NefMoto

Target lambda is set to one via a constant if B_falsh, B_lamlash, B_lamlshv, B_lamka, B_lamkh, B_lamnswl, and B_lambts are clear. Basically, if target lambda doesn't need to be controlled for a specific purpose then the constant sets it to one. Changing this constant should not interfere with functional control of target lambda.

The constant first gets moved to lamdiag_w, then ultimately to lamsons_w, lamsbg_w, lamvoa_w, and lambas. It can be limited based on tmot via LAMLGMTM.

I still have not figured out what all of the repercussions of changing this constant are, but it appears that lean target lambda will force open loop. It also looks like it may disable ATR unless LATRO is raised.

I was able to define it in my XDF and change the constant to 1.1. ME7Check reported checksums were corrected and comparing to the original revealed checksum changes.

I have defined for M box as follows:
0x3BFB2
16 bit LoHi
X*0.000244141

This is completely untested at this point, so attempt this at your own risk.

Functions that reference lamsons_w:
DLSH
DLSAHK
DLSV
BGMSZS
LRAEB
TEBEB
LREB
LRKA

Functions that reference lamsbg_w:
GK
Also lamsbg_w is used for axis SLX06TMUW.

Function that references lamvoa_w
ATR

Functions that reference lambas:
MDBAS
ZWGRU

Is this discussion for a narrowband or wideband ECU?

I've been running 1.07 lambda in closed-loop on both my wideband cars with no issue.

There are several, separate constants of 1 used throughout all those functions, but only one needs to be found and adjusted. The constant referenced in lamko-lamdsk is the one to adjust.  This is going off memory. I'll look at my file again.

Re-reading your post... that exactly what I've done. No issues. LAMLGMTM was high enough stock.

a smaller DLAMFAW plus one  increment below 1 on lamfa to bring target close to 1 when needed.

The information I posted was derived from the B5 S4 M box narrowband ECU.

Your input is appreciated though, I'm glad to see someone else has done this.

I think that the only difference may be that the wideband ECU can still operate in closed loop, but I believe that this will force open loop on the narrowband ECU. That may be a good thing though as the narrowband sensors cannot accurately represent these lambda values and calculated fuel mass is usually very accurate as long as injector calibrations are good and there are no hardware issues.

IMO this is much a better approach than changing USR or forcing open loop by other means and using KFLF or FKVVS. This way does not impede any other fueling functions and is only active when lambda would normally be stoich.

Unfortunately, I have a lot of mini projects right now and testing time is scarce.

I think trusting lean AFR to open-loop is a bit.. scary.

call me crazy.

Agreed, however, closed loop is just as bad if not worse because narrowband sensors cannot operate outside of stoich.

I look at it like running boost above the MAP limit.

Unless we find a way to convert to wideband, this is the best option IMO for those looking to go this route.

(regarding wideband ECU)
What is the name of the constant are you talking about?

I would like to try running lambda 1.07-1.12 on low load Closed-loop. Will all targetvalues in all lambda tables be affected.
Does the constant affect WOT?

Read the original post again.

It is a constant in the code, it is simply a value of 1 and not intended to be calibrated.

You should be able to find it by comparing the hex at the location above in the S4 M box to your file. However, if you are off you will be modifying code and that could be bad.

Ok...another thought.

Change the constant in LREB that lamsons is evaluated against to match the constant in LAMKO (described above) and adjust USR accordingly.

This should keep LR active, provide the proper IOT, and not fight trims. The benefit over just using USR is that the target amount of fuel is getting injected, so this way we are not forcing ourselves to trim to the preffered lambda.

I'll have to dive into LR a little deeper when I get the chance.

Set LAMBDIAG, LREBRI, and LREBLE all to desired target lambda. LAMBDIAG is the constant in LAMKO that sets target when there are no other influences dictating lambda. LREBRI and LREBLE are the constants in LREB that enable/disable lean/rich protection.

Lean/rich protection will disable regular lambda regulation (B_lr) if target lambda (lamsons_w(2)) from LAMKO != LREBLI and LREBRI. This function can take action if lamsons_w(2) < LREBRI (rich target lambda) and usvk < USR (lean sensor voltage) or lamsons_w(2) > LREBLE (lean target lambda) and usvk > USR (rich sensor voltage). These conditions can set B_lrnd.

All three constants should be set the same and USR will need to be tweaked to find what your target lambda translates to in O2 sensor voltage. This will preserve lambda regulation (B_lr), but accuracy is lost the further you stray from stoich.

LAMKO constant
LAMBDIAG - target Lambda constant
0x3BFB2
16 bit LoHi
X*0.000244141

LREB constants
LREBRI - Lambda constant for rich protection
0x85DBC
16 bit LoHi
X*0.000244141

LREBLE - Lambda constant for lean protection
0x85DCC
16 bit LoHi
X*0.000244141

USR - controller theshold for lambda control upstream catalyst
0x11ADE
8 bit
X*0.005216-0.2

Not to confuse anyone, the constants aren't assigned labels as they are simply a constant. The names above are just how I assigned them in my xdf.

Again, this is untested.

Would the nb emulator on most aftermarket wb controllers be usable/trustworthy?

This is very interesting info, one day when I get my fueling sorted I will look into this myself.

A thought I had was just to hack the 02 sensor linearization, and shift it say, 0.05? The ECU sees lambda=1 but it is infact 1.05.

But now I come to think that would probably not work on a narrowband sensor.

