Thanks for the reply, I may have to consider a crystal oscillator, since I also need bipolar output, while the slayer exciter has the secondary grounded at one end. (I still get bipolar output now, but only because my 'ground' consists of a small heatsink on the bjt). Do you think that tuned-circuit oscillators like Colpitts or grounded-base would also be disturbed by the coupling of the secondary to another coil capacitively?
Or could I have a single oscillator driving two primaries (see diagram), which induce currents in two different secondaries, such that the phase shift between the secondaries is the required phase shift. For example, the first secondary could be slightly capacitive (relative to primary) by half the needed phase-shift (31.5 deg), and the other secondary slightly inductive, providing the other half of the phase shift. Would a crystal oscillator topology allow for something like this?