DIY: Retuning the 2.5i

 

In this thread I will be covering my tuning adventures with the 2.5i (EJ253) on my 2012 Outback. I plan on tuning the car for first no knock, then best power on 87 and 93 gas with all stock components.

 

Ignition Timing:

87 Gas Tuning

 

Fueling:

Disabling Open Loop Learning

These cars are really not happy with 87 gas, the timing maps are way too aggressive. The only time I've been able to keep the IAM at 1.00 is with 93 gas, but even then again it would still go down to .6 in hot Florida weather.

 

Timing & MBT

Highlighted are the cells that control IAM, it's actually a very short range that ECU is looking for knock in to control the fate of the rest of the tune.

http://legacygt.com/forums/attachment.php?attachmentid=260562&stc=1&d=1515594698

 

 

Here is where it becomes fun, finding the Minimum Best Timing (MBT). The principal of MBT tuning is, you use the least amount of timing that gives you the best torque. You know that you are past MBT when your power plateaus or you start loosing power.

http://legacygt.com/forums/attachment.php?attachmentid=260566&stc=1&d=1515594921

• Stock Tune - IAM(0.321): First I did a base run with the car the way it is, at this point it had an IAM of 0.321, this gave about 109whp, 119ft-lbs. This also is the richest run, this is because of D learning.
  
• Stock Tune - Full Timing: Next I disabled IAM and added all of the KCA timing to the main tables. I saw major knock at 5k RPM, FBKC pulled 5*, which is the max ECU allows it to pull stock. This got me 115whp, 125ft-lbs.
  
• Stock Timing -5deg: I pulled 5* across the whole map, still some a little FBKC at 5.5k, this got me 116whp, 124ft-lbs
  
• Stock Timing -10deg: By this point I was expecting to loose power, since I haven't yet I pulled another 10*, this got rid of all of the FBKC. By this point I was expecting to loose major power instead I gained some 117whp, 123ft-lbs. This just shows how far past MBT the stock tune is!
  
• Stock Timing -15deg: At this point I'm pulling 15* and finally started to loose power, this is crazy. This map got me 109whp, 110ft-lbs
  
• Tuned: Finally I went back to the 10* map, and smoothed it out majorly. You can see how bad the stock map is vs the final product below. This final map got me 121whp and 128ft-lbs, the car pulls you into the seat really nicely now (seriously impressed me for an NA car).
  

 

Stock Map:

http://legacygt.com/forums/attachment.php?attachmentid=260565&stc=1&d=1515594839

 

87 Tuned Map:

http://legacygt.com/forums/attachment.php?attachmentid=260563&stc=1&d=1515594698

 

 

Dyno graphs of the changes above. These logs were all taken on 3rd WOT pull (to maintain power consistency).

The dotted line is not boost instead it's a mapping of the total timing.

http://legacygt.com/forums/attachment.php?attachmentid=260564&stc=1&d=1515594698

 

What's funny is, even at fully tuned 121whp, at 23% drivetrain loss, that's only 157BHP. Something tells me Subaru did their dynos with 93 gas that they don't even recommend for this car . It could be my motor at 100k is not as good as it could be, but the motor's compression test was excellent at 180-190psi. For this car to make the advertised 170BHP, it would have to be putting down around 130WHP.

 

 

Fuel

All but the first run were tested after the ECU was reset, which means ECU hasn't had a chance to make fuel adjustments. The first run is much richer because the ecu self learned in the D Learning slot, this is where ECU takes the closed loop learned value right before open loop and applies it to open loop (not a good thing). I didn't really tune the fuel yet, it's actually pretty decent stock after reset, before D learning makes it be turbo rich.

 

 

Fuel Economy

I haven't really tested the gas mileage of running 10* less, so far it doesn't seem like it hurt it, gauge is reading 22.8 with a bunch of WOT pulls and city driving.

Are you going to tune for 93oct at some point or just stick with the 87 cleaned up tune? I need to do this for my 2.5i again! Nice write up and visuals

Yup planning on doing a max power tune on 93 gas eventually too, who knows might be a better dollar per HP ratio then most bolt ons.

