I might be able to get away with sing a 1N5821G Schottky diode for D2 an d moving it back next to the input fuse. That will of course alleviate the reverse polarity vulnerability. It is awfully tight, however. Assuming t he main supply is delivering a full 12V, the current should be no more than 1.3 amps. At that current the voltage drop across the diode should be .4V or less. That leaves us with 11.6V at the charging circuit input.
I could employ an FMMT720 for Q6, in which case the CE saturation voltag e for an emitter current of 300mA or less is .1V max. The voltage at the i nput to X1 would then be no less than 11.5V.
The 1N4148 diode should reliably have a forward bias voltage of between
0.6 and 0.7V. Assuming the worst case of .7V, then the charging voltage at F2 can be as high as 10.8V. That yields a very thin headroom between the terminal voltage of a fully charged 9.6V NiMH battery and the maximum avai lable charging voltage. It may be serviceable, however. Note the power su pply must absolutely deliver no less than 11.8 volts or so.
The circuit is quite serviceable with a power supply input of up to 18V. I will play with this when the prototype boards come in. They should be here in a couple of weeks.