Yet another AA5 (an Emerson 708B) modified with some subminis and designed to run off 8VAC. Why 8V you ask? This dates back to my late 1960's model railroad days when I ran HO scale streetlamps off 8V instead of 14VAC, to prolong bulb life and make the lamps look more realistic. Built a 8VAC 5 amp bus for this. Still have it. Anyway:
To get plate voltage I used a 110V to 12V transformer run in reverse 8V to 72V. And I connected the new 72V secondary atop the 8V line to get 80V (unloaded). Then voltage doubled it, though the secondary's resistance lets the resulting B+ sag to around 135V. The nature of the current filter cap recharge is that it looks like pulses with a duty cycle of around 5%. This radio's current draw on the B+ is about 40ma, and this combined with this 5% duty cycle makes for peak currents 20 times larger, here about 0.8A. And with the 72V secondary resistance of about 40 ohms that makes a loss of 32VAC. Subtract 32V off the 80V and that yields 48VAC, and with the voltage doubler (x2.8 ) that produces 134VDC. About what I measured. Secondary resistance is effectively multiplied, making it worse than you'd think.
You don't see this with hot chassis silicon diode rectifier power supplies, as American 120V 15A branch circuits have an effective resistance of about 1⁄3 Ω.
What it looks like:
I edited the tube placement chart, printed it out and glued it to the cabinet bottom. It's here as an insert in this picture.
More such adventures at my web pages and sites http://pw2.netcom.com/~wa2ise/radios/AA5submini.html