After spending some time talking to a knowledgeable friend it became apparent that all the inputs and outputs required to implement a traction control system are already fitted to the car, they just need linking together in the right way.
So, the theory is that the ABS sensors on the front wheels can be used to generate a variable voltage depending on the wheel speed (frequency to voltage conversion using National Semiconductor LM2917 ICs). I've got one of these wired already using existing ABS connectors, so as not to cut the regular wiring (thanks to Seamus Burke for letting me have the socket from an old ABS sensor). I now have two of these working, which are calibrated to give the same readings in a straight line.
Now the clever stuff is done by a PIC16 microcontroller, into which I will feed the left and right wheel signals (along with a rev counter input for low rpm condition monitoring - don't want the traction control to stall the engine ..) This controller compares the two wheel signals and when they are sufficiently different, ie. more than just cornering tightly, will reduce power to the wheels. The power reduction will be achieved by cutting the earth path to a series of injectors, two seems to be enough though the circuitry is available for four.
After some experimenting with the PIC (16c71 for anyone who knows them ..) using the controller wired into the car and lighting an LED when the wheelspin condition was detected I was able to refine the code driving the microcontroller in order to have a reasonable success at detecting it. I was initially getting false signals when travelling in a straight line, and this was mostly due to stray signals from the engine. Changing the pickup wires for high quality microphone cable and adding some more smoothing at the F->V converter circuits totally eliminated this. Once this was working reliably I wired up one injector to a cut-off relay, but found this was not enough as especially in the wet the torquey V6 could still keep the wheel spinning. The second injector was added and this helped things a lot, it is now fairly impressive when in use but not intrusive enough to be annoying.
The controller board -
Wiring schematic -
As an addition to this work I found that the spare channels on the PIC could be utilised for various other signals. Two of them are configured to provide outputs as the revs rise past 5k and 6k rpm, which I will ultimately wire to some LEDs on the dash as a sort of gearchange indicator. A third is setup to monitor the throttle angle and RPM which it uses to disable the air-conditioning compressor under high engine load conditions (basically over 70% throttle or over 3750 rpm will cut-out the compressor). Although the Honda ECU is supposed to do this itself I found in reality this is not the case, as a LED I wired across the compressor would not go out under such conditions.
Mail me comments