Not sure I understand why Wayne used an opamp as an active buffer for that biasing on pin 3 of the THAT ic. I suppose he wanted a low impedance source but I think the input impedance is low enough on pin 3 of the THAT IC that a simple decoupled resistor divider would suffice. Use around 4k7 up to 10k instead of the 100k resistors, tie that to pin 3 of the THAT IC (get rid of U2A) and remove R2/C4.
The other possible solution is to connect the part to the 24v rail and ground but AC couple all I/O through capacitors. C1, C2, C3 are already coupled. Try removing everything connected to pin 3 of the THAT IC except R2/C4 and make C4 something large like 100uf. The part should self bias around half rail (12V) as long as there is no other DC connection. This is, however, guessing since I don't know the part at all but I've done it with other diff amps with no problems.