Jump to content
LegacyGT.com

Time to identify once and for all the source of my FKC


xt2005bonbon

Recommended Posts

D learning will apply closed loop corrections to open loop. Generally this is a big no-no, since it will richen or worse, lean out, your WOT portion of the map.

 

My WOT AFR's would vary day to day before I discovered D learning and disabled it. I believe Subaru did this incase something goes wrong in MAF or fueling, but on our modded cars D learning leaning out will kill the motor before the MAF or fuel system has a chance to fail.

05 LGT 16G 14psi 290whp/30mpg (SOLD)

12 OBP Stock 130whp/27mpg@87 Oct

00 G20t GT28r 10psi 250whp/36mpg

22 Ascent STOCK

Link to comment
Share on other sites

  • Replies 130
  • Created
  • Last Reply
So it is a good thing to have that D learning disabled then. OK. Interesting.

 

One thing to keep in mind is that your D learning may not actually be disabled, depending on your tune. I found it very unlikely I'd ever be in closed-loop and in the D learning ranges when I was running my Cobb SF intake, since I'd be on the verge of running in open loop at the higher MAF ranges of the C learning range. In other words, by the time I was in the MAF ranges that the D learning range covered, I'd be over 1g/rev load, and the car would be in open-loop (which means the car won't learn anything in the D range). The only way I'd be in closed loop in the D learning range would be if I was revving very high with little throttle in a low gear (let's say cruising on the highway in 3rd/4th instead of 5th/6th).

 

If you log and see that you're running in open-loop (either via the switch parameter, or the STFT being pegged at 0%) when in the D learning ranges, you'll likely never get a LTFT in the D range, even if it is "enabled." Disabling it, by the way, just means lowering the D max range value below the D min range value, which is a hacky way of making the stock ECU fueling code skip learning in the D range entirely.

Link to comment
Share on other sites

So it is a good thing to have that D learning disabled then. OK. Interesting.

 

There is an age old debate, some say Subaru did it for a reason so we should leave it, others say it's pointless and does more harm. I'm more in a latter camp, but I do acknowledge that there are possible benefits of leaving it enabled.

 

Benefits in theory are: If MAF or if fuel system goes bad, it will lean out around 1-5psi, so ECU will add up to 15% of fuel and it might prevent you from leaning out.

 

But in reality most of the time I see negative D learning (fuel being pulled). I think if it was only in one direction (only add fuel, but not remove), I would leave D learning enabled.

 

Another first hand experience is, a friend's MAF went bad. Was reading way lower g/sec then it should have been. Car was still dangerously lean, even with all things corrections adding fuel to max. In this case D learning being on may have helped prolong the engine, but it also masked a faulty MAF. If the car was allowed to lean out to even 14.7 it would have stumbled and told the owner(s) something is bad, instead of ~12.5:1 AFR's.

 

 

As for learning ranges: D learning on stock cars starts at 40 g/sec and ends at 80 g/sec. Here is a map trace from highway driving. 1.60g/rev was about 6psi of boost on my car. This basically shows that stock ECU tries to keep up to about 6psi in closed loop.

 

http://legacygt.com/forums/attachment.php?attachmentid=262439&d=1519672304

05 LGT 16G 14psi 290whp/30mpg (SOLD)

12 OBP Stock 130whp/27mpg@87 Oct

00 G20t GT28r 10psi 250whp/36mpg

22 Ascent STOCK

Link to comment
Share on other sites

Don't know if I am following everything :spin:. So if D is disabled (or hacked), we are sure that the ECU does not rely on what's been happening in trim C in order to adjust the open loop regime?

My current A-C are positive (e.g. ~5-7%) with B being almost pegged at 15%. Car runs fine but still...

Link to comment
Share on other sites

Correcto, once you bypass the C range's limits it should stay at 0% to redline. I personally have mine set to D=60 and I have my Max airflow range set to 59g/s.

 

A-B being pegged at +15% could be a vacuum leak or fuel related issue like clogged fuel filter or injectors, but it could also be an issue with the tune.

05 LGT 16G 14psi 290whp/30mpg (SOLD)

12 OBP Stock 130whp/27mpg@87 Oct

00 G20t GT28r 10psi 250whp/36mpg

22 Ascent STOCK

Link to comment
Share on other sites

OK. I am not seeing any misfire or anything like that. But I think I may have an issue with the inlet hose clamped at the turbo. It looks pretty bad i think. May have to change that. Although it is just strange that trim B would be the most affected. I think something is up with the tune too...
Link to comment
Share on other sites

When my blue tee hose blew off the evap on/off valve, I had pegged AF learning maxed on A (my A is as big as A+B on stock tune), no misfires or anything.

