Simple soft start for tube heater

[rankot]

I was thinking about this - because I have read some articles about tube longevity etc. -  is this a good way to slow start a tube heater? Putting 1000 or 2200 uF capacitor in series with heater will provide both slow start/stop, as this oscilloscope graph shows. But is it good enough? Or not necessary at all? 

[Groovenut]

IIRC tube filaments when cold are high resistance and get lower in resistance as they heat up and pass current. Most tube filament circuit are AC and do not include a cap downstream from the tubes. This means the inrush current is limited by the high resistance of the cold filaments. Thus I don't know that a soft start is really necessary. Could you provide a link to the article(s) that were suggesting a soft filament start?

Also you may want to use an AC voltage source as V1 in you sim. It will provide slightly different results

[R.G.]

Actually, tube filaments are lower resistance when cold. They do pull a larger current at turn on.

However, this tends to be a non-problem with small signal tubes like 12AX7s and other non-power tubes. Especially with an auxillary resistance in series, they're going to last a long time - as in many years. There are various ways to improve this, but they all involve more circuitry and will slow down the tube's warm-up time.

If this is for a pedal, I would ignore it. If it's for a guitar amp with several tubes, I can suggest some managed-start circuits.

[rankot]

R.G. is right - resistance increase with temperature, so the same is valid for filament's resistance. That means they are sensible to inrush currents! There are many articles about that on the Net, just Google "tube heater soft start". I had in mind to use this for tube based pedals and with DC powered heater of course, but if it is not something to be concerned of for low power tubes, I will not worry.

[Mark Hammer]

Steve Daniels sent me a tube-based overdrive to test out, and I was surprised to see how quickly the 12AX7 warmed up.  When I took it apart to see if there was available space for any mods worth doing, I was rather startled to see the presence of 2 small red LEDs under the horizontally-mounted tube, that shone through the tube as soon as power was supplied.

Yes, it's true: LEDs turn on faster than tubes do! 

[rankot]

Yes, it's true: LEDs turn on faster than tubes do! 

I've designed my last Thiagosarurus Ranx that way - with 3mm orange led below 6J6 tube. Looks nice!

[Groovenut]

Well hell....

Goes to show you should never trust your memory before coffee or as you age past 50.....dammit....

[PRR]

> articles about tube longevity

You can read about anything.

How many burned-out heaters have you ever seen?

Not counting 110V series-string radio/TV, I've seen less than one a decade.

Tubes fail because the guts shake apart. Oxide flakes, grid wires distort, etc. This is inherent to how tubes are made; worse because tubes today are CHEAPER (in real money) than any time in history. OTOH tube heaters are mature incandescent lamp technology, mass produced until recently, run far cooler than a lamp, and have been very reliable. You will replace a heap of hissy rattly or shorted tubes before you find a burnt heater.

If you "must" be clever: re-think the whole heater thing. Low-volt AC heat was convenient in 1930. Good DC heat throws less hum, costs more, but the cost is declining. 6V heat is awkward with 0.6V rectifiers because losses are high. Study the Master. Peavey often stacks heaters to 24V, sometimes derived from a 30V DC supply with enough resistance to limit cold-surge and to improve filtering. Switch-mode DC supplies have got incredibly cheap, though so cheap you have to allow for occasional replacement (with probably a different shape/size).

[thermionix]

^ I've had to repair a lot of DC filament supplies, but I reckon some are designed better than others.  And they do reduce hum well.

Might be a better idea to have heaters come on immediately. and slow-start the B+ voltage.  5AR4 can be good for this in amps.  I have read contradictory things about cathode stripping, maybe it's not a concern in our world.  But if you fire up 5AR4-equipped amp without using a standby switch, the tubes and filter caps get a slow, steady buildup of DC.  It seems friendlier to the components, but maybe just in my mind.

[wavley]

^ I've had to repair a lot of DC filament supplies, but I reckon some are designed better than others.  And they do reduce hum well.

Might be a better idea to have heaters come on immediately. and slow-start the B+ voltage.  5AR4 can be good for this in amps.  I have read contradictory things about cathode stripping, maybe it's not a concern in our world.  But if you fire up 5AR4-equipped amp without using a standby switch, the tubes and filter caps get a slow, steady buildup of DC.  It seems friendlier to the components, but maybe just in my mind.

Yeah, I see a lot of stuff go bad in heater supplies that make me want to keep them simple as possible.  Balancing resistors and balance pots on AC also burn up and make the amp hum.  I personally like to just use a center tap transformer, twist and dress the wires really well and never worry about my heaters again, because the tube will get crappy well before the heater does.  If you really want to get crazy, you can add 1uF 100v film snubber caps across then heaters and not have to worry about hum or crap in your heater supply going bad.

