The problem is that is not how a Super Scaler works.
Instead of a small input transformer feeding the power tubes' control grid (G1), a Super Scaler uses an output transformer which feeds the power tubes' screen grids (G2).
You take a typical output transformer and turn it backwards, so that the input is on the (4 or 8 or 16 ohm) secondary, and the usual primary is used to deliver high voltage and some current to the screen grids. The screens are considered quite linear, and need both a voltage and current input, which is power, and is why O'Connor says they function in the power domain.
How much power? The idea is to take a Champ-style amp with a few watts of output power, and use it to feed a pair of 6L6's to make the larger power output. The screen voltage determines the plate current, and (if I recall correctly) the screen is near d.c. ground. The control grid is also near 0v, but the tube can't pass current (or much current) without screen voltage. That means almost no idle current in these tubes. When there is an input signal (which is roughly 200vac or so), the rising screen voltage allows for a big rise of plate current. These things run in roughly class B, but work well.
You can't just make a large-voltage input signal, because the screens also draw current. So you need a power input. The end result is claimed to be a very accurate increase of power output from a small 4-10w amp to a total of 50-60w.
The schematic (in TUT 3, maybe?) is quite simple and the parts count is low compared to a complete amp. This is just a power supply and output stage, but you do need 2 "output transformers." One is an actual output transformer, and the other is for the input power.