RomRaider

merchgod or mickey... are either Lance or I right on this or are both of us off base?

it would be 12.2:1 AFR, assuming, of course, no other corrections to fueling.

Yes, assuming no other corrections to fueling. Then Lance is right and I am wrong. I thought it was a 10% change in enrichment vs. stoich and not a 10% change in absolute AFR ratio.

Can you include a cool icon to go with this cool program?
I've got this on my desktop with a geeky windows icon.

Cheers!

Sure, someone want to make one? That is not my best area.

So the fuel trims are ignored when the car is in open loop? I had assumed that the car continued to apply them, so if it uses 5% extra fuel to get open loop dialed in, it would still add 5% extra fuel in closed loop, and therefore be pretty close to the expected AFRs.

(Assuming for the sake of discussion that the D-range fuel trim is reasonably accurate in spite of the fact that the car spends little or no time in closed loop (and thus learning) at the upper end of D's MAF range.)

There's no "dialing-in" of open loop. It simply uses the D-range A/F Learning which is based on patterns of a/f correction made in this range when in closed loop. If the d-range threshold is 50 g/s (that would be 50+ g/s - no max limit) and let's say that, generally, you are in closed loop between 50 g/s and 60 g/s and it learns +5% there, then that +5% would be applied whenever airflow is 50+ g/s, whether that is 60 g/s, 100 g/s, 300 g/s, etc.

Crap, I had the "open" and "close" swapped in that sentence. Let me try again:



And it sounds like that's correct.

Guy on the corner asked: "If AFR Learning D is -10%, what would be the actual AFR at WOT in open loop for a map that calls for and was tuned at 11:1?"

So, shouldn't the actual AFR still be 11:1?

(Assuming for the sake of discussion that the entire D range is uniformly off by 10%.)

No.

Is this correct?

IPW = latency + (some_constant * MAF * (1.0 + enrichment) * af_correction * af_learning * etc)

Where some_constant is whatever it takes to get one gram of fuel per 14.7 grams of air.
"Etc" is other correction factors, to be ignored for the sake of discussion.

In closed loop:
* enrichment is 0
* af_correction is calculated in real time based on O2 sensor signal
* af_learning is whatever is stored for the current MAF range (it's adjusted slowly over time, based on af_correction values).

In open loop:
* enrichment is from the current cell in the OL fueling table
* af_correction is fixed at 1.0
* af_learning is whatever was learned for the current MAF range.

the injector constant is the ipw necessary to achieve stoich fueling per gram of load. If your primary fuel map is calling for 20% more fuel for example (i.e. 14.7/1.20 = 12:25:1 AFR) and A/F Learning #1 D is calling for 10% less fuel (1.20 * 0.90 = 1.08 = 14.7/1.08 = 13.6:1 AFR), ignoring all other fueling corrections, then obviously your AFR is going to change in open loop.

But if your AFR learning is calling for 10% less fuel, that suggests that your MAF scaling is either 10% too low (in the range in question) or your injector constant is 10% too high, or some combination of those. AFR learning + MAF scaling error + injector settings error should cancel out in closed loop to achieve stoich. (Or to achieve stoich with minimal AF correction, anyway.) So they should all cancel out in open loop to achieve the desired OL AFR as well.

Again, assuming that the MAF scaling D range is 10% off across the board (or it's perfectly accurate, and the injector scaling is 10% off).
I realize that the D range is broad and the only really "learned" at its low end.

Well, the ECU is working on the basis that your MAF sensor scaling is correct. If you are tuning for aftermarket intake, though, the airflow is not going to be off the same amount across the entire D range as it would be in the narrow range that D would be calculated in closed loop. So, it would be difficult to determine how much the AFR would be off compared to your fuel map.

Seems to me that half the reason the fuel trims exist is that the ECU engineers know the MAF scaling is not going to be accurate. So the gave the ECU the ability to learn (via what we call fuel trims) to compensate for MAF scaling error. So having 5% fuel time is basically the ECU's way of saying that the MAF scaling is off by 5%.

If your D trim is +5%, then your open loop AFRs at the low end of the D range should match your open loop fueling map at the bottom end of the D range. The 5% fuel trim is there to compensate for a 5% MAF scaling error that the ECU has identified.

The high end of the D range may be off, but if (and only if) the high end of the MAF scale if off by some value other than 5%. I guess what I'm trying to say is that AFR errors at the high end of the D range are entirely due to MAF scaling error, not due to fuel trim. Fuel trims are there to reduce AFR error, even in open loop.

I realize you're the one who has viewed the code, but if the people who wrote it did things any other way I'd be really surprised. It just doesn't make sense for the fuel trims to reduce AFR errors in CL and create AFR errors in OL. MAF scaling yes, but not fuel trims.

Am I making any sense?

Well, I can't read the intent for doing something from the code, but I would imagine the reason for long-term fuel trims are due to mechanical problems that might occur. For example, if you had a post-MAF intake leak or you had an injector stuck open.