[Alex Pogossov]

If you are experimenting on a running tube amp, you might be at some stage
pulling a hot output tube from its socket while the amp is running.
Alternatively you might accidentally send a high negative pilse to a grid of
the tube while working on the circuit.

In any case, an abrupt interruption of the plate current will cause a
voltage spike on the OPT primary. The stored magnetisation energy of
Lp*(Ia^2)/2 will have to be dissipated. Lp can be quite high in a decent
amp. If a load is connected, then the energy will be dissipated in the load
giving you a loud crack in the speaker. A residual smaller energy stored in
the leakage inductance Ls*(Ia^2)/2 will most likely be safely dissipated in
a snubber of say 2200pF+5K usually connected in parallel to the OPT primary.

But what is the load is not connected? Then the huge magnetisation energy
will either cause:
- arcing in the tube;
- breakdown in the OPT winding insulation;
- breakdown of the subber capacitor (say 2200pF);
- arcing elsewhere in the wiring.or in the tube socket.

Would it be a good idea to placa a varistor rated slightly above the +B
across the primary? Or a spark gap of some sort, or gas discharge surge
arrester tube (the later have miniscule capacitance and very reliable). I
remember in vertical deflection stages of old TVs such varistors were a
must, since the abrupt cutoff of the current (during flyback) in EL84 often
used for this purpose was the mode of operation.

However I have never seen any discussions on the varistor protection issue
on this site.
What is your opinion?

Regards,
Alex