First... the circuit diagram is fine, easy to read, it helps a lot.
200V and 20mA is about what I would expect using the ECL80. That's 4W. The ECL80 is rated at 3.5W max anode dissipation, and 0.5W for the screen-grid, the triode, and a bit of balance current, all adds up fine.
Circuit values all look OK. You could probably dump the 47pF capacitor between the anodes. It just adds a bit of high-frequency roll-off.
When you say the frequency response improved, I take it that this was at the low-frequency end? Because core saturation reduces transformer inductance and this hurts LF response. (Fantastic news that you have access to a signal generator and a 'scope.)
You might also try adjusting the bias. If you use more bias the ECL80 pentode will pass less current. That will reduce available output power of course, but it will also give the transformer an easier time, so you may get back as much as you lose.
Connecting a capacitor across your balance-adjust resistor is virtually equivalent to connecting the capacitor directly across the balance winding. And by transformer action it's then equivalent to connecting said capacitor directly across the transformer primary, which you can see would just shunt away your audio, as you have found. What you actually want is just the opposite - a big choke in series with the resistor, or even better, a constant-current source such as a pentode valve. (But if you use a valve, then you might as well drive it rather than have it just passing steady current, in which case you have push-pull!)
Carry on the good work... in the meantime I will have a look as I am sure I have a few small output transformers that I removed from dumped TV's years ago, which would be 'near enough' for the circuit... and would be interesting to compare performance against the mains transformer.
200V and 20mA is about what I would expect using the ECL80. That's 4W. The ECL80 is rated at 3.5W max anode dissipation, and 0.5W for the screen-grid, the triode, and a bit of balance current, all adds up fine.
Circuit values all look OK. You could probably dump the 47pF capacitor between the anodes. It just adds a bit of high-frequency roll-off.
When you say the frequency response improved, I take it that this was at the low-frequency end? Because core saturation reduces transformer inductance and this hurts LF response. (Fantastic news that you have access to a signal generator and a 'scope.)
You might also try adjusting the bias. If you use more bias the ECL80 pentode will pass less current. That will reduce available output power of course, but it will also give the transformer an easier time, so you may get back as much as you lose.
Connecting a capacitor across your balance-adjust resistor is virtually equivalent to connecting the capacitor directly across the balance winding. And by transformer action it's then equivalent to connecting said capacitor directly across the transformer primary, which you can see would just shunt away your audio, as you have found. What you actually want is just the opposite - a big choke in series with the resistor, or even better, a constant-current source such as a pentode valve. (But if you use a valve, then you might as well drive it rather than have it just passing steady current, in which case you have push-pull!)
Carry on the good work... in the meantime I will have a look as I am sure I have a few small output transformers that I removed from dumped TV's years ago, which would be 'near enough' for the circuit... and would be interesting to compare performance against the mains transformer.
