The "Importance" of SFI

[SappySE107]

I haven't seen a table anything like what the 2008 3900 code has for intake port temperature, but I don't code anything either.

[Futures Passed]

direct injection is quite an awesome concept.... if only it were more easily controlled. if a non-direct injection PCM were used without any modifications, there is a pretty good chance that the injectors would be firing when the exhaust valve is open.

Are there any fuel injection systems which put the nozzle in the intake valve seating, (embedded deeply so it doesn't get interfere with the seal) so it can't fire unless the valve is open, but it is still direct injection? (Or pretty close to it; the spray could be aimed away from the valve.) That way the nozzle is protected from combustion, too.

[robertisaar]

sounds too complicated for a street engine, IMO, so it's likely no OEM would make anything like it, if it's even feasible to begin with.

[geoffinbc]

You would have to mount the injector in the head. If your going through that effort than true DI is not much more effort and a much bigger payoff.

[LZeppelin513]

Nice article

[robertisaar]

so, not that i care about SFI at all, but i'm pretty sure i've managed to visualize a working scheme to convert a MPFI ECM into a SFI setup using some external electronics...

however, the same principle could be used with direct injection, which is why i even consider it worth mentioning, since that seems to be an excellent technological advancement.



anyways, imagine, if you will:

stock MPFI ECM, generally set into double-fire mode, meaning 2 injector events per 4 stroke cycle, as opposed to single-fire, which opens the injector once per 4 stroke cycle. the only time there are exceptions to this are when the ECM goes into Asynchronous mode. this can happen when there is a sudden increase in the amount of fuel needed due to rapid transient conditions. direct injection either completely eliminates or severely reduces the need(i haven't looked into it much) for that due to the injector spraying directly into the combustion chamber, eliminating fuel that would otherwise stick to the intake valve or port wall. anyways, the hardware wouldn't require any modifications to function with port SFI that do need transient fueling accounted for, and it will be done sequentially as well.



anyways, the idea is that if possible, force the ECM into single-fire mode(if not, you'll have SFI..... kinda. read on, and it should become clear) so that only a single pulse is done for the entire 4-stroke cycle. it won't matter how many cylinders the engine has either, the add-on hardware will take care of that. give this hardware a cam signal(with the signal pulse being an important difference between port and direct injection), an engine speed signal like the 3X signal the 60V6 ICM creates(basically, the same number of pulses as there are cylinders in a 4 stroke cycle, the 60V6 ICM creates 3 pulses per revolution, with 2 revs per 4 stroke cycle, hence 6 pulses, all evenly spaced), connect to the ECM's injector output circuit and connect to all injectors with their own dedicated circuits for sequential control.



here is where the magic happens.... when the ECM triggers the injector circuit to start the injectors, this hardware(it's a microcontroller, like an arduino or something else simple yet capable) will hopefully already have seen a cam signal, but if not, let's have it fire in batch-fire until it does see one, but we all know how that operates, so i'll skip it. anyways, it sees a cam signal and based on that and the 3X signal from the ICM, it now knows which cylinders are doing what and how fast they are doing it. back to the start of the injector signal, assuming a cam signal has been seen: a timer now starts to see how long the injectors need to remain open to deliver the amount of fuel the ECM wants to deliver. so the module turns on the first injector while counting time and watching for a pulse on the 3X circuit to indicate that another cylinder is coming up to TDC. the 60V6 ICM actually pulses the 3X signal 60* before another cylinder comes to TDC, so now another timer starts that counts between 3X pulses to create an engine speed calculation, which will be used to help turn on the injectors at the correct time.

i keep getting sidetracked.... anyways, for a port injected motor, let's say the cam pulse happens as soon as one of the cylinders(doesn't matter which one) closes it's intake valve, so that injection will start when it is most advantageous to do so. the module now knows engine speed and which cylinder to fire and when to fire the next one based on the 3X signal and cam signals, so while it's watching for the 3X signal for the next injector to start firing, 1 of 2 things will happen: either the ECM will quit signaling to fire the injectors and the counter will stop, then the next injector will open and close based on that time period or the time to start opening the second injector will happen before the ECM drops the injector on signal, in which case injector 1 will still be open and it's timer continuing until the signal is dropped, with more injectors opening as necessary. with a really high injector DC(above 83.333% with a 6 cylinder), all injectors will be open at the same time, but the timer for cylinder 1 will still be counting and waiting for the timer period to keep all injectors open and close them in sequence as well. should injector DC climb to 100%, the timer will never time out and all injectors will be open at all time, which is exactly what is desired. when they drop below 100% and the ECM drops the injector on signal, the timer will finally come to a value and the injectors will start closing as intended.

async fuel will be dealt with sequentially as well, since async is done after the main fuel pulse, so the signal drops for whatever time, then kicks up again for a small amount of time, the back down, so when the module sees that, it will mimic the same behavior, with each cylinder done in sequence.

if done in double-fire in place of single fire, pulses will be done sequentially, but twice as often.... essentially, injectors are fired 1-2-3-4-5-6-1-2-3-4-5-6 for 2 revolutions as opposed to 1-2-3-4-5-6 for 2 revolutions as single-fire would do. if you think about it... if injector 1 fires as soon as the intake valve closes, rotate the crank 360* so that the injector will start opening again and from the looks of the diagram i made in post 1, you'll at the very least start firing against a closed valve. short pulse widths will complete by the time the valve opens, but longer ones will still be firing when the valve opens.



as you can imagine, this isn't as sophisticated as OEM level SFI control, since each cylinder will be getting exactly the same amount of fuel, as opposed to individual cylinder trims as done with some systems. it also doesn't allow for adjusting the injection point other than varying where the cam signal happens. which from what i can tell, with port FI, you want it done as soon as the intake valve closes, with direct FI, as soon as the exhaust valve closes.



food for thought. i can actually imagine all of this at a programming level so much easier than how i can explain it. maybe it's just me?