heaters in small tubes routinely ran 10,000 hours, and sometimes 100,000 hours. I've overseen many tube systems that were left on 24/7 for over a decade.

The only exceptions were series-string radios and TVs. Series operation of different tubes causes transient stresses, and such products usually shipped with the very cheapest tubes available. (One other exception: a 6SJ7 in a 200AB which had been thrown in a dumpster.)

And what we probably want is "amplifier life", not heater-life. If a cold heater lasts forever, but the starved tube looses gain or "tone" due to underheating, is that a success?

"Thermionic" devices work on HEAT. We NEED to stimulate the little electrons to break-loose and do our bidding. (A lot of cold-cathode devices have been proposed, even exist, and they are all lousy audio amplifiers.)

I'm also looking at my tractor engine. Over-heating might hurt it, but its problem is under-heating (stuck thermostat). The oil never really got warm, nor had its water condensation boiled-out, and the sludge has mucked-up the rings bores and bearings. Tubes are not tractors, but just suggesting that machines usually have an "optimum zone" of heat, and there can be problems on *both* sides, hot and cool.

That said.... the 12AX7 is perhaps a candidate for "cool" heater. It is a very small (guts) tube with the same heater as the much fatter 12AT7 12AU7 types. This is one reason it has higher performance (transconductance per milliAmp) than the others; but it probably came down to using a stock heater spec rather than design a new smaller heater just for 12AX7.

*Always plagiarize the Ancients!!* They did it all before most of us were born.

Back when tubes cost much more (in real terms) than today, most everybody ran tubes pretty near their ratings.

One major and pertinent exception: Dynaco's tube preamp ran 12AX7 at 11 Volts (on 12.6V connection). This may have been more about hum (Dyna ran poorly filtered DC) than any concern about tube life. Also the Dyna runs the 12AX7 far below their limits (like 0.4mA in a 2mA tube). This is a known-good system.

> the saturation current

English has too many words already, but here we need more. "Saturation" is used in very different ways in BJTs, FETs, and vacuum tubes.

The very energetic experimenter can plot the effects. Variable heater supply. Monitor the tube voltages (thus currents). From 7.0V to 5.7V the tube voltages will change very little (but they will change, which is A Problem in sensitive DC amplifiers). From there down to maybe 5V the tube voltages will change (tube current is reduced by cool heaters) and you can observe that small-signal THD numbers actually go down (due to parasitic "resistance" in the starved space-charge) (but do we really want lower THD in a guitar amp??). Lower heater temps will probably throw the tube far away from the nominal conditions you stole from Fender/Marshall/Kay, and you will have to seriously up-scale the tube's resistors (reduce tube current) to stay near a good operating point. Since these resistors are already scaled for the external load on the tube, this is a bad way to go.

I'd keep a 12AX7 in typical guitar circuits well above Dyna's 11.0/12.6 point (5.5V on 6.3V connection).

[merlinb]

Well I grossly underestimated the saturation current! You have to drop the heater voltage seriously low to see saturation effects creep in!

[R.G.]

The 12AX7, as Paul notes, has a LOT of emission for such a small plate current. It is likely that as Paul said, they used the same heaters as the other 12A?7 family.

Tube life was often determined by the loss of emission, as evidenced by the popularity of the "tube testers" in drugstores and other stores. I actually remember seeing these as a small boy. Over time, the oxide cathodes get poisoned and emission drops. The tube is worn out when emission drops too low and the bias shifts too much because of that.

I've read notes on small receiving tubes getting lives in the 100,000 hour range, as Paul again noted.

However, filament life in both tubes and light bulbs was studied to death back in the 50s and 60s. Theoretically, life varies inversely with the thirteenth power of the applied voltage. So even tiny over-voltage/current/power reduces life a lot. Modest heater voltage reduction pushes heater burn out way out past cathode poisoning. Tomer's "Getting the most out of vacuum tubes" is fun reading, as is Acheson's papers on tube life factors.

[PRR]

> drop the heater voltage seriously low

Merlin, what is the vertical scale on these plots?

A full-heat 12AX7 should be ~~2.1mA at zero bias and 100Vp. http://www.mif.pg.gda.pl/homepages/frank/sheets/093/1/12AX7.pdf -page 3

If your "10mA" is per-division, you show (for your 9.7V-heat graph) 55mA, which seems impossible.