 

Tune would only be an issue if your MAF scale is not stock. I'll look to see what all is disabled on your tune later.

05 LGT 16G 14psi 290whp/30mpg (SOLD)

12 OBP Stock 130whp/27mpg@87 Oct

00 G20t GT28r 10psi 250whp/36mpg

22 Ascent STOCK

Link to comment
Share on other sites

Yeah. You don't have to but thanks for that.

 

I am 99.999% hoses connected to the blue tee are fine. I just checked two weeks ago while I was chasing almost pegged negative fuel trims across all ranges (except D). Turned out to be bad connection between MAF wiring harness and MAF itself.

Link to comment
Share on other sites

Looks like your ECU flash doesn't have the AF1 learning ranged defined, thus I can't see if it's set lower, you can ask for your ROM to be defined here: http://www.romraider.com/forum/viewtopic.php?f=40&t=8121

 

Your A/B/C/D Ranges are stock though.

 

Your MAF scale is not stock either. Looks like it might be meant for an intake of some sort (more then just a drop in filter).

05 LGT 16G 14psi 290whp/30mpg (SOLD)

12 OBP Stock 130whp/27mpg@87 Oct

00 G20t GT28r 10psi 250whp/36mpg

22 Ascent STOCK

Link to comment
Share on other sites

See "A/F Learning #1 Airflow Ranges" table.

 

I've seen that post too, I touched on those points couple posts back too. While I agree on on the reasoning, I think this is another one of those Subaru over-complicating things. This is the only car that I've dealt with that applies trims to open loop, kind of like it's the only car that forcefully stays in closed loop in boost (CL/OL Delay).

 

If the D learning was only allowed to add fuel (safest direction) then I would probably leave it on, but since it can pull fuel too that makes it more dangerous then a possible part failure.

 

Update: I like to reference newer Legacy's tunes since Subaru had time to learn new tricks. 2010 LGT has D learning set to 80g/s, but they did also raise the Max ceiling to 120g/s, so closed loop way into boost now. Also notice that they allow up to 25% of fuel adjustment (which also means up to 50% fuel increase/decrease before CEL is thrown).

http://legacygt.com/forums/attachment.php?attachmentid=262456&stc=1&d=1519705154

1437334477_2012AF1Learning.png.2f3aaa8115247f9ec4ebe88696444763.png

05 LGT 16G 14psi 290whp/30mpg (SOLD)

12 OBP Stock 130whp/27mpg@87 Oct

00 G20t GT28r 10psi 250whp/36mpg

22 Ascent STOCK

Link to comment
Share on other sites

Looks like your ECU flash doesn't have the AF1 learning ranged defined, thus I can't see if it's set lower, you can ask for your ROM to be defined here: http://www.romraider.com/forum/viewtopic.php?f=40&t=8121

 

Your A/B/C/D Ranges are stock though.

 

Your MAF scale is not stock either. Looks like it might be meant for an intake of some sort (more then just a drop in filter).

 

Huh.

 

Where did you see that the A/B/C/D ranges are stock?

 

I do have a K&N drop-in filter.

 

Also, I stumbled upon this post regarding that trim D disabled. Looks like Invar is against it. Thoughts?

 

http://legacygt.com/forums/showpost.php?p=4298782&postcount=23

 

 

The definition for the "A/F #1 Max" value is trivial. Add 4 bytes to the address of the D learning range to get the max. Incidentally, the min learning is 4 bytes ahead of the A learning range. In other words, for the AZ1G201G ROM detailed in that romraider thread:

 

C7AE8 - A/F #1 Learning Min Threshold

C7AEC - A/F #1 Learning Table (This is the A max/B min)

C7AF0 - A/F #1 Learning Table (This is the B max/C min)

C7AF4 - A/F #1 Learning Table (This is the C max/ D min)

C7AF8 - A/F #1 Learning Max Threshold

 

The original definitions for the A/F #1 Learning table is silly since I assume the min/max thresholds are always adjacent to the rest of the table in all the ROMs.

 

Invar is saying the same thing, where there are cases the learning D could be safe. However, in actuality, as covertrussian pointed out, there's also the common case where learning D will learn negative trims, and then continue to pull fuel outside of closed-loop operation (which can be dangerous). I personally leave it enabled, but lower the threshold down from 80g/s to avoid learning in spots where the combination of my MAF curve and engine load compensation aren't well tuned (i.e. low gear, high RPM, where the car might actually be in closed loop at higher MAF operation).

Link to comment
Share on other sites

btw, I want to thank the both of you, covertrussian and solidxsnake for the info you keep providing.

