2004 Subaru STI SD comp question

I am tuning a 2004 STI on Cobb Speed density. I am logging fuel pressure, and it is consistent. I completed my VE table tune with plenty of data to get great tight fuel trims at idle. The next day, I'll start the car and it will run very lean, like 10-15% lean. My IAT's are within 10 degrees of ambient.

THEN, i'll make adjustments to the VE table to tighten the trims again. Drive it around for 10 minutes, then the car will be RICH.

I was considering adjusting coolant temp comp table for VE, but it doesn't make sense, since the coolant temp before I start driving and after is similar.

I have to be missing something. Any ideas?

Have you tuned the after start fueling? The VE table is for a baseline kind of condition (basically, engine warmed up, not too hot, not too cold). Typically that might be something like a 25C to 45C air temperature. Then you have to adjust compensation tables on top of that. That includes after start fueling, IAT compensation, water temperature compensation.

You also have post start enrichment tables that may need to be adjusted. What injectors do you have and what fuel are you using? Is it E85 or normal E0/E10 ?

Are you using MAF mode, speed density mode, or hybrid mode? How did you build your VE table? Where did the injector data come from? Why are you speed density anyway? Do you really need to go that route for your setup or is this more of a science experiment?

Whew, lots of questions!

These symptoms are after all of the enrichment has decayed away.

I’m using ID 1300x2s, ID data, And it’s a flex fuel setup.

By build VE table I’m not sure what you mean, but I used a combo of closed loop fuel trims and comparing commanded fuel final and wideband data.

The car is setup to run maf, sd, and hybrid. It pulls enough air mass to be right at the top of a 73mm MAF, so I recently started the SD tuning.

The symptoms I’m describing do not happen in MAF mode. Only in SD.

Cobb gives you iat, baro, and ect comp tables for the SD model. I am only using iat comps. The fueling changes are not related to ect.

im beginning to think it’s related to the temp of the intake manifold. (Btw, my iat sensor is in my charge pipe right before throttle body, and it’s not heatsoaked.)

Why don't you set it up for hybrid mode based on MAF voltage, so that you only have to use it when MAF voltage is exceeded? Then you won't have to spend hours and hours trying to populate a gazillion cells in all the various correction tables.

Ha! I do plan to run it in hybrid mode.

But I was trying to understand why I couldn’t get it to run perfect in SD.

I have most of the table dialed, it’s just the idle area where I see this weirdness.

I mean, I’ll tune it one day and it’ll be driving great, super smooth accurate afrs all around, then the next day, idling lean by 20%. Then I’ll correct after car is up to temp, drive it around for 20 minutes, then it’s pulling 10%!

You can't get it to run perfectly in SD because it's a lot of work to run in speed density in terms of setting up all the calculations/algorithms and tuning them, then testing under every circumstance to make sure it works right. It sounds like a flippant or dumb answer, but Subarus were never designed for speed density and one guy at Cobb wrote a few algorithms to insert in there. Look at a modern Ford or BMW. They have speed density with multiple air temperature and pressure sensors in the intake tract, and a gazillion compensation tables to account for cam position, air temperature, fuel temperature, etc etc.

You have to lower your expectations. Cold start accurate AFR control (without fuel trims swinging back and forth) is REALLY REALLY hard to do and it is very sensitive for emissions. That's why OEMs have entire departments writing code and tuning just that narrow area of operation. You can grind it out for hours and hours trying to adjust the VE table and the combination of IAT and Coolant temperature compensation tables to get your fuel trims closer to a little closer to 0, you can rely on the closed loop feedback like you are now, or you can just run it with a MAF the way it was designed.

I mean I know a guy who literally got paid six figures to tune the first 30 seconds of a cold start on the current generation Corvette during original development 6-7 years ago. That's all he did.

On most standalones lots of people just focus on driveability of cold start/idle/drive away. Deviations from Lambda 1 in that area, which is what your drifting fuel trims represent, are mostly just tolerated because emissions aren't that important to most people and standalones don't have all the closed loop learning zones that the Subaru ECU (with shoehorned speed density) has.

I really appreciate your answer here. With my level of experience, it’s hard to know what’s acceptable! I’ll just focus on the higher map areas.

Following up on this. I recently solved some weird low load fueling issues by zeroing out my per cylinder fuel comps. Made sense to me since I have a built motor with a completely upgraded fuel system etc. The conditions the comps were created in, were long gone.

Anyways, I thought I'd go down the SD rabbit hole again since I thought my previous issues could be related to these comps. Turns out they weren't.

But upon further investigation, I zeroed in on the Coolant temp affecting my VE and this seems to be the issue. I logged coolant temp vs fuel trims and noticed that the car was running lean during warm up, but once up to temp the trims were great. I plugged my errors in the SD coolant temp comp tables and the car now warms up like a champ with tight fuel trims.