If it is "10mA" top-to-bottom (for 8 divisions, heavy math) it gives 6.9mA, which is still far above the boogie-value of 2.1mA.
____________________________

Thinking deeper..... Under-heating a cathode will flat-top the curves. We always see them rise up from zero in a graceful curve (3/2) and apparently go up to infinity. An under-heated metallic cathode will go up so far and then level off fairly abruptly. Under-heated oxide cathodes will quit rising along ^3/2 and tend to level out but the transition is very broad. And in practice (assuming anything like decent heat) we never get anywhere near this zone. But if we do, life may be short. Short-term, the mono-atomic surface of Barium is stripped. Long-term, the low temp does not bring-up fresh Barium from inside the bulk oxide

Your curves show no strong sign of flat-topping. Maybe a wee bit at 6.3V.

But maybe more important-- all old curves for 12AX7 show a "reverse bend" for zero-bias and 0-25Vp 0-0.75mA Ik. While we don't often operate in this area, the "boost" it gives to the zero-bias line in the higher V and I area defines the plate-bottoming and clipping action of Large-signal 12AX7 amplifiers.
________________________

> Acheson's papers on tube life factors.

I found Acheson's papers for the IEEE. His papers in the Sylvania Technologist are hard to access quickly. The IEEE papers speak in generalities; not odd for a maker's employee speaking to a wide audience.

But I still find it hard to worry about the life of one, or a dozen, tubes, in light of Acheson's remark: "{1956} ....not only are 25-tube television sets common, but also that 1500-tube payroll and accounting machines are common in modern business. The modern bomber may use about 3000 tubes...". Those 25-tube TVs used the cheapest bottles on the market and ran for years at a time.

The problem of "vintage" tubes needs special concern. The best (as well as some of the worst) tubes were made 1945-1965 in America, England, and Europe. These tubes in like-new condition command high prices today.

However more than any heater or plate ratings, IMHO what kills tubes in portable service is SHOCK. Thowing gear in the dumpster can break heaters. That long and bumpy road to Lodi can shake-loose bits of oxide, or crack welds which then release gas. If I had to consult on "reliable" tube road-gear, I'd urge a LOT of shock protection. (Even with the understanding that part of "tube sound" is mechanical feedback from speakers to tube internals.)

[PRR]

Here is an extract from Bell Labs' work on tubes for the trans-atlantic cable, using 306 tubes way-down-deep under the sea where they "must not" fail for expected service-life of a decade or more. (In fact the cable ran failure-free for nearly 20 years, by which time additional cables and capacity made it not worthwhile to keep running.)

http://i.imgur.com/UtTXItF.gif

The "life optimum" temperature is somewhat below conventional radio-tube degrees, but they note "serious instabilities" when temperature is a little lower.

[PRR]

A paper which simplifies tube error rates to two factors: heater voltage and glass temperature.

PREDICTION OF TUBE FAILURE RATE VARIATIONS
M. P. Feyerherm
Radio Corporation of America, RCA
Reliability and Quality Control, IRE Transactions on  (Volume:PGRQC-9 )
Jan. 1957

Tube guitar amps run their 12AX7s far-far below the glass temperature ratings (over-stuffed aircraft gear in thin air can run much much hotter) so this method becomes just heater voltage.

While over-voltage can be estimated as a rising exponential, UNDER-voltage is good only down to a point and "Below this, the failure rates often increase, as determined from actual cases cited in published information."

http://i.imgur.com/kdLWLid.gif

[Jdansti]

I don't worry about tube failure. RadioShack will replace my bad tubes for free!

[PRR]

Application of statistical techniques to electron tubes for use in a 4,000-mile transmission system
Van Haste, W. ; Bell Telephone Laboratories, Inc., Allentown, Pa.
American Institute of Electrical Engineers, Part I: Communication and Electronics, Transactions of the  (Volume:75 ,  Issue: 1 ) March 1956

"In two sections the line current was reduced to 95 per cent of the normal operating value and was left at normal in the other two sections. The effect was operation of half of the tubes at reduced heater power and the remainder at normal heater power. The experiment was balanced to permit statistical treatment of the data from coaxials and power sections. While about 15-per-cent improvement in average tube life was noted at the reduced power, this was not considered sufficient to warrant a change and the system continues to operate at the original power level."

This is probably like 5.5V on a 6.3V nominal heater. (Heater current does not drop as fast as voltage.)

[PRR]

Such a great picture.



Running "117V" TV sets on 130V (11% high) more than doubled the failure rate. (8.3 power of Voltage.)

Note though that this is not just heater stress. Many parts of a TV are stressed by voltage. Almost all parts are stressed by heat, and 130V/117V in a simple resistor is 23% more Watts of heat in the box. 480V on a 450V cap sure will cause early death. And unlike steel beams or back-porches, electronics is often designed and sold with "no" (unity) safety margin.