[Chas]

There is a failure mechanism at work when a tube radio is playing and reception is during an electrical storm.

Plate current via the audio output transformer is abruptly cut off and the resultant magnetic field in the core collapses. This induces a very high peak voltage of several cycles. Without any capacitor to absorb this spike the transformer will arc internally. When that occurs windings may still conduct but there will be a carbon path and the generation of acids, that, in time, will corrode the winding and open the transformer. Factoid: copper in an arc/flash expands 67k times...

Most radio designs provided for reduction in adjacent channel (10khz whistle or "tweet") with a filter just after the detector but also with a capacitor across the primary of the output transformer or from the output tube plate to B-. The spike is suppressed but the voltage value of that capacitor must be high enough to withstand that static generated spike. A thicker paper/foil condenser works as was installed OEM. In modern times, generic metalized caps cannot propagate this high frequency spike across the metalizing even if the cap is rated for 1kv. The metalizing disconnects permanently but the insulating film does not perforate. To the user, there is little difference (in audio) until the metalized cap looses almost all of its capacity. Then, the output transformer becomes vulnerable to static destruction. A occasional phenomena of a failing metallized cap is noise, random, very high frequency spikes as the disconnected area of metalizing connect and disconnect. Often heard is 'ticks" in the speaker. The noise can be captured easily with most scopes.

That said, simply increasing the working voltage of a replacement metalized (common) cap in the plate bypass is not a method as it is not the plastic that is failing. The metallized is ill suited in that circuit, its "plates" are just not thick enough. Choose a metal foil/poly cap that has a published dv/dt rating. The metal foil will not deteriorate and the pulse current will be absorbed without damage to the dielectric. In some radio designs this same cap provides a dual function of tone control, that cap should also be of the same type with a dv/dt rating. Some radio designs have both a transformer shunting cap and a cap(s) used for tone control. Can the tone caps be plain metalized? IMHO yes, as the pulse is being absorbed effectively by the foil/poly cap.

I have read of many radio repairs failures of output and even some interstage transformers failing and though amateur transformer forensics reveal corrosion and often associated with moisture, the green dots in an OPT winding can be the result of static induced arcs and acid production.

I would give some thought after installing a Hammond (costly) replacement or preserving an OEM functioning transformer...

YMMV

Chas