1. U1A is shown as an LM324 because that was in stock but only two of the 4 opamps are utilized. Use LM358 if buying new.
2. R6 and R14 set the output level based on how much current is coming through the mirror, which is set by the 6GM8 bias point (times mirror gain). The shown values make it a bit low for the autobias to work. U1A compares the push pull output level to half V and increases grid drive, thereby increasing current, to bring the output up to the 9V midpoint.
3. As the volume pot moves toward zero volume the impedance seen by the opamp drops causing the opamp to max out attempting to drive it. The 470Ohm resistor on the bottom tail of the volume pot reduces that effect, but not entirely. Although, at such a low volume the output voltage swing is so small that it doesn't matter if the output level is not centered. It could probably use some refinement but works as is.
4. During power up there is a second or two of hiss as the filament warms, the tube begins to conduct, and the autobias settles in. At power down there is a speaker pop characteristic of capacitor isolated single supply class AB designs.
5. Q1-Q2, Q5-Q7, and Q6-Q8 are bonded together for thermal equalization.
6. Q3 and Q4 are heatsinked to the metal enclosure.
7. The small enclosure and decision to use it, itself, as the heatsink made the original 360mA idle a ridiculous notion so the circuit was revised to Class B at 85mA idle.
I built the solid state portion of each channel on separate perfboards mounted one per side on the upper, inverted 'U', section of the chassis and the filament regulator is mounted to the bottom center. Total dissipation for both channels and the regulator is on the order of 7 watts and that stretches the capability of the small steel enclosure so she gets a bit toasty after a while. Not dangerously so but enough to make the meek wonder.
R10 increases cathode voltage so the opamp output operates around 4V and input cap C8 DC isolates the input to avoid loading of the opamp output. Plus a minor revision to the bias point sensing filter.
I haven't gotten around to implementing these changes because, as noted above, the original circuit works as is.
(c) 2004-2008 All Rights Reserved