[Image] Evacuated Envelopes [Image] The On-Line Valve Amp Magazine DATE: SEPTEMBER 1995 --------------------------------------------------------------------------- ITS A HOT LIFE FOR TUBES As you should know all too well, tubes have a finite life span ( well thats what everyone says! ) Here are some extracts from a thread on R.A.T HEATING Hugh Gibbons All in all, most wearout mechanisms are directly related to heating. They can be reduced by lowering the filament voltage to the minimum required for proper operation. Correct but remember... Axel Rohde Underheating is as bad as overheating!!! The golden rule for heating is: the heater supply voltage should be the specified voltage +-5% FLASH TURN ONs Don Borowski I have found the those tubes in which the heater flashes on brightly when the power is first applied are often short lived. They would easily pass the 72 hour test, but might not make it 72 power-on cycles. Then again, others have reported no problem with such tubes. Mark Garvin I believe that unfortunately includes a lot of Telefunkens I've seen! Incredibly bright power-on flash. Andy Moss You must be thinking of tubes like the 7025 (12AX7A). They have a series filament that has a piece of metal with opposite thermal properties to the heater. It's there to limit the initial turn-on current to the filament. : Then again, others have reported no problem with such tubes. {Don} Yes. Case in point my radio at work is an old RCA ca.1960 that has these kind of tubes in it. The tubes are the original ones and the radio is working fine. That's 35 years of trouble-free service. Try that with a Sony. IN PRACTICE Fred W. Bach Back in the days when I was designing and building amps, I liked to push tubes for as much as I could get out of them. So I learned a lot about heat failure. You can drive a tube a LOT harder and raise the voltages a lot if you keep the glass and the socket cool. If the glass of the tube gets too hot, it can outgas more. Air diffusion across it, while small, increases. The pins on the tubes oxidize or corrode more the hotter they are, and so do the tube sockets on the chassis themselves. It is not uncommon for the solder on the joints to crystalize and crumble, and then require re-soldering. I found that cooling the chassis (around the sockets) AND the output tubes themselves with forced air was most useful. I could run a pair of good 6L6GC's (normally good for 60 watts CW output power) at very high voltages (I seem to remember 600 volts) on the plates. I mounted the tubes on another chassis with tiny holes around the sockets and air forced up from underneath. I could run this all afternoon. Without the cooling, the tubes would have failed. Also, it is important to add little resistors in series with the screen grid and add some cathode resistance to prevent runaway and nasty RF parasitic oscillations which can further heat the tubes and shorten their life. With these precautions and the extra chassis and envelope foirced-air cooling, I fired up a pair of 6l6GC's that would *brightly* light a 100-watt light bulb off the 70-volt line tap on the output transformer while playing a Hank Snow record. I knew it was clean power because I listened across a speaker hooked in series with the light bulb (I had a power rheostat across the speaker to shunt away most of the current from the speaker). I figure the amp was putting out about 120 watts on a single pair of 6L6GC's. Not bad. Not to be forgotten in amps where the driver stage is capacitively coupled to the output tubes is the possibility of leaky coupling capacitors. These leaky caps drain off negative grid-bias, causing the tubes to overheat. In mild cases the effect is not noticeable but the tubes overheat and die early. You must make sure that the grid bias is solid. Be warned, this is very tricky to measure with ordinary meters since they load the circuit so badly (grid circuits usually have a very high impedance). Reading the voltage across the cathode resistors (if they're separate) can give an idea by inferring the grid voltage from the cathode current. Swapping the output tubes with each other can tell whether anomalies are due to the tube or a leaky coupling capacitor or an otherwise faulty grid-bias circuit. Shielding Someone mentioned that putting a shield around a tube can increase its temperature. This is, of course, true. An increased envelope temperature leads to an increased plate temperature which leads to a temperature increase in the whole tube. What I have heard of in some cases is that you can lower the filament voltage a touch on such tubes to compensate. Another trick I've heard of but never used: Many tubes develop cathode-to-filament shorts. So, for tubes heated with a floating AC filament, since