As from what you all told me, you can't. What I'm saying is disable closed loop at low loads, run lambda>1. Then at elevated load, switch to open loop and run fueling as suggested here via KR, BTS or however you like.

I forget the map name to disable closed loop, but Rick mentioned it in the post I referenced above ^^^

Edit: Found time to look and see what map I was referring to; "RLLRUN". Is this idea not feasible? Somebody enlighten me.

Sorry, mentally flip flopped the maps. Obviously the screenshot tells you which map is modified.

I am totally new to this, but I was attempting to read the function and data sheet and noticed the following on p. 676 regarding the stochiometric ratio (it seems to imply that it can be adjusted (rich or lean) away from 1 through LAMSBG_W [fuklamsbg (1515A in 4Z7907551R)]:


Variable Source Type Description
LAMSBG W LAMSOLL EIN Desired Lambda limitation (word)
NMOT SWADAP EIN engine speed
RK2 W GK AUS relative fuel mass Bank2
RKA2 W LRA EIN additive adaptive correction of the relative fuel amount bank 2
RKACO W EIN Additive correction of the rel. fuel mass for setting idle-speed CO
RKA W LRA EIN additive adaptive correction of the relative fuel amount
RKTE W TEB EIN relative fuel part of the purge control
RKUKG W ESVST EIN rel. fuel mass transition compensation
RK W GK AUS relative fuel mass
RL SWADAP EIN relative air charge
RLP W BGRLP EIN rel. air charge predicted for injection calculation (Word)
FW GK 13.10 Fixed Values
Parameter Value Description
FB GK 13.10 Detailed description of function
The function GK calculates the relative fuel mass rk w necessary for the predicted relative air charge rlp w of a
cylinder, for combustion at Lambda = 1.0. The standardisation of the two variables rlp w and rk w was chosen such that, at
100% air charge 100% fuel is also required for combustion with the engine at operating temperature and where Lambda = 1.0.
The stochiometric ratio of air mass to fuel mass in the cylinder is designated here as lambda combustion chamber,
as opposed to lambda sensor, which is measured on the lambda sensor and may deviate from lambda combustion chamber due to the
introduction of secondary air. In normal engine operation lambda-combustion chamber = lambda-sensor.
The function ESVST delivers the pre control values fgru, fst w, fnswl w, fwe for lambda-combustion chamber = 1.0.
This lambda combustion chamber can be shifted to "rich" or "lean" by way of the desired value of lambda LAMSOLL to improve
engine running (engine protection, lean running limit) or to comply with exhaust specifications.
The function GKRA includes mixture adaptation (rka w, fra w), canister purge (rkte w), canister purge diagnosis and
lambda control (fr w).
The continuous lambda controller controls each bank to the required lambda (lamsbg w, lamsbg2 w), which is
formed in the function overview LAMSOLL.
During secondary air the input for lambda controller is lamsbg w =lamsons w,lamsbg2 w=lamsons2 w, lambda-combustion chamber
is # 1.0.

The point is that open loops' AFR could really be anything several % points either way of what you think it is based on fueling tables. You will run few tenths of lambda either side of desired amount... and god forbid your maf curve, or more precisely the *clusterfaq* (had to use that word...) KFLF and KFKHFM is (some insane person calibrated these), doesn't match your intake. If you changed ANYTHING on the intake path, MAF diameter, turbos, intake pipes, cone filter, etc, these maps essentially don't apply anymore. They are calibrated for stock intake. When you forfeit using closed loop, you start to exclusively rely on them.

If you really want to see how inaccurate they are, log the car and do some mild acceleration through the RPM range and look at lambda corrections... you'll be scared once you realize that if not for the closed loop, you'd be hanging 10+ % on either side of lambda 1.0... that's 13...16 afr. Closed loop ensures that the system self regulates to stick to the o2 switching point which is 14.7 with narrowband o2 sensor...

One more thing. If you see ANY long term fuel corrections, it means that you have exceeded 10 or 15% (I forget what the threshold is) of lambda corrections in the past.

So... if you force open loop for all the time and you flash that, then it means that:

1) you disabled clsoed loop
2) your system will never store long term LTFTs
3) you might be running REALLY off the target fueling.

Just make sure you wideband o2 hooked up to monitor real AFRs.

On the side note, the KFLF and KFKHFM are a mystery to me. More precisely, the mystery is why they subtract from one table and add to the other creating overly complicated relationship where just doing KFLF would do.

I know, somebody will say "but if you just do KFLF then you will inbalance something in the ECU as you're playing with fuel mass correction etc, etc". Well, just go and look at RS4 factory map. The way they did that one makes far more sense. They left KFKHFM alone (defaults to 1.00 for all points) and only adjusted KFLF. You can finally see the full picture when looking at just one map.

I think I will reset my MAF corrections map as well and only do KFLF.

Cheers.

KFKHFM is 1.00 in the RS4 tune exactly like you said.  Also FKKVS is 1.00 in the RS4 tune.  As you suggested, I agree that KFLF is the best way to tune corrections.  My belief is that the correct way to do it is as follows.  Setup a full logging session, ideally 30+ minutes of 10 Hz for all the relevant variables.  Take the following:  Corrected Load vs RPM vs Average of Lambda 1 / Lambda 2.  Thus you'd have a very lengthy time series of these three variables synched up.  From here you take a two-dimensional curve-fitting tool, which bins and self-groups the data to build a surface.

This surface will then have the axis of the KFLF tables, one can visualize not just the local average but also the standard deviations of lambda corrections.  The necessary changes will be obvious and relatively smooth.  Then take this this map and create a blending table.  The blending table takes x% of the value from the "theoretical curve surface" and (100-x)% from the 1.00 constant table.  This is a goal-seeking technique known as "mixing", we'd likely start with x=50.

This multiplicative inverse of this blended table with x% is your new KFLF.  The likelihood is you'll need to recenter your LTFT's after the first KFLF revision.  My guess is that further KFLF repeats of this procedure and effective refinements using this methodology will require progressively less KRKTE/TVUB adjustments.

Heh, no fault there.   MATLAB all the way.    If you get me those vectors, I'll attempt it...    i can try to use an old log today.

Your method assumes KFKHFM = 1.00 everywhere is correct... which is generally far from the case

You are tuning fueling assuming your MAF data is correct (which isn't true).

In fact, I used a variant of what you suggest to tune my KFKHFM table, assuming the fueling is correct (which also isn't true)

If you really want to do it "right" you have to figure out how to tune one without depending on the correctness of the other.

Ha. No kidding