First...the schematic you posted came from a trace in FSB
here. By all accounts it was never fully verified. Everything apparently worked but for the Sub Octave section.
Here is the VERIFIED bipolar version with the CA3094s. Both schematics are practically identical with exception to some value changes and some specific handling of the conversion to LM13700.
There is a major difference at the Sub Octave tracking section. The verified version uses a 3094 into a FET but the Bernard version uses a PNP transistor driven by the first amp output stage. No one tried to BB that section since the FSB thread so no one can say if the Bernard version is correct or not.
So I'm wondering if it's worth subbing in a 13700 for the 3094 in this section as well.
The major differences between the 3094 and a 13700 are
3094 has 1 diode drop at Amp Bias (Pin 5) while the 13700 has 2 diode drops (Pin 1 for first half, Pin 16 for second half)
3094 has a buffer permanently connected to its output but you can tap the output BEFORE this buffer using Pin 1. So this equals the 13700 output pins 5 and 12.
So this brings up a few questions. As the buffer is internally connected to the output on the 3094 but is isn't on the 13700...must we still connect the 13700 output to the buffer input (pin 7 and 10) EVEN if we don't use the buffer? That's the way it looks with the 3094 anyway.
Then theres the matter of the Darlington pairs collector connections. Internally connected to V+ on the 13700 but you can tap into the collector of the second one on the 3094 through pin 8. There are instances on the Verified scheme that Pin 8 is connected to Pin 7 so exactly like the 13700 BUT there are a few times when Pin 8 is going somewhere else. That would explain why the Bernard scheme sometimes uses an external darlington pair to provide feedback i.e. he ignores the internal pair as he can't tap one of the collectors so he builds one externally.
So with all that in mind....
So my advice pick a section that uses a CA3094, convert it to 13700 based on the above rules and BREADBOARD it lol! I reckon go with Bernards scheme and replace the Sub Octave with a 13700 conversion from the other scheme.
I hope some others can chime in here. Let me know what you think anyway.
EDIT: Added the schematics of both the CA3094 and the LM13700 for people to compare.