the filament is hotter than the cathode, biasing the filament supply just slightly (a fraction of a volt) positive to the cathode should help maintain the insulation between the cathode and the filament. I have never tested this theory. Any comments? (It would be a problem for tubes with cathode resistors, the filament winding would have to float at slightly above the cathode voltage. Sort of tough to do if the cathode is 40 volts above ground!) Orientation One more thing about tubes and lifetime: some tubes like to be mounted in a certain orientation with respect to gravity. This is said to be helpful for preventing cathode-to-filament or grid-to filament shorts if the cathode or filament flakes (they do), or element misalignment should they get cherry-red like plates often do. And remember, hot air rises, so the air space above the tubes must be adequate. DONT BLOW A FILAMENT Dave Cigna All this talk about heater lifetime confuses me since I have rarely encountered open heaters. The catastrophic failures that I've seen are more often a short (cathode to plate), but even those are rare. It seems that it's far more common for tubes to just start sounding bad. The lose their luster or maybe become microphonic long before they blow up. So, I don't mean to be knocking you guys. Must be that I'm missing something, but I don't think I'll be building any fancy circuits to save those precious heaters when they've never been a problem for me. What gives? Don Borowski I think you have hit the nail on the head. Heater problems are so far down the list that they really don't matter that much. About the only piece of equipment where heater failure is a real problem is the "All-American Five", the common 5-tube radio of the post WWII period. The heaters are all in series, and are connected directly to a low impedance power source (the power line). The turn-on surge current will eventially kill the two highest voltage heater tubes (the rectifier and audio output). In a transformer power device, there is enough impedance in the heater winding to limit the turn-on current. This is easily confirmed by connecting a voltmeter and watching the heater voltage. Series-string-heater equipment with a heater voltage dropping resistor is fine too. Even an "All- American Five" radio can be backfitted with a thermistor to control the turn-on to good effect (or even just a resistor helps a lot, though the heater voltage will be a bit low). Martin Ackroyd, a net-aquaintance of mine in the UK, states that he has never seen a blown heater on a post-WWII series-string-heater radio because they all had thermistors to control turn-on surge currents. Summary: On transformer powered equipment, if the operating heater voltage is OK, be happy and don't worry. A BIT OF REGULATION Fred W. Bach Back in the 60's I was the repairman for a Fender dealership in Edmonton, Alberta. I resoldered a lot of Fender Bassman amps (and I fixed the cigarette burns and the beer/drink stains). One day we got an old Silvertone in on trade-in. Two channels, vibrato and reverb. It had a lot of problems, so I decided to rebuild the amp (to some extent). Besides the grid bias, I decided to add a little bit of cathode bias on the twin 6l6GC output stage (OK, so I effectively robbed a few volts from the plate-cathode voltage. So what.). Cathode bias is very helpful in stabilizing tubes and preventing runaway conditions. It so happens that if I connected their cathodes through a couple of small resistors to a single hefty 12-volt zener diode stud-mounted on the chassis and bypassed by a large electrolytic, that I could use this 12 volts DC to power the front-end tubes with nice quiet regulated DC !!! What a great idea! It got rid of considerable hum, and it treats the input tube filaments very gently so they're going to last a nearly forever! I liked the amp so much (why not?, I built much of it) that I bought it from the boss (and I needed a new amp anyway). I used that sucker all over Alberta and a lot of BC. I haven't replaced a tube in that amp for about 20 years, although I have blown up a filter cap. I still use the amp a couple times a year and it still works great! Eventually I ditched the separate Silvertone speakers & cabinet, although I used them for a while in Alberta and a little out here in BC. But the speakers were too weak and the cabinet too flimsy, so I built my own cabinets using some $5.00 surplus speakers from Admiral (clearance of old stock). Nice bass drivers, (they probably didn't know what they were getting rid of) but I had to add horns for the high end. We pumped organ, electric piano, guitar, and voice through the amp (not all at the same time....) One thing I can say about the Silvertone, it had a fairly high level of quiescent