Originally Posted by
Epstein
Figured out how the ignition dwell settings work in the S13SR ROM's. Got to borrow a scope from work to help out. It's pretty sweet.
So apparently the stock map for dwell time dips down between idle and about 4k rpms. Above 4k, it's basically flat. The idea with dwell time is to charge up the coil for a certain number of milliseconds. That charge is then fired through the plug. The longer the dwell, the more charge, up until you've saturated the coil. So you need enough dwell to fire it, but not so much that it gets too hot. At redline, the map has about 25% more dwell, so we know that it can take that amount of time. So, just making the map linear with the same amount of dwell across the rpm range will be an improvement. Calculating is a pain in the ass, since they generate this number based on degrees of rotation (dwell angle), so you have to divide by 360* to get the duty, then multiply by the time per rotation to get a number in mS.
The result is a much smoother car in the 1000-4000 rpm range, with stronger return-to-idle qualities. My car has been a pain in this band due to combustion chamber reshaping, large-ish cams, and a greddy intake mani.
For those programming at home, 0x0510-0x051F is the dwell voltage coefficient. Less dwell is needed with higher voltage because you're charging the coil faster. 0x0C40-0x0C5F is the dwell duty coefficient versus rpm (400 rpm per box). This table increases with rpm because there is less time per rotation at higher rpms, so you need to charge for a larger percentage of the time. Stock, the dwell time is about 2mS above 4k.
I ran into the dwell problem with a Mustang back in the cut. We really could have used someone who could pick apart the ECU like you can but this was back in 1998.

Our simple fix for lack of dwell (with one coil, the problem is greatly exacerbated) was an aftermarket ignition amplifier box. It worked but never gave us a finite dwell that was satisfactory.
Good stuff as usual, Chris.