For what ever reason Subaru thought that it was a good idea to take closed loop AF Learning and to apply it to Open Loop fueling. This means the open loop section could be much richer or worse, much leaner then what the map calls for. You could really see this in my previous post's Timing Dyno:

http://i160.photobucket.com/albums/t188/covertrussian/Cars/12%20OBP/Tuning/87%20Tuning/12OBP%2087%20Gas%20Timing%20Tuning%20Dynos.png~original

 

The first run was before I reset the ECU thus it had +7% learned addition. All other runs were right after ECU was reset thus haven't had a chance to learn anything for the D range.

 

Looking around the tune I noticed that Min range D is set to 30 stock.

http://i160.photobucket.com/albums/t188/covertrussian/Cars/12%20OBP/Tuning/D%20Learning/AF%20Learn%20Ranges%20-%20Stock.png~original

 

Now AF 1's Max Learning limit was not defined for my tune, but thanks to this RomRaider thread, I was able to figure out the hex offsets for A/F 1 Max Learning. Using that I was able to see that Max Learning is defined to be 60 g/s, which normally at 80 g/s for turbo Subaru's. I then continued to set the D learning range to be 61, which would be right above the Max Learning limit, thus the ECU should leave D alone.

http://i160.photobucket.com/albums/t188/covertrussian/Cars/12%20OBP/Tuning/D%20Learning/AF%20Learn%20Ranges%20-%20D61.png~original

 

After a day of driving here is the learned values of before and after change:

http://i160.photobucket.com/albums/t188/covertrussian/Cars/12%20OBP/Tuning/D%20Learning/AF%20Learning%20Stock%20vs%20D61.png~original

 

Here are the dyno graphs of before and after. Red line is before the change right after ECU was reset, blue line is after ECU has learned, and the green line is D learning disabled about 5miles after reset.

http://i160.photobucket.com/albums/t188/covertrussian/Cars/12%20OBP/Tuning/D%20Learning/AF%20Learning%20-%20Dyno.png~original

 

Disabling D learning had another interesting effect, on a fresh ECU reset this car always had a lean spike around 3500rpm, well this change eliminated that lean spike! I'm thinking this was related to closed loop to open loop delay, either way I'm happy to see it gone, makes for a much more linear fuel curve.

All nice work. This may be an odd question but given the shift to COBB for tuning among the major turbo tuners where can us regular people go to get tuned? If I had a COBB AP for the 2.5i I would skate right over to Mach V but all I have now is OS.

 

I get questions from other owners about tuning is there any shops that will:

 

1. Tune a 2.5i

2. Tune via OS

 

?

 

 

Thanks

I would say your asking in the wrong thread (more on that below) and the wrong guy, I started to tune my own cars because I was in the same dilemma about 10 years ago when I couldn't find a local tuner for my SR20DET swap.

 

Decided to teach myself on how to do it instead, this thread is to help those trying to tune the cars themselves . I can see that my thread title is not specific enough, thus I asked the mods to change it to be "DIY: Retuning the 2.5i".

Thanks! And my intent was to may help someone out doing a search of the forum for 2.5i tuning.

 

I have been tinkering on my own tune as well started with MAFv during an attempt at 2.5RS intake now im altering tables to smooth out revs between shifts and OL/CL behavior.

Roger that, yeah I would like to keep this thread to DIY tuning, since a lot of vendors can already tune 2.5i's if a person is willing to pay them for it .

 

How well did the intake work? I'm considering one myself, but since my Outback is an off roading car I need it to be high water safe .

Roger that, yeah I would like to keep this thread to DIY tuning, since a lot of vendors can already tune 2.5i's if a person is willing to pay them for it .

 

How well did the intake work? I'm considering one myself, but since my Outback is an off roading car I need it to be high water safe .

 

 

Didn't work at all I couldn't get the MAFv to calm down across the powerband it merely (the MAFv spikes) merely moved up the RPM range.