I have no experience emulating narrowband O2's, but I think some members have had some success.


From what I've seen, LALIUS is not present in narrowband ECU's and lambda sensor linearization is only performed on the secondary O2 sensors.

Regular lambda control just trims fuel to match sensor voltage to USR.

Have you tested this more and how much does this have real effect on mpg readings?

I run it this way full time on both my cars. But truthfully, I noticed more from disabling kat heating and tuning LAMFA. The lean burn can gain you 10% better mpg but only for the time you're in that driving mode (cruise) where you're already burning a lot less that any other mode (barring overrun). Disabling kat heating (SAI) and tuning LAMFA so that ECU doesn't go pig rich every time the gas pedal even thinks you're about to press it more, will gain you more in mpg than lean burn. That's just my experience.

Ditto. And in ME7.1, there is no SAI, and LAMFA is disabled anyway.

Gee, I wonder why?

What is proper way to edit lamfa? Is it modifying axis data so that rich areas are in somewhere near 98-100% request or just put higher values on the last fields (in those which are allready on rich side from factory). And then make fueling depending actual load not just requested?
I have allready disable sai and katheating ,or att least i think so. Can you tell how to this also proper way? Cwkonabg maybe?

Also zeroing out kfmres and leaving kfmresk stock helps if you leave the car in neutral (without pressing clutch) while idling in traffic lights. One thing I have noticed after disabling cat and kfmres  is that it takes some  time for the car to come to operating temperature and the idle fluctuates a bit.
Also I am loosing coolant , can it be related. ???

So, if I understood right regarding the narrowband cars, phila_dot's post #8 is basically the same method using USR that Julex posted about, but modifying those other 3 constants makes it so that the car's not fighting fuel trims.  Correct?  Anybody experimented with that any more since?

Can anybody confirm that the constant to change in a HN ori
(like this: http://nefariousmotorsports.com/forum/index.php?topic=286.msg1659;topicseen#msg1659) 
is at 0x67944  ?

Thanks!

... confirmed: Now my lambda target is always at 1.086.

Problems I've found so far:
- ignores LAMFA and KFLF enrichments (where I tried to set back to 1.00 smoothly) on lean areas. I get target 1.086 all time when it supposed to be 1 originally. I only see richer target at high loads (>100) sometimes, probably when BTS kicks in.
- lambda control is massively negative. sometimes it hangs on the -25% limit and cannot lean to target lambda. Don't know what other fueling tables need to be changed?

Little progress here, I was able to enrich through LAMFA with a small DLAMFAW value, but only for LAMFA < 1 values.

I guess the lamfa_w path is maximized to 1.00 from a different 1.0 constant, hence no correction higher than 0.99 possible here.

It would be nice to set some smooth fading from 1.1 -> 1.0 through the LAMFA (or similar table).

Anyone succeeded with it?

Did you change LAMBDIAG, LREBRI, and LREBLE all to desired target lambda, like phila_dot mentions in post #8?  http://nefariousmotorsports.com/forum/index.php?topic=2515.msg24531#msg24531

Only LAMBDIAG.
I couldn't locate the other two, and not sure they're even exist in a wideband car.
If my understanding is right, those two supposed to limit the LAMBDIAG value, and not related to trims(?). 

For limiting I've found LAMLRAMX only and that was around 1.09 already

I didn't know your car was a wideband car, and I haven't delved into the FR yet regarding those other variables to have a full knowledge, but just going off of what phila_dot said in that post, it seems like those are the only changes necessary to not be fighting the trims on a narrowband car.

Sorry, I didn't emphasize it's a wideband.
Yet to see when LTFT values settle for long term use, but my first impression was that I need to set other constants too. 

I hope others will chime in with wideband experience.

I'd ran lean lambda on my 7.1 a while ago by cahnging USR, if you search forums you will see the thread. Look at LALIUS to give you idea about V to Lambda scaling of O2 sensor.

Caveats:

Narrowband can only accurately measure a very narrow AFRs accurately, we are talking like 0.05 lambda either way.
My engine doesn't like anything over 14.7. It starts bogging down at 15.5 pretty significantly so I saw no point in running higher than stock lambda.

To my best understanding of the german 7.5 FR,  two-point lambda control with USR is narrowband-only, I don't have any practice with that.

My car runs fine around lambda ~1.1, even at 750rpm idle. 
I guess the small stock injectors, larger plug gaps and TSI coils help here :)

I am also struggling to get narrowband ECU run leaner than 1.
At least 1.08-1.1 lambda.
I have read so much but can't get real answer on my question. Is it possible? For economy.
We can gain good economy running around 1,05-1,1.

For kflbts fueling lambda is probably limited by
LATRO - Upper limit lambda target for exhaust gas temperature control
018ch ecu - 0X1D11E (16BIT LOHI)
offset: x*0.000244140625

latro doesnt seem to exist in 518al

Can anyone confirm my finded adress please?
0x65C62?

ive found the address for these ecus so far and they are confirmed working by me. i found them using winols just comparing the 1 address someone else found till i found the most similar patterns i could in the random coding towards the end of the ecu which honestly took a bunch of hours between all the different bins. I dont have time to do it for you but you can use one of these to find it.


offset for lambda - X*0.0002441406

8e0909518ak 368072:  0x72456 
8e0909518al 386802: 0x72F1A
4b0906018ch 360101: 0x6654A
4b0906018dp: 0x5F8B0
b5 s4 m-box: 0x3bfb2

Ok thanks i search for them.