useful stuff here
http://www.roostamps.co.uk/roostschematics.html
Owner switched off amp when he saw one of the bottles melting , leaving a
deep dimple over the anode but not punctured. Its matching polarity one had
starting to do the same. (Reminder to myself , leave some slack in axial
leaded components, regimented stuff looks fine but this is the practical
downside). 100K bias supply resistor lead had snapped because tightly
joined to 10K phenolic preset wiper, itself mounted on turret tag strip, not
sheet, so plenty of scope to bend with humidity/temp/vib.
At the moment trying to find from the owner what the sequence of distortions
were before final meltdown. As one of the double pack HT electrolytic cans
(for the preamp) is kaput and the reverb transformer seems also kaput.
I would expect .5 to 1R // 1k to 2K but this measured DC and AC R and L , is
0.2R // open, I can never be that sure with high L. Reverb tank disconnected
measures 0.8R and 165R. I'm fine with restuffing can electros ,Main HT
seriesed electros seem ok, and cannot see how loss of bias would blow preamp
stuff. Unwinding/Rewinding small reverb tr will be something different.
The output matching transformer is 27R/27R ie the same , does that mean
bifilar? seems unusual

That was quite painless. I suppose as audio, between mains frequency and
SMPS f., T&I iron lamination but air gap, all E in one varnished lump and
all I in one lump. Between interlaced I&E iron of mains and ferrite
airgapped of smps. 34.48gm of bobbin,terminals, thick wire and the broken
0.13mm wire . On outside winding , should count off easily via paint
stripper softening or hot air and weight if not. No obvious overheating on
the outside or telltale melting of the bobbin at the now revealed centre.
I suspect just marginal overheating combined with too much wire , it is
seriously squashed up on the bobbin to have got it slid over the E plates,
and pehaps just a mechanical break somewhere, but it did occur in the same
session as the loss of bias meltdown.

Counted off with no breaks, paint stripper on outer layer and then blast of
hot air if the wire was "grabbing" . No hot spot or shorted turns. Break 5mm
from the end due to mechanical strain presumably.
No wonder only 0.2R over the inner coil , it is .78mm diameter wire. 3 amps
of audio ac expected ? , I can see 0.6mm wire going back there ,or whatever,
for 1 layer of same number of turns
1 and a part layers , the part layer had ruckled under the tape allowing
the thin wire turns to drop off at the ends , hence grabbing on counting
off, and pushing radially over the wire lead-out, especially as a few part
turns to take up the recess from the thick wire part turn. Surprisingly no
obvious punch through between primary and secondary (which is which in this
circumstance?) as ruckling had allowed fine wires to touch thick wires it
would seem, bypassing the intended interlayer tape

anyone aware what the innards of valve/tube output matching transformers
look like? do they have air gapped soft iron aggregated E& aggregated I
laminations ?