3

Some valves on the market today are rare and expensive. GEC Marconi KT66's retail for about £300 /pair today for example. Some others clearly show a bright flash as their heaters initially power up which is worrying when you see it for the first time.

One concern I have is that the traditional constant current approach may not be sufficient to initially raise the temperature of the heater element sufficiently. I understand that some over current is necessary to begin the thermal resistance increase. Without over current, the heater will just sit there and may not get sufficiently hot. I'm not sure about this statement though.

One modern aspect of this question is that I'm not trying to be totally authentic with all of the circuitry. Only the valves. I'm prepared to consider any modern form of protection in what you might call a hybrid design. An anachronistic design featuring relays or micro controllers would be fine as well as DC drive.

What modern or traditional means can be used to reduce the initial current /thermal surge on a valve heater to extend it's life?

NOTE. A single small valve might require 300mA @ 6.3V RMS.

Paul Uszak
  • 7,517
  • 5
  • 40
  • 76
  • 1
    What "traditional constant current approach"? Traditionally, tube filaments are driven from a low-voltage transformer tap (such as 6.3 volts) and this is best modeled as a constant voltage. – WhatRoughBeast Jan 12 '17 at 04:17
  • tradionally a NTC thermistor was used, they just looked like a black slug with wires attached to each end. they ran so hot that no attempt was made to paint them. – Jasen Слава Україні Jan 12 '17 at 04:41
  • @WhatRoughBeast Sorry, I was referring to the traditional method of a constant current supply when people discuss heater protection, not what the ancients did. My bad. – Paul Uszak Jan 12 '17 at 12:42

4 Answers4

2

In order to limit the rushing current in the cold heaters + the current in the empty capacitors, use a 33 ohms inrush current limiter like a negative thermistor (NTC) just made for that purpose in the primary circuit of your power supply transformer. Here is an interesting article about that.

When it’s cold (at start) the NTC presents maximum impedance to the rushing current. After few seconds, the NTC heats up and its impedance lowers down to a ohm or so.

greg
  • 443
  • 3
  • 6
1

Sure an NTC could provide some benefit if it was sized correctly. Greg has stated this.

An old school approach is to use a separate filament transformer which has just one secondary winding of usually 6.3 VAC. This transformer is rated for the job. It is not much bigger than it has to be. A grossley oversize transformer makes this cold inrush current worse. On most valve systems the anodes and screens take much more power than the filaments. This means that the traditional multi secondary transformer makes inrush worse.

What was sometimes done was making the leakage inductance greater to limit the cold current even more. When you encounter such a high leakage filament transformer the Volts does depend more on the loading so you should use them within their spec. Of course you would not overload it but if you wanted to run it at say 30% capacity you may find that the filament volts are too high. The traditional approach here is to use a series resister to bring the filament volts into spec. Some filament transformers had primary tappings to achieve this. Even if you need a resistor you wont be wasting as much power as the LM317.

If this high leakage filament transformer is unobtainium and the NTC can’t do enough then bypassing a primary current limiting component with a relay will work .Another possible modern solution is to employ 2 back to back low on resistance mosfets and slowly ramp up the gate volts at power up.

greg
  • 443
  • 3
  • 6
Autistic
  • 14,616
  • 2
  • 28
  • 65
1

First, in a modern circuit it's better to run the heaters from a DC supply. Even AC heaters always coupled a little of the AC to the signal.

With DC, it's easy to apply PWM at startup. It doesn't need to be anything fancy. A open loop ramp is fine. Just ramping the duty cycle linearly from 0 to 100% over a couple of seconds or so is good enough.

If the 6.3 V (or whatever your heaters use) DC supply is a switcher, then it is already essentially run from PWM. Some integrated chips have a soft-start feature, or you can use heavy capacitive coupling from the output to the feedback input of a integrated IC. That causes the supply to ramp up slowly.

If you are making your own supply with a spare microcontroller PWM output, then it's even easier. Just ramp the maximum allowed duty cycle up slowly, regardless of what the control algorithm says it wants.

All these methods have the advantage of basically not being in the way during normal operation.

Olin Lathrop
  • 313,258
  • 36
  • 434
  • 925
  • Is my concern regarding the need for some over current to begin the warming up process/resistance increase unfounded then? There are some on line stories of people's switchers not being adequate because the current limiting remains active constantly and the heaters do not warm up. They say this is caused by the initial surge current. It's so difficult to get objective science from the valve amplifier forums! – Paul Uszak Jan 12 '17 at 12:55
  • @Paul: If you're worried about that, then don't bother with current limiting. As I said, open loop ramping up the maximum allowed PWM duty cycle should handle it nicely. This is what used to be done for incandescent bulbs for long life. – Olin Lathrop Jan 12 '17 at 13:10
0

There is no requirement to run the heaters on 6.3 V AC, so firstly I'd suggest you convert to run them on DC. You don't need to run them constant current either, though I have seen that done. A simple voltage output DC supply with a long risetime (say 5 seconds or so) allows the heaters to come up to temperature smoothly.

You could use an LM317 with a circuit like this (from the datasheet) which will ramp up from 1.25 V to the final voltage set by the R1, R2 divider.

enter image description here

R2 set to 910 + 56 Ohms (966 Ohms) would give you 6.3 VDC instead of the 15 V shown.

Jack Creasey
  • 21,729
  • 2
  • 15
  • 29
  • Startup may be fine, but problems likely come after - 300mA per typical 6.3v tube filament will add up fast when using a lossy regulator like this, so even with an ideal input voltage to it you'll need a substantial heatsink. – Chris Stratton Jan 12 '17 at 06:27
  • 1
    I shouldn't have used a "small" tube as a current sample - the mentioned KT66's take 1.27 amps each, in theory an LM317 can handle one if adequately cooled, but the large dropout makes it really not look like a good choice. – Chris Stratton Jan 12 '17 at 06:32
  • @ChrisStratton. I totally agree with you, switching supplies would be much better. Though somewhat less trivial to design. However I would suggest you want no more than 2 heaters per supply with a risetime of only a few seconds. It's years since I worked on valve amplifiers but the heater resistance cold/hot variation can be a 10:1 ratio. However just like conventional incandescent lighting the aim is to allow the heater resistance to rise not to it's final value, but to some intermediate that reduces the surge. – Jack Creasey Jan 12 '17 at 06:53
  • @ChrisStratton. //cont'd// And given that a multiple valve amplifier may well have 20+ watts of heaters running, the extra dissipation in the linear LM317 seemed acceptable. The LM317 could of course have a boost transistor around it to increase current capability, though I still agree that a switching solution would be best. – Jack Creasey Jan 12 '17 at 06:56