My new FR-S!

[serialk11r]

So this morning there was some traffic so I didn't get any constant highway cruise readings, and after poking through my old logs I realized the data there wasn't actually clean either :/ I'll be taking a longer drive tomorrow and logging both stock and modified VVT map tunes on flat (hopefully) open freeway and figure out what's really going on.

I've been trying to think of a way to outsmart the rear O2 sensor but I just found out it's a narrowband, which means it'll be 200 bucks for a narrowband emulator to trick it :/ The other thing I could do is set the A/F learning mass airflow limit higher to basically disable the downstream sensor. There's a value called "Max range B/Min range C" which is set at 12g/s, which I think if I bump up to 17 will avoid learning because highway cruising uses about 16 However, there's a "Max range C/Min range D" value of 20 which I'm too scared to touch.

I could also outright disable learning by setting learning limits to 0%, but that makes me feel uncomfortable too.

These new ECUs are too smart! Slightly older models could either be commanded to run lean in open loop, or tricked into it by triggering open loop -_- I suppose I could set a majority of the 14.7 cells to 14.6 to trigger open loop everywhere but I would literally have to disable closed loop operation everywhere or else it starts learning again, and I don't really feel comfortable having it run open loop everywhere except idle.

[serialk11r]

Looks like a 0.5psi increase in MAP over the stock map, which I calculated to be about 150W in savings. Kind of underwhelming, considering that my light bulb swapping probably saved just as much power at the crank, but every bit counts.

I just realized that I scaled the O2 sensor the wrong way on the lean map though, so I was actually running a little rich LOL. Woops. I'm debating whether I should make a map that disables positive fuel trim A/F learning so that I can put E20 in the car and have it pegged at a lean burn. It feels like kind of a sketchy thing to do though.

The map I'm flashing tomorrow is going to be 15% AF learning limit, 15g/sec learning range B (so I can cruise at 60 without triggering the secondary sensor correction), and running 5% lean of stoichiometric. The new feature I'm going to try is increasing the DFCO throttle opening (overrun mass airflow target). Sometimes I want to DFCO but it brakes the car a little too much so I end up kicking the clutch in once in a while which gets annoying. With less engine braking effect, I can DFCO in 6th gear a little more and just drop down to 5th if I need more braking going downhill. The other advantage of this map is that if I take the car on the track or go through heavy hilly terrain, the greater airflow will cool the cat off a bit.

[serialk11r]

Update: Even setting it a little bit lean failed miserably lol. AFRs quickly dropped really low, even with the learning limits changed. Seems like lean burn isn't going to happen :/ I'm going to just do what I can and set the CL fueling target tables to slightly lean so the engine tends to run slightly lean rather than slightly rich, then set more cells to 14.7 in the primary map.

Also, I realized why overrun mass airflow target is the way it is. Today I was driving around with my map and I realized that below 2000rpm I set the target airflow so high that the revs would hang for multiple seconds in low gears because DFCO is disabled -_-

I just noticed that the minimum value you can put into the VVT table is -40 deg. I'm cranking it to -15 first to see if it'll actually do it.

[BabyDiesel]

Have you considered bleeding voltage from the MAF as it is going to the ECU? That is how Tygen1 enabled lean burn on his 98 ZX2.

Of course, you have a completely different beast. It is still worth a shot to see if it is possible!

[serialk11r]

Originally Posted by BabyDiesel

Have you considered bleeding voltage from the MAF as it is going to the ECU? That is how Tygen1 enabled lean burn on his 98 ZX2.

Of course, you have a completely different beast. It is still worth a shot to see if it is possible!

The problem here is that the post-cat O2 sensor is the final authority on fueling, and it's a narrowband sensor that can't be manipulated through software. I can change anything I want, and people do change MAF scaling as well as O2 sensor scaling (useful for turbo setups where the stock sensor's inaccuracy can be a problem), but at the end of the day at cruising conditions the post-cat sensor will figure out what stoichiometric really is and whatever scaling you set on the MAF or wideband O2 will be reverted in the long term fuel trim.

I just flashed my car with the -15 deg intake map though, and the logs say the engine is cooperating! I'm going to try -20 deg on the way back from work, but I may be running out of VVT adjustment range. I set the entire AF compensation table to positive and the engine now likes to run slightly lean more often. I'm really really hoping these changes can get me to 45mpg cruise.

