RomRaider

I've noticed in the past that my AFR drops to around 12:1 and stays there when I'm at very low throttle. It happens consistently on a mild hill on the freeway heading toward where I live... I let the the throttle get very low as I go down the hill, to avoid picking up speed, but not quite low enough to force the ECU into no-fuel-at-all engine-braking mode, AFR drops and hovers around 12, and it will stay this way until the road levels out. If I give it any throttle, AFR goes back into the high 14s again. If I lift throttle entirely, the ECU closes the injectors completely and AFR goes to 20:1. It's particularly evident when I'm on cruise control and going down a moderate hill.

What I noticed just recently is that the car remains in closed loop when this is happening, and AF correction goes to -25%, and AF Learning #1 A will slowly decrease. I've seen the learning drop down to about -10%, and I'm sure it would keep dropping if i had a longer hill.

Has anyone else seen this? Does anyone have theories about what the ECU is trying to accomplish?

I don't think this quirk really matters, but it's weird, so I'm curious.

Its going in to open-loop when this occurs? I would say figure out whats causing it to go into open-loop and you will find the problem. Is it going into a weird area of your open-loop fuel target table?

No, it stays in closed-loop. If it was in open-loop, the AF Correction would be zero and it would not be tweaking the AF Learning.

This happens at low load and cruising RPM, so it's the upper-left corner of the fuel table, where the fuel table calls for 14.7.

Your original post says open loop. Maybe an injector latency issue, are the PW very low when this happens?

Or maybe just a really low mafv that you have never hit while tuning and needs to be scaled in the maf table

Whoops... I fixed the first post to say closed loop, thanks for pointing that out.

MAF is around 8 g/s, which is pretty normal. I might be able to fix this scenario with MAF scaling, but I think I'd end up being really lean any time I cross through 8 g/s under other scenarios, like revving up from idle any time I start from a stop.

Pulse width vs. latency is an interesting idea, I'll look into it. I have a pretty strong hunch that this issue doesn't happen every time I see that pulse width though - it's just something the ECU does for light engine braking. The fact that it screws up the fuel trim might be a bug. And it might be the reason the stock tune has separate fuel trims for 0-5 (idle), 5-10 (very hard to reach, other than light engine braking), 10-40 (cruise), and 40+ (cruising up steep hills).

What does the corresponding cell(s) in the MRP correction table look like?

Yeah the latency is a bit of a long shot but definitely possible. It could be tested easily though by reducing your latency settings and observing how much of an effect it has on the issue. If it was purely a latency error(doubtful) it could be adjusted mathematically:

issue afr = 12
target afr = 14.7

latency + (((issue afr/target afr) - 1) * PW-during-issue) = new latency

^^^ but depending on the PW at the time of the issue, that could make a huge change to the latency and mess things up. So it would only work if the issue was entirely latency error


I would think that it probably has more to do with the MAF scaling than anything. If you scale with romraider logger for closed loop you would probably never hit this MAFV area. If you did it probably wouldn't have any time to make an afr learning adjustment. So when you interpolate and scale the MAF a bunch of times it could be screwing that area up a lot.

I'm now certain this is not related to latency or MAF scaling. I get minimal AF Correction from 5-10 g/s under normal circumstances. This rich AFR & -25% AF Correction phenomenon only happens during very light engine braking. I've tried disassembling the code and logging some previously undefined corrections, but those other corrections are always zero (which might be when Merchgod never defined them ).

Do you have a VTA BOV with a weak spring?

Nope, full recirc and stiff spring. I get a little bit of surge when I lift off the throttle.

I think there's something in the ECU causing this but I haven't got proof yet.

Did you ever make progress on this? I see it on mine all the time, especially when I let off the throttle slightly but not enough to fuel cut. One of these things I had been meaning to look into with all my free time.

Could this be the problem?
viewtopic.php?t=8883

Maybe at very very light loads the IPW is getting down to a point where this comp is being applied therefore forcing your AFR down due to the extra fuel...???

Mine was doing that as well with a shared MAF value.... I turned down the Low injector compensation to activate only under 1750 and it made a difference in quite a few areas of the rpm range. only place that is still a pain in the butt is 2500rpm... it has somethign special right there on all subarus ive tuned as well as mine... its like this dead spot..... especially when you get big cams... but put a hydra or AEM in there and there is nothing... wonderingi what is transitioning there...