A few years back I came up with what I called the Immortal Amplifier: replace everything that wears out with things that have very, very long life spans. The number one target was electrolytics. Parts of this suite of mods were listed in my article in Premier Guitar back in 2008: https://www.premierguitar.com/the-immor ... lifier-mod
Very early tube gear (up to about the mid 1950s) generally used film capacitors for the filter caps. They were heavy, expensive, and lasted semi-forever. Consumer goods went to electrolytics when they got cheap enough, because they were small and light weight. Today, we can make film caps that are only a few times bigger than twist-lock cans, and not dramatically more expensive. Motor-run film caps exist that can be stuffed or mounted on a guitar amp chassis instead of electros, and these will last about as long as the chassis itself. An amp built with motor run caps for the rectifier and dropper string caps would never need those replaced, even for an expected 60-100 year life.
You can't make tubes effectively immortal: that's why they're in sockets, per the engineer's mantra - if you can't make it last forever, make it easy to replace. They're just going to wear out someday. In a guitar amp, the real baddies are the power tubes, both output and rectifier. These can fail in a way that takes out other stuff. You really don't want a failed rectifier or output tube to take out a transformer, as the PT and OT are usually the second and third most costly parts to replace.
So for a tube rectifier rig, it makes sense to put in back-up diodes in series with the rectifier tube. The tube can then fail shorted and not kill the PT, OT, first filter cap, or output tubes. The DC voltage just rises a bit. You might not even notice. I've recently done a batch of work on making diodes zeners and resistors curve-matched to the V-I characteristics of a tube rectifier, so a tube rectifier might not even be needed, even for those who maintain they can hear solid state diodes. A curve-matched rectifier setup comes really, really, really close to the actual tube.
Transformers are another biggie. They normally don't wear out, but other faults kill them. The long-term cure for transformer death is to (1) add MOV clamps to keep transients from punching through the wire insulation and (2) fuse each separate winding so that overcurrents can't get the wires hot enough to decompose the wire insulation in hot spots.
The stacked diodes from ground to output tube plates is one form of transient protection. This dodge uses the diodes in avalanche mode so that the leakage inductance flyback on sudden current stoppages for the opposite-side tube can't arc the insulation inside the transformer. This works mostly, but is not all that predictable. A 3kV string of TVSS diodes across the OT primary would probably be more predictable.
MOVs on the AC voltage in would protect against AC line transients, and are small and cheap. It is seductive to think that a single AC mains fuse will protect the PT. Sadly, this is not so. That fuse is to keep an already failed amplifier from burning down the building and is required to pass most safety standards. A fuse in each side of the high voltage winding, one in each half of a center tapped heater winding, one in the rectifier heater winding if it's a tube rectifier, and so on will limit the current for each winding, so that the fuse trip can be set for that specific winding. On the OT you can probably get away with a single one in the CT as that's the only place current comes from.
The list goes on: think about how amplifiers you have known and loved have died. If a simple part substitution can prevent a class of deaths, think about whether the time and money to put in a protection will remove that class of amplifier deaths. The power supply and power output stages are the most fertile places to look, as this is where most non-tube deaths happen, along with the known wear-out of electrolytics.
This isn't by any means all-encompassing. It's more of an invitation to think.
Every time I have posted this, someone notes that adding more stuff can make reliability worse by adding more stuff to break. This is true. But if the projected lifetime of the added stuff and projected failure rate and ease of repair makes it quicker, cheaper, and simpler to repair added-stuff failures than what it protects, this can still be a winner. The biggie seems to be all those fuses; fuses DO have a wear-out mechanism, and contact oxidation, etc. can make them less reliable. True - so use good fuse holders and fuses. Same goes for any protective equipment; it doesn't make sense to buy only the cheapest fire extinguishers.
