Results of air testing are promising but I was concerned about over modulation so I built a SL6270 VOGAD followed by diode limiter and low pass filter seemed to work ok but still wasn't happy.
Instead I installed a negative peak stay alive combined with a mic gain ALC based
on this but with an LED / FET optocoupler rather than an LDR.
I used an H11F1 mosfet opto coupler as the control element, much faster response time than the 20mS or so from an LDR.
Seems to work very nicely once I had adjusted the component values.
However initially I was getting asymmetric modulation positive peaks of 60% and negative peaks of 80%. Changing the polarity of the modulation transformer secondary balanced the +ve and -ve peaks quite nicely.