The voltage at the supply capacitors is 7 V. If the voltage were
lower, the amp would be less efficient and would give lower output power, if the voltage were higher,
the amp would be less stable. The author says, that with the FET's he used in this amp the negative
additional supply might have a higher voltage - about 9 or 10 V, because otherwise the negative half-period
is limited barely earlier. This is however nothing important. This amp is fed with +/- 56 V of the main
6Va(1)<->(13) ;\ no problem if you confuse these two,
6Va(2)<->(14) ;/ the phase doesn't matter
41Va(2)<->41Vb(1)<->(5) or (6) ; = common ground
6Vb(1)<->(15) ;\ see (13) and (14)
6Va means the first six-volt winding, 6Vb means the second. It's the same
with the main windings.
The number in brackets means one end of the winding or a pin of the amplifier.
Just be careful to connect the main windings correctly, so that you get 82V AC
between 41Va(1) and 41Vb(2) - if you get a zero, swap ends of one of the two windings.
Apart of this it doesn't matter under which phase you stuff the AC voltage into the
rectifiers, just connect the common pin of the main windings to ground and
keep the small windings insulated from each other and from the main one.
I calculated the AC voltages from nominal DC values given by author and
usual voltage drops of Si diodes (0.7 V per diode). These voltages are not
crucial, especially the one of the main winding. It will drop with load
anyway. If you use lower voltage, the amp will give less power and may
sound more tubish, as a matter of fact. If you increase the main voltage,
you risk burning some parts - but I think +/- 63V DC is still bearable.
The trafo should be able to give some 6 amps, (or maybe less, this is the
consumption at full throttle) and it must hold the nominal voltage (2*41V)
at 200 mA which is the DPA380's bias current.
The small 7V DC supplies should be more precise, though. If you can't get
exactly the 7 Volts, but maybe 12 or 15, use some cheap stabilizers. I had
to do this myself - I had 10 V DC, so I used a darlington transistor of
apropriate wattage, a resistor and a zener diode in each 7V supply.
Ok, and how much amps must the 2*6V windings do?
Not sure about this one, I've never measured the consumption. Definitely
less than 0.5, probably less than 100 mA. Suppose you've got a 250W trafo
per channel: SQRT(250)=~16 (cm^2 - square of the trafo's core) => you need
45/16 =~3 turns of winding per AC volt, that's some 18 turns of wire. I
don't think 36 more turns is a problem. As for the wire's diameter: 0.5A
=> 0.15 mm^2; 2*SQRT(0.15/PI)=~0.44 mm. The wire should be 0.4 or 0.5 mm
thick, even less would do.