 

Back on topic :redface:, I will probably go ahead and try what you've suggested at the beginning covert (decrease timing in specific cells). I'll see how it goes. I should note that the 06 has very similar timing table and I never see -5.98 FKC. But I do see sometimes -3.87 FKC in a similar load/rpm range..

Link to comment
Share on other sites

When in doubt always refer to the stock timing map, which is actually very aggressive. As you do any mods to the car, it should actually need less timing instead of more, let me explain why...

 

Think of igniting timing in as total time for fuel to burn: The more ignition timing the more time the fuel has to burn, as you expect the less ignition timing, the less time the fuel has to finish burning. This controls when (at what crank angle) the peak combustion pressure happens at. The optimal peak combustion pressure is said to be at 15* After Top Dead Center (ATDC).

 

As you replace parts with more efficient parts (ie free flowing exhaust, turbo, or intake) the more efficiently the engine can drag air(oxygen) into the cylinder. Since there is now more oxygen in the cylinder the mixture will burn faster, and hit the peak combustion pressure earlier. If you keep your ignition timing the same, your peak cylinder pressure might now happen at 10* BTDC instead of 15* ATDC. This means mixture wants to push down while the cylinder is still moving up, this creates a lot of pressure that the other cylinders now need to waste power to overcome (to visualize this at home: take a turkey syringe without the needle, cover the spray hole and try push the plunger up). This is where you need to reduce your ignition timing to move the peak cylinder pressure back to happen at the 15* ATDC crank angle again.

 

This is why you will usually run less ignition timing but make more power on our modded cars compared when compared to the stock tunes. If you see more ignition timing then stock, you either reduced the engine efficiency and tuner had to increase timing, or your tune is bad.

 

TL;DR: The whole point of ECU tuning is finding that right combination of AFR and timing to give you that peak cylinder pressure at around 15* ATDC.

05 LGT 16G 14psi 290whp/30mpg (SOLD)

12 OBP Stock 130whp/27mpg@87 Oct

00 G20t GT28r 10psi 250whp/36mpg

22 Ascent STOCK

Link to comment
Share on other sites

Thanks for the info. Appreciate it.

 

Random question: is your LV consistently free of FLKC?

 

Mine generally only gets FLKC in two cells (rarely, might I add): the lowest load/RPM cell and the highest load/lowest RPM cell. The lowest load/RPM cell happens occasionally on the switch to open/closed loop, since my tune is not well refined and fueling is all over the place in that area. The highest load/lowest RPM cell picks it up sometimes, I assume, when I hit the gas a bit too much at very low RPMs in higher gears (instead of downshifting). The car probably needs some work to get it in tip-top shape, and the tune also isn't well-refined, but I'm lazy and I think it's safe enough so I haven't been motivated to do any work on in :lol:

Link to comment
Share on other sites

Thanks for the info. Appreciate it.

 

Random question: is your LV consistently free of FLKC?

 

On most days yes, only knock correction that I ever get is at very low load/rpm ranges (<1.25g/rev and <2krpm) and it's around -1*. Rarely do I have FKLC anywhere else on the map even though my map is tuned to MBT (where giving it any more timing didn't gain any more power).

 

With that said, I do get a false knock if I drive the car for more then 30 min in the city, and it actually happened last night:

http://legacygt.com/forums/attachment.php?attachmentid=262582&stc=1&d=1519915345

 

Reason I think it's false knock:

- No audible can of beans rattle, at -10* timing pulled it would have been audible

- Water temp was nominal ~178*F

- Post intercooler charge temp was barely above ambient at ~62F

- Engine Load was low 0.77-0.85/grev

- RPM was at 1800-2000

 

Reading more about it, and that load/rpm range seems to be a common false knock zone, probably has to do with the frequency of our engines at that rpm. See this RomRaider post.

 

 

I'm actually getting sick and tired of this false knock, I felt the car being much more sluggish on the last 4 miles of yesterdays drive and surely enough that's when I saw my FKLC being at -10*. Thus I'm raising my FBKC and FLKC listing ranges to be above 1.0 g/rev. Even Subaru did that For 2007 LGT's, with FLKC starting to listen only after 1.30g/rev, but FBKC is about the same as our 05's.

2013244235_0228-02_57f_-10flkc_after24mCity.png.d31767f9449bfab0e16787277352d097.png

05 LGT 16G 14psi 290whp/30mpg (SOLD)

12 OBP Stock 130whp/27mpg@87 Oct

00 G20t GT28r 10psi 250whp/36mpg

22 Ascent STOCK

Link to comment
Share on other sites

-10 is still a lot though.

