Looking at the photos, your spark plug is upstream of the two orifices, in 'dead space'. It also looks like the two orifices are pointing directly at each other, both at 90 degrees to the igniter axis, and 180 degrees apart from each other. (I figure this is what you mean by 'unlike impinging', but of course I can't tell from the photos whether this is precisely the case.) That's a really good way to produce nondeterministic flow patterns: which way the splatter will go depends on how, precisely, the two streams meet: what the mixture ratio is going to be at the spark plug is anyone's guess, and could easily oscillate between too rich and too lean. If I've misinterpreted, and the two flows are set to miss each other (and it sounds like this is Robert Watzlavick's design), there will be less nondeterminism, but I still wouldn't like to predict what the mixture ratio will be at the spark plug. Besides oscillation between too rich and too lean, that dead space also could get filled with nonburnable combustion products, which then got cleaned out by eddies of the flow, followed by a minor explosion and a return to the nonburnable state; you could have oscillation that way. A 'popping sound' is, of course, a small explosion. Which is sort of okay for an igniter, since the whole point of having an igniter is to make the 'hard starts' small and containable. But of course you want just one of them, followed by smooth combustion, not a "machine gun" like sequence of them. -- Norman Yarvin http://yarchive.net/blog