tube hiss (even with the volume controls down). I suppose I should have re-designed that circuit as well, but I never really got around to it. Incidentally, about THE BEST little short course I ever saw on vacuum tube electronics, especially with a bent towards audio, was the front section of the RCA Receiving Tube Manual. There are a couple of really great amplifier designs in that book. I think that book is a MUST for all tube-audio fanatics! SPACE Heat is the primary consideration, both in tube life and component life. Other things being equal, a 25-40 watt amplifier and preamp on a 19 X 12 X 3 inch chassis will last a lot longer than one on a 12 X 8 X 1 inch. FRY YOUR TUBES ( in oil? ) Dirk Wright Did you ever try the old ham trick of taking a metal 6L6 and imersing it in oil to get more power out of it? Chris Christensen I believe that to make this method effective you should use a finned heatsink on the tube. The problem with this method (as I see it) is that ass soon as the oil heats up you will have lost it as a suitable cooling media. One would have to cool the oil. Dirk Wright Well, perhaps you're taking this a little too seriously. I would never do this, and current hams probably don't either. I think it was sometimes done during the 40's and 50's when transmitting tubes were very expensive. But, if we want to stretch this a bit, with our tongue firmly planted in our cheek, we could of course provide a recirculation system for the oil, have a suitable heat exchanger hooked up to a nice 20 ton air conditioning unit, and be done with it. But, I digress... David S.A. Stine You might have your tongue in your cheek, but the engineers at places like Varian/Eimac do not. If you examine the cooling and power requirements for modern metal-ceramic power tubes (triode, tetrodes and pentodes) in the 800W to 150kW range, you'll see that massive cooling is absolutely critical to tube life. As in, if you don't start the blowers or the water before you apply the filament power, you will lose the tube by the metal-ceramic seal failing. Oil is infrequently used for cooling tubes; water and air are the two most dominate cooling media. There are a few tubes which use vapor-phase transition cooling (ie, the tube will boil the cooling water, which is then re-condensed in a closed loop system), but they are not in common amateur use. As for the 6L6 cooling issue: I, as a kid, ran a 6L6 inverted, with the base pulled off, in a coffee can of ice water, for a "betcha I can do it" class-C amp running 100W input, 750 volts on the plate worked just fine for CW. Not bad for a tube with 18 watts of plate dissapation as I recall. TUBES HAVE THE BEST LIFE Don Borowski I was digging though my 1943 Radio Amateur's Handbook over the weekend. There was an RCA ad in the back, talking about getting the best life out of transmitting tubes, which were in short supply during the war. Part of it is applicable to high power audio tubes. Here is the text (minus the part about mercury vapor rectifiers): Long life--not maximum output--is the keynot of transmitting tube operation today in many servies where, because of war restrictions, it may prove difficult to replace tubes that wear out. Since care in the use of tubes--even far beyond that what might ordinarily be considered necessary--should pay big dividens in longer life, the following suggestions are worthwhile: HIGH-VACUUM TYPES For tube types using pure tungsten filaments, a reduction of only 5% in the filament voltage doubles tube life. A reduction of 15% increases it almost tenfold! Decrease filament voltage to 80% of normal for standby periods of less than two hours. For longer periods, the tubes should be shut down. Care should always be taken in starting up tugsten filaments. Never should the filament current exceed, even momentarily, a value of more than 150% of normal. For types using thoriated-tungsten filaments and oxide-coated filaments, the filament may be operated on the low side--as much as 5% below normal voltage--if the loading is light. The filament voltage should be increased gradually to maintain output. Towrad the end of life, additional service may be obtained by operating the filament above its rated voltage. During standby periods of less than 15 minutes, the filament voltage may be decreased to 80% of normal to conserve life. For heater-cathode tube types, where some operating delay can be tolerated, it is good practice to dro the heater voltage as much as 20% during long or frequent stnady periods. For all types, reduce dissipation of grids and plates to a minimum to avoid overloading and, thus, obtian materially longer life. THE TRUTH / MINI FAQ R.G. Wearout mechanisms are * filament evaporation This is how ordinary light bulbs die. The