 

Moved on to more responsive tweaks like throttle plate angle and requested torque.

So the biggest issue is MAF oscillations? I wonder if using OEM MAF housing would alleviate that, though I hear even removing intake resonators causes oscillations.

Roger that, yeah I would like to keep this thread to DIY tuning, since a lot of vendors can already tune 2.5i's if a person is willing to pay them for it .

 

How well did the intake work? I'm considering one myself, but since my Outback is an off roading car I need it to be high water safe .

 

Do the intake, and work on the Maf part. The intake with a nice tune will work very nice. I know I still have my setup (K Chamber) with fog light ramair tuned and the car is running GREAT!!! I do run 93, and with the cold weather the car is crazy.

 

can't wait to see the outcome covertrussian. I also have an outback . Car is strong for it's high miles.

 

Note: I would do fresh plugs, and fluids if possible, this made a big difference.

What are you using for an editor? I am using ROMRaider, but I don't know how to get the total timing editor if it even has that function. My 2012 stock timing looks a lot different than your table.

I use a mixture of RomRaider, ECUFlash and NSFW's Timing Editor.

 

There are 4 maps for timing (Low cam/high cam and cruise vs non cruise maps). Once you add the IAM based maps to them you'll get the values I got.

I installed the NSFW Timing editor and I see that my stock timing map matches yours. I had my car etuned 4 years ago, but it doesn't like 87 octane at part throttle. The tuned map has less timing at high loads like yours. The high rpm (4800+) looks very similar to yours but the lower RPM's have more timing. One thing it does have it a lot of timing at the lower portions of the IAM range. The knock correction advanced map is way different; it peaks at just over 3 degrees compared the 12 degree the stock adds back in (I assume that is how that works). All my 4 AVCS/non-AVCS cruise/non cruise timing tables are also all exactly the same, the stock stock timing maps have some differences at low loads/high rpm.

 

It looks like some have reading to do and then I need get out to do some dataloging and play around to see what my engine likes.

The knock correction map just adds displayed value to the map as long as your IAM is 1.0, if it's say .75, then it should use about 75% of the knock correction advance values.

 

All 4 shouldn't be the same, AVCS related maps are the high lobes (which are used 90% of the time). Regular maps (non-avcs) are for the low lobes that seem to be only active during cold starts and maybe slow acceleration. Either way the timing requirements are going be slightly different due to different airflows based on cam lobes.

So the biggest issue is MAF oscillations? I wonder if using OEM MAF housing would alleviate that, though I hear even removing intake resonators causes oscillations.

 

 

Can someone explain how MAF oscillations are a problem, how they affect drivability or tunability?

Can someone explain how MAF oscillations are a problem, how they affect drivability or tunability?

 

MAF oscillations are up and down jumps in maf voltages, when in the airflow should be fairly linear. So the MAF see it's sucking in less air, then more air, then less air. It's very possible that the airflow it self is being non linear, which is not good either, since we want consistent airflow.

 

This causes the AFR's to oscillate too, all of this leads to hesitations and other poor driving conditions.

MAF oscillations are up and down jumps in maf voltages, when in the airflow should be fairly linear. So the MAF see it's sucking in less air, then more air, then less air. It's very possible that the airflow it self is being non linear, which is not good either, since we want consistent airflow.

 

This causes the AFR's to oscillate too, all of this leads to hesitations and other poor driving conditions.

 

gotcha...that makes sense...its like it misleads the ecu into adjusting something that doesnt really need to be adjusted and upsets the balance of things...

gotcha...that makes sense...its like it misleads the ecu into adjusting something that doesnt really need to be adjusted and upsets the balance of things...

 

 

Also why you shouldnt mess with the intake tract on these cars unless you are able to scale the MAFv table appropriately and even then these EJ253 base Subarus can still exhibit problem.

Also why you shouldnt mess with the intake tract on these cars unless you are able to scale the MAFv table appropriately and even then these EJ253 base Subarus can still exhibit problem.

 

to borrow a line from a famous cartoon character...be vewwy vewwy kehhful..hehe.

