> I think that it should be possible to use this without the feedback-loop.
Sure, but it will suck. The driver is heavy and has a falling frequency response. It also has a rising impedance. The classic way to fix the response is to drive with a high impedance, 5 times the coil nominal ompedance. This is natural with tube drivers. Opamps normally give a low impedance output. The $0.06 trick is to put in a resistor. But now 5/6ths of your opamp output power is wasted. Since the coil needs very little power to get muddy, this may be acceptable. The other trick is a small series resistor, and take feedback from there. The opamp is set to hold the resistor voltage constant, resistor current is constant, coil current is constant at all frequencies.
The beefier opamps "should" drive the 8 ohm coil with >40 ohms dead series resistance and conventional feedback (before the resistor). 100-200 ohms may be needed to keep the opamp from straining too bad.
Dang near any NPN-PNP pair will work. If you went back in time to when those Accutronic schematics were prepared, I bet you would find that those were the cheapest 30V parts around. The PN3904/3906(?) jellybean pair may be easier for US-folks to find today.