The timing of the spark for inductive systems (points or the current automotive transistor stuff) is controlled by the points opening, where the spark timing on CDI systems is controlled by the timing of the discharge of the capacitor (about 300 volts) through the primary of the coil. Thus, inductive sparks occur when the magnetic field collapses, while a CDI spark occurs as the field is being built. Inductive sparks are longer in duration for a given energy, but CDI sparks have a very fast rise time (time between when charge starts going to the plug and when the plug fires). Short rise times make fouling less of a problem. This is because a spark plug acts like two circuits, a capacitor in series with the gap, and a resistor in parallel to the gap. If the charge does not build up enough charge fast enough on the capacitor to reach a voltage sufficient to jump the gap, the charge will bleed through the resistor (deposits on the core nose) and the plug will not fire. Small gaps help prevent fouling because a lower voltage is required to jump the gap. When an engine is spinning slowly, the charge coil may not be able to charge the capacitor in a CDI system to the normal voltage (again, about 300 V). So, if you ride trials and you want to avoid killing the bike when the RPMs drop a little low, try a smaller gap. Probably the most active factor in determining the required voltage to jump the gap is the stoichiometry in the gap. Lean mixtures require a LOT more voltage to ionize the gap.
Doc