My recommendations are use good, hi capacity AA NiCds, make all wiring as short as possible, in thick flexible wire and make sure that your solder joints are top rate. Also Make sure that your on off switch is capable of switching the currents involved without a significant volt drop. You only have a very low voltage across the battery in the first place, so you cannot afford to lose any voltage at bad joints!
If you use say an OK Super 60 (a la Bob Zambelli) then the points are up to automotive standards (I use Nissan auto points on mine!). This means that you are unlikely to get much volt drop across the points (compared to minature points you see on a lot of engines). With the OK engine, the points are at the back and don't tend to get oil soaked.
For minature points at the front of an engine, I would use a transistorised unit to be on the safe side. You can't go wrong with Floyds's circuit, there are plenty of other designs about that work reasonably well. If you are new to the game, use Floyds, then you cannot go wrong.
Almost as important, make sure you have a good plug, most of the old Champions can leak and cause all sorts of problems. Take a look at Bob's writeup on his Super 60, lots of excellent tips there.
Our small Kettering system is in fact a form of CDI ignition. The turns ratio on our minature coils are only several hundred to one, with around 3volts LT available, this would only give a few kilovolts at the plug, insufficient for a spark. The back emf generated by the coil (when the points break) is sufficient to charge the capacitor up to 200 to 300 volts, this is then dumped across the LT winding of the coil and with the turns ratio as it is, this gives the 20 kilovolts or so across the plug and hence the spark. So a good high voltage capacitor is required! I suspect I may be challenged on this interpretation of events. However anyone with a storage scope can see these high voltages across the capacitor.
For control line use, the addition of microprocessor control is a waste of time. Mind you for radio assist, old timer flying, it is great to retard the ignition by applying a voltage from the reciever output directly to the processor. This eliminates a servo to actuate the mechanical advance retard. This is just me enjoying playing with processors. For control line flying, use the KISS principle! Or a variation on Occam's razor principle, keep things to the necessary minimum and there is less to go wrong!
Regards,
Andrew.