[bikerhifinut]

Thanks for the input.
A couple of possible design errors of mine have come to light.
One in particular is my idea to reduce the grid leak on the input valve from 1 meg to 100k, My thoughts were as it would reduce noise a bit and all my sources are very low impedance it wouldn't be a bad idea. What I haven't taken into account is that I left the grid leak at 22k. With the input at the junction of the 22k and a 100k to ground I've inadvertently created a substantial potential divider at a point where I don't really need it. Shorting the socket with my shorting plugs has the obvious effect and the hum totally disappears. But with a source and even with a potentiometer on the input it's a different tale. So thats the first point of action in the morning. I'm minded to restore the 1meg grid leak resistor rather than reduce the grid stopper. But I'll have a scribble round the input circuit.
Another thing is with the EF86 valve triode strapped and a 150k anode resistor, theres potentially more gain than needed and a reduction here with a commensurate alteration of the compensation network on the anode might make a big difference too. Thanks to Terry for pointing that one out. And back to the curves.
I'm fairly certain the problem, such as it is, lies in the input stage. The output is totally hum free with no input to it So the heater wiring isnt injecting 50hz where it isnt wanted in that part of the amp. And none of the power valve heater wiring gets anywhere near the signal valves.
All fairly simple fixes I reckon. But one at a time so I can see what if any effect they have. I've already lifted the heater ground references to 40V above 0V which if nothing else gives the cathodes of the phase splitter an easier life. No Hum advantage as they are already warmed with regulated DC.
The amps going under the scope tomorrow if I get time and that hopefully will identify the issue better. I'm having to do some hasty revision on oscilloscopes as my memory aint what it was these days.

A.