 

All I did to mine was drop in a K&N element. But I am thinking about modding the intake side of the factory airbox to get more ram air effect at speed...wont make any difference in city driving of course, but might help highway mileage a bit. Also playing games with the different grades of fuel to see what effect they have without messing with the ecu...for my driving, changing down from 93 to 89 octane cost me about 2 mpg in overall driving. It still feels better than it does on 87 but not as good as on 93 from a seat of the pants perspective. Next step is to go all the way back to 87 and see how bad it gets.

 

I'm looking forward to seeing what the horsepower numbers are for the difference between 87 and 93. That may be the least expensive horsepower we will find for these cars. And its certainly the easiest to get.

to borrow a line from a famous cartoon character...be vewwy vewwy kehhful..hehe.

 

All I did to mine was drop in a K&N element. But I am thinking about modding the intake side of the factory airbox to get more ram air effect at speed...wont make any difference in city driving of course, but might help highway mileage a bit. Also playing games with the different grades of fuel to see what effect they have without messing with the ecu...for my driving, changing down from 93 to 89 octane cost me about 2 mpg in overall driving. It still feels better than it does on 87 but not as good as on 93 from a seat of the pants perspective. Next step is to go all the way back to 87 and see how bad it gets.

 

The drop in filters don't do anything. I measured only about 1 inH2O pressure drop reduction with K&N filter vs Stock.

 

I'm also working on a custom intake for the Outback, just needing to find a spare maf housing .

 

 

I believe it on MPG drop, though I did only measure a 1.5mpg drop between 93 and 87, but that's with the car not getting it's ideal MPG anyway.

 

I'm looking forward to seeing what the horsepower numbers are for the difference between 87 and 93. That may be the least expensive horsepower we will find for these cars. And its certainly the easiest to get.

 

I've been busy on the Legacy so haven't had the time or brain power to dedicate to finishing my 93 fine tune. After that I do want to do one more highway MPG test on 93 and then move back down to 87.

do you know of anyone who has tried using an accessport on a normally aspirated subie engine with good results? Seems like cobb has the tuning software already figured out as well as a convenient way to get into the ecu via the accessport. Admittedly, its a bit on the expensive side, but not much worse than a day on the dyno. And you can use it again and again. Just seems easy. I know they dont say it will work on the NA motors, but why wouldnt it? None of the boost parameters would come into play obviously, but I would think everything on the vacuum side of things would translate right over. Any thoughts?

Software is only a small part of it, the bigger hassle is defining every single ECU. Cobb probably doesn't want to waste a bunch of hours to define an ECU that only a handful would buy the Access Port for. Plus every time Subaru releases a code update for the car, a new definition needs to be built.

 

This is where Open source, aka Tactrix + Romraider, are a it. better bet. For my 2012, my ECU version was actually undefined, but Romraider guys already defined the latest version of the ECU, so I simply upgraded my code to that version.

 

Plus Cobb forcing a $100+ class just to get access to the tuning software pisses me off. I can't help Cobb guys with their tunes now, since neither of us have access to the software to make the changes. Thus I recommend open source to friends, since I can at least help with fine tuning.

 

None of the boost parameters would come into play obviously, but I would think everything on the vacuum side of things would translate right over. Any thoughts?

 

It's not that simple, motors themselves vary a lot, so do the actual firmware code. This puts all the tables at different addresses, aka needing a new definition per ecu.

Sadly that makes all too much sense, I am sorry to say. I would like to see what performance difference there is for example in 0-60 runs between 87 octane and 89 and then 93...with no other changes...then compare what I am getting doing that what you are getting with your tuning efforts...and if its a significant improvement, I might spring for the hardware to do open source. But it does scare me a bit...no rational reason for it really, but just dont feel like its as well sorted maybe as Cobbs solutions...but you made good points about all the different versions of ecu and no code for certain ones...its a quagmire for sure going that way too. But at least with open source there are some people using it who know their way around and are willing to help sort out problems...so that is surely worth a lot. What is the approximate cost for the tuning software and the linking cable?