Leaving the heater of a vacuum tube on all the time vs cycling it
A heater's failure mode is typically a stress-related fracture of the tungsten wire or at a weld point and usually occurs after many thermal on/off cycles. One way to mitigate this of course is not to turn off the heaters at all (one of your options). Another is to employ a negative temperature coefficient (NTC) device such as a thermistor in the power supply servicing the heaters. to allow the heaters to reach operating temperature more gradually.
When the ENIAC computer was built in 1946 using over 17,000 vacuum tubes, the failure rate was initially several tubes a day. Of course they were already on all the time. They derated the voltage (and current) going to the heaters and reduced the failure rate to one tube every two days (longest time recorded without failure was five days).
Leaving tubes on all the time can accelerate failures which occur over long periods of time (thousands of hours of operation). Cathode depletion is the loss of emission after thousands of hours of normal use, as it is poisoned by atoms from other elements in the tube. However, according to page 34 of the 1960's era book getting the most out of Vacuum Tubes, this is fairly rare since by the time the cathode has lost its emission, the tube is pretty much dead for other reasons.
This same book, on page 14, makes another suggestion re keeping the heaters on; during standby, reduce their voltage to half instead of either leaving them on full voltage or turning them off.
I would personally probably provide a VERY soft start heater powering circuit and just possible an (almost unheard of) soft stop circuit (ie ramp up and down heater current over some seconds to minimise thermal shock and inrush current. You could even use a "maximum current circuit" to ensure Imax was hardly greater than I_operating_warm at any time.
Note that while general advice is that lowering Cathode temperature and reducing operating time increases lifetimes. one reference below, which very loudly claims to be authoritative, makes some radically unconventional claims. I'd be wary of taking his claim at face value unresearched - but also wary of not doing so.
TCrosley notes that
"Cathode depletion is the loss of emission after thousands of hours of normal use"
However, 7 hours a day = 2555 hours per year and 24 hours a day = 8765 hours per year, so you have "thousands of hours" per year in both cases but 24/7 = 3.4+ times as many thousands if operated for 24 hours a day rather than 7.
Extending vacuum tube life targeted at thoriated tungsten directly heated transmitter tubes but has some good general advice.
- They claim a significant lifetime gain by reducing cathode temperature.
Tungsten Filament Life under Constant‐Current Heating -1969
Abstract only of for $ article BUT notes -
- The life of the wire under constant‐current conditions is shown to be substantially less than that under constant‐temperature conditions.
Wikipedia - Vacuum tube says a number of useful related things, but of special note is
Tubes on standby for long periods, with heater voltage applied, may develop high cathode interface resistance and display poor emission characteristics. This effect occurred especially in pulse and digital circuits, where tubes had no plate current flowing for extended times. Tubes designed specifically for this mode of operation were made.
Cathode depletion is the loss of emission after thousands of hours of normal use. ...
BUT this authorative SOUNDING page contradicts advice from a number of other sources Vacuum tubes and vaccum tube failures
The big enemy of high gain power grid tubes (or valves) using metal oxide cathodes is grid current and excessive cathode current, or low cathode temperatures.
The ARRL, as careful as they sometimes try to be, has published more than a few incorrect articles about amplifier and tube life.
A thoriated tungsten filament tube can be run "hard", to the point of complete filament-cathode emission saturation, and the life will be no shorter or longer than when run easy,
Provided the elements being bombarded by electrons or the envelope does not overheat and suffer permanent thermal damage. We can lower filament voltage in a thoriated tungsten tube, and provided it doesn't get contaminated from prolonged operation at excessively low voltage, all that happens is peak clipping.
A metal oxide cathode tube can quickly suffer damage if operated that way. This is why they sometimes have to start on a timer that prevents current before the cathode is fully heated, which sometimes can be the warm-up time of a rectifier tube and other tubes in the system! Lower the filament voltage in a metal oxide cathode tube too low, and you can ruin it in seconds!
Within emission and thermal damage limits, tubes basically do not wear any faster or slower if just idled or operated. It isn't like a mechanical engine, where high RPM operation greatly increases wear by mechanically loading internal parts increasing friction. As a matter of fact, too cold is often much worse than hot.