 

It is, but if it's false, it's false. It went from -2 to -4, stuck there for a while, then just jumped to -4 to -10. It could also be a timing retard induced knock, I've seen this happen on my Nissan when I wasn't running enough timing. One knock count would happen, ECU would retard more timing, more knock happened, and ECU pulls more timing, ends up spiraling down until your running negative timing. Once I increased map timing to be higher, no knock at all :lol:.

 

Basically what happens is, when you retard timing, more raw fuel ends up in the exhaust, where it finishes burning or explodes in the exhaust (this is why retarded timing can cause high EGT's). This can lead to false knock since noise is happening by the engine still. Fun fact: This is also how timing based anti-lag systems work, they retard timing so much that raw fuel ends up exploding in exhaust and spinning the turbo up.

 

 

How do you get the 'post intercooler' charge temp? Did you install a temp sensor there?

 

http://legacygt.com/forums/showpost.php?p=5755331&postcount=1206

 

And you're OK with such a high fuel correct at idle? Tune related or actual vacuum leak?

 

I'm not, I fixed a ton of vacuum leaks weeks ago and have been in negative trims. I saw it being high again this week, I tried to find vacuum leaks using carb cleaner (which worked like magic last time). Haven't found anything, so it might be fueling related :mad:

05 LGT 16G 14psi 290whp/30mpg (SOLD)

12 OBP Stock 130whp/27mpg@87 Oct

00 G20t GT28r 10psi 250whp/36mpg

22 Ascent STOCK

Link to comment
Share on other sites

Eh Eh the -5.98 popped up this evening :lol:. After a closer look, timing was apparently already pretty 'low' when it registered knock. I need to take a look at the table now.

 

 

KNOCK DETECTED at 2018-03-01 10:10:57 PM

AFR: 13.9

Boost (psi): 0.20

Boost Target (psi): 3.1

Coolant (degF): 192

FKC (deg): -5.98

FLKC (deg): 0.00

Fuel correct (%): -0.0

Injector DC (%): 15.6

Injector PW (ms): 7.4

Intake (degF): 55

Load (g/rev): 1.13

MAF (g/s): 47.54

Misfire cyl 1: 0

Misfire cyl 2: 0

Misfire cyl 3: 0

Misfire cyl 4: 0

RPM: 2522

Timing (deg): 15.0

VVT int L (deg): 10

VVT int R (deg): 9

*fuel learn A: 3.80

*fuel learn B: 9.90

*fuel learn C: 5.40

*fuel learn D: 0.00

 

 

KNOCK DETECTED at 2018-03-01 10:10:56 PM

AFR: 13.9

Boost (psi): 0.60

Boost Target (psi): 3.2

Coolant (degF): 192

FKC (deg): -3.87

FLKC (deg): 0.00

Fuel correct (%): -0.0

Injector DC (%): 15.5

Injector PW (ms): 7.4

Intake (degF): 55

Load (g/rev): 1.14

MAF (g/s): 47.54

Misfire cyl 1: 0

Misfire cyl 2: 0

Misfire cyl 3: 0

Misfire cyl 4: 0

RPM: 2510

Timing (deg): 15.0

VVT int L (deg): 10

VVT int R (deg): 10

*fuel learn A: 3.80

*fuel learn B: 9.90

*fuel learn C: 5.40

*fuel learn D: 0.00

 

 

KNOCK DETECTED at 2018-03-01 10:10:56 PM

AFR: 13.8

Boost (psi): 0.70

Boost Target (psi): 3.3

Coolant (degF): 192

FKC (deg): -2.11

FLKC (deg): 0.00

Fuel correct (%): -0.0

Injector DC (%): 15.3

Injector PW (ms): 7.7

Intake (degF): 55

Load (g/rev): 1.19

MAF (g/s): 47.14

Misfire cyl 1: 0

Misfire cyl 2: 0

Misfire cyl 3: 0

Misfire cyl 4: 0

RPM: 2379

Timing (deg): 16.5

VVT int L (deg): 10

VVT int R (deg): 10

*fuel learn A: 3.80

*fuel learn B: 9.90

*fuel learn C: 5.40

*fuel learn D: 0.00

Link to comment
Share on other sites

It is the 'timing' variable that shows up in btssm. So I would say total timing. And this -5.98 does occur usually above 2-2.5k rpm.

 

I just checked the total timing table and around that load/rpm range, total timing is supposed to be 19-20deg. Interesting that the app shows 15deg...

 

By the way, how about the vvt angle? Is the reported 10deg somehow related to the total timing?

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.




×
×
  • Create New...

Important Information

Terms of Use