I fix a lot of amps, the easiest/cheapest ones to fix are the ones that keep sand as far away from tubes as possible, and to add to that, the most reliable amps I've found are the ones that keep PCBs as far away from tubes as possible.  (unlike the poor Mesa I just fixed that has carbon burns all over the place, not that tube sockets don't get those, but sockets are an under $10 part)

[thermionix]

^I'm with ya 100%.  I won't even accept Boogies for repair.  And Peaveys are a case-by-case thing.

[Transmogrifox]

I might be wrong, but I didn't ever think as soft start for the heaters as concern for the heater filament but for the tube.

Rapid temperature changes cause materials to expand and contract at uneven rates.  I always imagined rapid uneven heating in a tube could compromise the vacuum seal allowing some air into the tube during this time of stress, or eventually a stress failure creates a permanent leak.

Maybe I'm wrong because this is just my imagination, but I do know that the "blue glow" (other than Crate's blue LED's) is indicative of the presence of air in the tube.  It gets there by leaking.  It stands to reason the vacuum seal can become more and more leaky due to rapid and uneven thermal expansion stress.

I haven't looked at a comparison of metal and glass thermal expansion rates nor the stresses induced by heating and cooling so I really don't know whether this is a "real" failure mechanism.

Anyway, food for thought and some foolishness to egg on the guys who know about these things.

[Transmogrifox]

I fix a lot of amps, the easiest/cheapest ones to fix are the ones that keep sand as far away from tubes as possible,

So those tubes are made of what? Sapphires? Faux diamonds? Acrylic?

I just had to poke at that.  Yes they are better when you keep the silicon valley sand far away from the St. Petersburg and Saratov sand.

[wavley]

I fix a lot of amps, the easiest/cheapest ones to fix are the ones that keep sand as far away from tubes as possible,

So those tubes are made of what? Sapphires? Faux diamonds? Acrylic?

I just had to poke at that.  Yes they are better when you keep the silicon valley sand far away from the St. Petersburg and Saratov sand.

I guess I never think of the sand being integral to the function of the tube like it is for other semiconductors given that tube packages can be ceramic, quartz, or metal. 

But you got me there, there is sand comprising the envelope of most tubes.

[R.G.]

Tubes are so very fertile a ground for sideways ideas.   

Soft start for filament/heaters is a good idea. Incandescent light bulbs - which our USA government will not trust us to use properly any more - generally die at turn-on, not during operation, because the current surge overheats the smallest cross-sectional area of the filament, it fuses, forms an arc, and pops. Tube heaters are the same technology but run at much lower temperatures. When/if they fail, they fail because one section got too thin and went into runaway like a light bulb.

DC is not necessarily the best way to heat a tube filament, as any evaporation/erosion of the tube heater happens preferentially at the end that attracts the most positive gas molecules. DC bias puts this all at one end. AC is actually better for spreading out the heater filament wear. Which doesn't matter much for low power receiving tubes because these usually have lives in the range of 20K power on hours anyway. They tend to die from low emission or shock/microphonics first, at least in guitar amp use. In pedals?? Who knows. Stomping may be even worst than living near a speaker at 120ds SPL.

Voltages on cold plates don't cause cathode stripping below about 450-500V, so it's largely a myth in guitar amps, but is real in really high power stuff - which we don't use. It's non-existent on low voltage stuff, like pedals.

Applying sudden plate voltage surges to cold plates can be a problem to the high voltage rectifiers in guitar amps. It can damage the rectifier and/or filter caps. But it can be a worse idea to HOT tubes, as they suddenly want to conduct a lot. Some Marshalls turned off the bias supply to output tubes in standby, so when flipped out of standby, the output tubes were already hot and conducting, but un-biased and immediately conducted like a diode. Nasty things happened. AC 30s have such a bad setup for standby that the insider wisdom for them is to not use standby at all.

You can use a power MOSFET in a current-limiter setup to choke off current a level just above highest-operating peaks to limit both power on and off-standby surges. There's an article on it at geofex.

[amptramp]

You could use an LM317 as a current limiter set to 150 mA with an 8.33 ohm resistor from output to adjust pin and the current limiter would be in series with the power supply.  It would entirely prevent startup surge current.

[R.G.]

That works, but it may take a long time for the tube heaters to come up to temp and the tube to operate normally.

The 12A*7 series is part of the "modern" controlled-warmup time tube series where tube technology was mature enough for the designers to try to work out how to have assorted tubes of different sizes and types all come up in a nominal 15 seconds or so. That involved juggling the cold filament resistance, wiring size, and coiling so the heating was just so.

That process involved the filament being driven by a voltage (ish) source, and having V²/r be big at first, then tapering off as R rises. The biggest dissipation that gets the temperature up comes very early, and gets the first, non-operating part of the heat-up over with quickly. Running it from constant current heats as I²*R, but with I being constant, the heating increases linearly with R, so the "snowballing" effect of R increasing happens more slowly at first, and the initial heating takes longer.

Neither of these is a problem, but the user perception of taking a long time to heat up can be bad.