[serialk11r]

Update:

Drove home on -15 deg intake timing today, didn't notice much. Flashed a -25 deg map to see what it would do and the log registered some -25 but mostly -22.5 deg, which may be the limit. I'm going to try -30 next but I'm not sure if it'll listen to me.

The good news is that the MAP values look very high at -22.5 and -25, around the 8psi range, though the data is noisy and I need to log some highway cruising to get clean data. I got rid of some knock compensation and leaned the fueling out a little more, so hopefully that sees results too.

If this doesn't work, I may try bumping the exhaust cam retard up a little more to get some overlap back. I doubt that pumping losses from the exhaust valve opening too late would be a problem.

[kafer65]

I hear that the narrow bands are impossible to tune by since they swing so quickly it just becomes too noisy to work with. Ten years ago I could introduce a resistor into the wiring of the oxygen sensor to make it a blunt instrument. That's on a Forester XT. That allowed me the flexibility to go up to 14.7:1 (Map limited) in more places and get way more timing requests met (I think). It would make a lot more torque anyway and I got 1 more mpg out of it. It was all too limiting and I realized the power wasn't edifying anymore and I wanted to mess with diesel engines for awhile.
I think yours is an interesting balance of the mix of fuels with different octanes and BTU content. For example : If you're limited to 14.7:1 AF and you may be able to sustain 16:1 so you add alcohol which needs to be a little fatter 14:7 looks more like 16:1 (with that mix) and you get higher octane which helps retard knock.
I've been thinking about this on my Beetle since I can't get the ECU to make it lean enough off boost and rich enough on boost. If I run E85 where its rich now on pump gas on E85 it may be much closer to stoich. Then when boost comes on I can do a 50/50 mix in the charge pipe to cool and richen up top.

[ksa8907]

Congrats on the purchase! My brother in law is leasing a '15 fr-s. His is some limited production trim, only 1500 made. Has all the trd parts and body kit on it, looks awesome.

[serialk11r]

Originally Posted by kafer65

I hear that the narrow bands are impossible to tune by since they swing so quickly it just becomes too noisy to work with. Ten years ago I could introduce a resistor into the wiring of the oxygen sensor to make it a blunt instrument.

Yea like I said, what you need is to plug in a wideband sensor and a narrowband emulator. Costs around 200 but it'll definitely pay for itself within a few years in gas savings in most cases, though it obviously works best if you have the hardware to tune the ECU already so you can adjust the maps appropriately.

Update, -25 is the most the intake cam will move, I tried putting -40 into the table but it only gave me -25, so -25 it is. I attempted to add exhaust timing to increase EGR but it started stumbling at 2000rpm in neutral! I kept 5 degrees in the cruising load range cells because I figure the intake cam is retarded enough to make it work.

My gas + ethanol blending experiments are progressing pretty well, I had about 4.5 gallons left in the tank which was like 30% E85, the rest was 91. I filled the tank all the way with 89 (8.5 gallons or so) reducing the ethanol content down to probably something like 17%, and still zero knock detected at max load . I basically overpaid around 25 cents in fuel energy content for the E85 but got a whole ~92 octane tank for over a dollar less than a full tank of 91. Good tradeoff.

It's too bad the E85 station I fill at has very expensive gas prices, otherwise I could do 20% E85 + 80% 87 blends conveniently and at lower cost for the E85.

[wdb]

Originally Posted by serialk11r

My gas + ethanol blending experiments are progressing pretty well, I had about 4.5 gallons left in the tank which was like 30% E85, the rest was 91. I filled the tank all the way with 89 (8.5 gallons or so) reducing the ethanol content down to probably something like 17%, and still zero knock detected at max load . I basically overpaid around 25 cents in fuel energy content for the E85 but got a whole ~92 octane tank for over a dollar less than a full tank of 91. Good tradeoff.

It's too bad the E85 station I fill at has very expensive gas prices, otherwise I could do 20% E85 + 80% 87 blends conveniently and at lower cost for the E85.

Those are interesting experiements but speaking for myself I think I'd last about 2 tankfuls before saying "too much hassle!" and just filling up at a single location. (Although it might be a fun way to amp up a tank for the STi, hmmm....)