The 220 kohm bias supply resistor from the AC secondary had gone open and the output valves were badly damaged.
Will Marshall never lean to use decent resistors here or else two in series ??
When new Sovtek 6L6wxt valves were fitted results were poor, low power ( about 22 watts) and asymmetrical clipping into an 8 ohm dummy load. Checked the usual suspects in the phase splitter and even tried the amp with the new6L6s removed - I saw good drive waveforms on each grid pin on the CRO.
Eventually I found it - the amp had been mis-wired in the factory !!
The 4.7 kohm "presence" pot was NOT connected to ground as shown in the schem but rather was wired as a rheostat with the 0.1uF cap in series to ground. DC conditions in the phase splitter were well off since the only path for DC current was now via the 27 kohm feedback resistor to the OP tranny secondary.
Another point is that the Fender Bassman 5F6-A circuit the Marshall slavishly copied used four 10 inch, 8 ohm speakers in parallel - so a 2 ohm load. But dopey bloody Marshall used an OP with 4, 8 and 16 ohm secondaries and linked the feedback to the 16 ohm one !!!
Changing it to the 4 ohm tap produces 6dB more gain and better stability - as you might well expect. The amp now behaves well and produces symmetrical clipping under all loads.
The 0.5 amp slo-blo fuse in the HT holder was of the coil spring type. The fine fuse wire inside the coil had vaporised so current was being carried by the spring and the end wire.
On test, it passed 5 amps without trouble ........