hot filament actually evaporates some metal from its surface. Eventually this erodes the filament open. This is usually not the only death mechanism, as breakage gets the weakened filament first. Donald Borowski I don't think this is a significant problem, expect maybe in some of the old transmitting tubes which used thoriated tungstan directly heated cathodes. They indeed looked like light bulbs when they were running. Standard light bulbs are designed to run 750 to 1000 hours. Cutting the applied voltage in half to these bulbs will extend their life to beyond 1,000,000 hours. Even at that voltage, they are running much hotter than the heaters on tubes. Heaters do fail, but I think thermal shock is the big factor, not evaporation of the tungstan heater material. * filament breakage An eroded filament can break before it erodes in two. The key factor here is thermal shock at power up. When the filament is cold, it is a much lower resistance than when it is hot. This makes for a large current surge at power up if you don't somehow limit the surge. The rapid heating makes for rapid mechanical expansion, which can break the filament at any weak point. * cathode poisoning Residual gasses in the tube can get ionized and attracted to the electrically negative cathode. This can cause the cathode to stop emitting electrons by chemically reacting with the specialized barium and strontium oxides which coat the cathodes of most tubes. These oxides emit electrons at a much lower temperature than most materials, and as they are poisoned, the electron emission goes down until there is too little to support proper operation. * evolved gassing Leftover gasses in the glass and metal can be cooked out into the vacuum envelope by heating. Some materials can simply evaporate into the envelope. This eventually kills the tube by cathode poisoning. Overheating is a primary contributor. * shorts from overheating If you get the innards too hot, they can get soft and physically deform, changing characteristics of the tube, and eventually touching, shorting out the tube. You can actually melt spots in the plates if you try. I have seen a tube where the -glass- partially softened and started to give way as the tube died ( I think the innards were supporting an arc at the time - not normal). * cathode wearout The barium and strontium oxides that coat the cathode have two limits. One is that there is a maximum number of electrons they can emit without damage, and the other is they gradually degrade at high temperatures. Both lead to reduced emissions, and eventual uselessness. OK, what do we do? 1. Run filaments at or below rated voltage - never above. Signal tubes usually have eneough emission that at a slightly lower temperature they work with lower transconductance, but acceptably well. The life of the filament is inversely proportional to the 13th power (yep, that's right) of the applied voltage, so even a minor decrease lengthens life - a lot. Run signal tubes at -5% to -10% of rated voltage IF THE EQUIPMENT FUNCTIONS OK WITH THEM RUN IN THIS CONDITION. Some apps won't, some will. Don't go below -10% or you may not have enough emission to keep gas ions at bay and you'll you'll not only have low transconductance, they'll die early. Power tubes need all the emission they can get. Run these -*exactly*- at rated voltage. 2. Do not apply B+ before the filaments are hot. This ionizes any gas and lets the ions bombard the cathodes, poisoning them early. If the cathode is hot before the B+ comes on, the electron cloud around the cathode ameliorates the problem, giving a longer life. Preferably, run a sequential power up where the filaments get hot, then B+ is ramped up. 3. Limit surges in filament and B+ at turn on. Ramp them up gradually. 4. Keep the tubes as cool as you can. Overheating will eventually kill. use a fan run at a reduced voltage to make it quiet to cool the tubes. 5. Do not -*ever*- exceed the max cathode current spec for any tube, especially power output tubes. Even instantaneously. 6. Keep all element powers within specification, all the time. This specifically applies to screen grids, which can overheat and die like plates, except they are more fragile and can dissipate less. Keep screen grid current and power within the spec limits. Always. There was an article in the Sensible Sound a year or so ago with many of these concepts in it. The article indicated that 100,000 hour life was a reasonable goal. Standard hifi amps could use these concepts as is. Standard guitar amp usage will violate the current and power rules, and will require frequent re-tubing. But we knew that, right??? --------------------------------------------------------------------------- tubes@hillier.demon.co.uk © M.J.Hillier 1995