being somewhat surprised to find so many people apparently not understanding the distinction between the two different types* of electronic ignition used on cars, i dug up this which i should think explains it well:
i particularly like the burning paper analogy to explain spark energy requirement. anyone with access to a dual plug ignition system [like you'll find in planes] will be familiar with the fact that if you shut one plug bank off, power output of the engine will decrease significantly, which is about as dramatic a demonstration of spark energy requirements as i think you can find.
i find the whole obsession with cdi to be fascinating. it came to fame in the 1970's when consumer electronics was reliable enough and affordable enough to tinker in this way, and it was indeed an improvement over kettering ignition**, but there's almost no technical overlap between combustion physics and electronics, so i guess what might seem like a great electronics project would continue to seem like a great ignition idea if the spark energy vs. spark voltage concept is not understood.
inductive ignition on the other hand, particularly coil on plug, is absolutely the way to go simply because it provides not just sparking voltage, but significant total energy in a sustained event whose duration outlasts the whole combustion nucleation phase. kettering may not have had access to clean reliable switching [or dwell management] technology, but he absolutely knew what he was doing when it came to lighting off air/fuel mixtures.
- strictly speaking, it's debatable whether cdi /is/ actually "used" on cars - not a single oem that i know of uses it. it's the exclusive territory of aftermarket ignition - and a great example of inferior product getting mind-share over superior product simply because of better marketing. seriously, who in their right mind wants an ignition system whose only debatable benefit [multiple spark events - which are irrelevant if the thing could produce a sustained spark in the first place] cuts off at only 3,000 rpm and that fails to address the whole spark energy concept???
** kettering has a major problem: mechanical [points] switching is noisy. noisy in make, noisy in break. if you look at a fast oscilloscope trace of contact voltages at switch, there are great fuzzy regions where multiple rapid on/off contacts are made at each main event. that noise disrupts the coil input current in undesirable ways each time, thus translating to similarly "dirty" coil output. electronic switching is much cleaner on make and break, thus dramatically improving spark output. compared to kettering, even cdi output looks great!