A correspondent of mine, Brian Magee, has been drawn in on the discussion about fuses, with particular application to valve equipment that had been so ably prodded along by Nigel Machin.

Brian writes:
Fuses in Valve Amplifiers.
I have never had any problems, so have not been forced to think too deeply about the matter..
The following thoughts come to mind when designing power supplies for valve gear :-

(a) Any starting transients in the high tension supply before cathode emission starts may generate large grid – cathode fields in the tube.
This can reduce cathode life.
In frame grid tube such as the ECC88– EF184 and the E180F the grid wire diameter and grid spacing is so small,  field emission occurs and this  usually destroys the tube.
To prevent this I always put a small neon between the grid and cathode of frame grid tubes.
To delay the onset of high tension I use indirectly heated cathode rectifiers such as the 5AR4.

Under these conditions the greatest peak input current occurs when the switching occurs at the start of a cycle, driving the transformer core into saturation. Toroidal core transformers tend to have smaller primary resistance and so greater peak input
currents.

To set the fuse rating on any new design, I simply pull the rectifiers out and switch the thing off  and on about fifty times upping the fuse rating until it withstands the turn on transient. It goes without saying that a fuse should not be placed on the secondary of a transformer driving a full wave rectifier if its failure produces a DC component in the secondary.

<end quote>

I do remember learning, years ago, that the initial magnetizing current in a transformer primary will depend on the phase of the mains cycle when the switch on occurs. My memory had been that the magnetizing current transient peak could be up to about twice the continuous peak magnetizing current. One would not expect this to exceed the sum of the continuous magnetizing current and the load current.  I have to admit that I had never thought of saturation in this context. Something new (for me) to think about. I have never been aware of any evidence of excessive current on transformer switch-on. I have been caught out, however, with a potential problem when a transformer is switched off. If there is no load on a transformer, and it is switched off with a switch with low arcing losses, then there is nowhere for the magnetizing current to go, except into the charging of stray capacitances or the creation of a carbon track through the insulation. A bit of a lottery as to which will happen first. After having lost a high quality and expensive transformer by doing no more than swithing it on and then off again with no load, I now have a rule that I never switch a mains transformer off without a load on it (a MOV on the primary will do), or if just powering it up to check secondary voltages etc. then use the variac so that the magnetizing current can be reduced at a slow pace compared with a mains half cycle.