> What if you start out with AC?
AC is going out of style. Not just Vampire Load; it's got the the point that a 37KHz switcher is cheaper than 50/60Hz 120V/230V iron.
Note also that we have a lot of DC supplies; adding an AC wart to the mix increases confusion and possible smoke.
> thrift shops and pawnshops
> if you have a 9VAC source, you could use it to provide a 26VDC supply
No. Each stage charges to the Peak of the sine AC wave. "9VAC" makes around 12V DC, or 24 or 36 V DC.
The "Delon" plan makes a lot of sense for many purposes:
Ground the bottom, call it +24V. Ground the middle, call it +/-12V. The losses are small. When used single-ended (ground and +24V) the ripple is 120hz and small. (Worked bipolar, the ripple is two out-of-phase half-waves.)
If one end of the source must be common, or for more-than-double, we go to the Greinacher schemes. These tend to sag more, more-and-more as the leverage rises.
> a power amp chip that will "work" off a rectified 9VAC (12.7VDC), but provide more power at higher supply voltages, do I gain advantages by doubling or even tripling, or are the current limitations of the wall-wart going to over-ride all of that? Assume here that I have a 9vAC unit rated at 1.2A.
Assuming simple (not bridged) amps: The full-roar amp presents a load to the DC supply roughly six times higher than the load impedance. 8 ohm load reflects as 50 ohms to the supply. 4 ohm load reflects as 25 ohms to the supply.
Note also that a simple (not multiplier) DC load must be multiplied by about 1.6 to find the load on the AC dupply.
So 12V DC to an amp with 4 ohm load draws 12V/25= 0.5 Amps.
The AC supply must deliver 0.5A*1.6= 0.8 Amps.
Try doubler. 25V out, 25 ohm loading, 1A DC, 1.6A load on the AC supply.
Your 1.2A wart will sag or smoke at full roar.