It's to limit reverse transient voltage on that node.
Op amp U3 appears to latch up if the input voltage goes too much below ground (op amp negative rail).
A bit late I know, but on the original question, I've used the HCPL2531 quite successfuly for linear aplications as described in this app note: https://www.avagotech.com/docs/5954-8430E
A bit over the top for my needs - but thanks for the input. Opto-isolators are something I know hardly anything about, so the app note makes interesting reading
Crutschow's solution is almost bang on the mark for what I need - I'm just trying to tweak it a bit
With the extra transistor between the slave's opamp output and q2's base, the diode wasn't enough to prevent oscillation. A cap from the opto's led anode to gnd did the trick though. Off to the shed now to try it out. Huge thanks
The basics of the common type transistor output opto isolator is that the output transistor current (when normally biased by a positive collector-emitter voltage) is proportional to the input LED current and, for low currents, the relationship is quite linear.
The ratio (gain) between the diode input current and the transistor output current is called the current-transfer-ratio (CTR) and is given on the data sheet.
Note that it can have significant variation from unit to unit so you must account for that in your design.