NefMoto

ECU 8N0906018F, 512kb narrowband motronic 7.5.

never played with underscaling before, but i think this car is a candidate for it.

currently at 1.35bar boost, and i noticed im hitting 191 load, and capping out.. from what i can tell this seem to affect the airflow when logging as the MAF signal starts to swing up and down.. (probably due to throttle or timing interventions etc)

what is the smartest way to underscale this so i wont cap out the load?
i wont be boosting over 1.55bar, and preferrably i would want the load to remain the same up to ~160ish or so, that way i wont have to remodel timing etc too much.

ive seen KFKHFM and KFLF mentioned.. but how to do it? other ways?

i remember someone mentioned there was a variable that could be altered to change the whole conversion of the load, in some KISRM discoussion thread..


any advice appreciated =)

Log ps_w.
Also rl_w does not max out at 191, only rl does, so just log rl_w.

Pretty sure there's just something wrong with your tune, as mshfm is not affected by load.

hmm.. ok, so there is NO point in underscaling this?

i could basically just up ldrxn and request load above 191, only i wont see this in VCDS? (no me7logger used in this car..)

You are not hitting problems with maxed ps_w at 1.3 bar that's for sure, so yes you don't need to underscale anything.
VCDS is waste of time. Have fun tuning blind :P

Agree with prj on all counts.

haha, well.. the cluster/gateway refuses communication with the ecu.. and I myself refuse to hook up a direct k-line to it.. so yes, i might be tuning blind atm, not like its the first time.. ;D

first time ive seen the load column maxed out tho, and requested boost dont seem to raise even if requesting higher then 191 load, which is why i asked here about underscaling.. =)
how would i raise the requested boost further?

car is an 1.8t 150hp engine, stock internals, fitted into a VW Lupo, k04 turbo, 1.3bar boost at 27deg timing.. airflow ~215g/sec atm.. this little green thingy is QUICK, the tune is spot on up to 1.3bar boost.. but we want to boost MORE..! =D
theres more flow to extract from this k04..  ::)

As I said before, your problem is not ps_w or load limit.
And you won't find out what it is by going blind.

ok, thanks for clarifying that prj =)

ill get to the bottom with this..

hmm.. i just noticed 215g/s is ~4.36v on the MAF, perhaps the maf have difficulties reading higher? or will it operate all the way up to 5v? any limitations in this that needs to be lifted?

Just about 5v, KFMLDMX and MLMAX notwithstanding.

KFMLDMX is allready corrected.. MLMAX, does it really affect? i know i read a discoussion about this earlier..

http://nefariousmotorsports.com/forum/index.php?topic=769.msg6689#msg6689

why im asking is, its set to 152g/s atm, or 550kg/h, and as mentioned we are allready way above that without any problems at all.

From looking at the FR again, I can't tell if MLMAX affects anything. Good find. Perhaps prj can comment and confirm?

Just diagnostics from what I can see.

hmm.. ive found something.. my KFMIOP load axis ends at 195.. can this cause the boost to not rise any further?

KFMIRL is requesting 204, and LDRXN allows 204.. alltho from ~191 and up i cannot really see any noticable differences in requested boost.

It'll go higher, using the last values in KFMIOP for any load >195

Please. Log. Stop guessing.

ok, thanks  :)


WHAT and HOW should i log when me7logger wont connect?

im very much aware that im tuning blindly here.. and so its a bit of trial'n'error, i can see enough variables thru VCDS to figure out the requested boost wont raise, just need to find what im overlooking.
spent several hours tuning the PID, and the boost is really spot on at all loads.

logging a ton of variables is nice for sure, but not a necessity to solve this.. if rl_w and ps_w are not capped as i thought when posting, then ill just go ahead and raise KFMIRL, stretch KFMIOP axis and further increase LDRXN.. should do the trick.

thanks for all replies, advices and clarifications guys  :)

Cut the k-line from the ecu and obd port and run a new wire directly to the cable.

I've had luck doing this when I couldn't connect.

i have to unplug the 2 cluster fuses and me7logger work , strange but works.
also you need to rise IOP , which prevent  you now from more boost.

yeah, i did this on my own car, works like a charm! ;D

..this car, not so much my own, and while it would be great for the tuning, we are allmost done.

the owner is more then satisfied with the car as it is right now, but we did one more quick run, and the MAF-signal "bouncing" seems to be more related to the boost PID stabilizing at 2320mbar, then a load cap, so no worries then, well add more cylindercharge tomorrow..!  ;D

thanks for the advice, ill look into it!  :)

Raising IOP is never a good idea....

You need to log rlmx_w, rlmax_w, and rlsol_w.

I don't see how people are giving advice without this information.

Why ? in many 1.8T 150/180hp IOP ends with 160 , which stops you on max 1bar boost, so what do you advice in this case ?
i always increase iop values and its fine. without increasing IOP in those files i coudn't boost more :(

BUT-  YES , in standard S3 APX/BAM file i never need to rise IOP more than factory 190 - and always i could boosting to 1.5-1.55bar, normal or not ?

You mean the load axis, or IOP itself?

Taking out both fuses for me didn't work... I only took out the first fuse (so that the clock was still showing, but the needles were dead.  If you pull out both (so that the whole cluster will be blank) it might not work.

I've had issue with nefmoto and vagcom not connecting on S4s with bad ABS modules, when this happens I just pull the ABS fuse and it connects again no problem, not sure if this is related but worth a try..

I mean load axis only


One fuse for me wasn't enough in my case, but taking out 2 fuses work good.

this is my experience too.. the requested load/boost wont raise above the last axis value.


the file is indeed S3 APX(programmed into a 06A906032AC ecu), but the engine is an earlier 150hp with big-port head etc.. so 1.3bar boost on this one might actually represent higher load then 1.5-1.55 bar on a normal s3 engine.

This makes no sense.

The axis itself isn't going to cap anything. It's the torque value that matters.

i never completely understood the KFMIOP map and its relations, but im sure you are correct phila_dot  :)
mind explaining the nature of the KFMIOP map in simple words?

some brainstorming below..  ::)

KFMIOP seem to represents maximum torque% of the engine at a given load, and the characteristics of the engines torque output itself is set in KFMIRL by specifying the cylindercharges relation to torque..?

following this would mean 204 in my MIRL at 100%, will be capped to 88.7% = ~181 cyl charge once requesting 195 or more.
raising the last column will make the interpolation change and allow higher load.. to a certain point..
raising the MIRL values and correcting the IOP for them should prove more efficient..?

http://nefariousmotorsports.com/forum/index.php?topic=3765.msg38179#msg38179

This post ^^^ sounds like you are saying that rlsol won't exceed the highest value in IOP load axis, but your last post sounds fairly on track.

Raising the IOP map values (z, output, w/e you want to call it) may be required in the high rlmax columns. Just make sure that you don't exceed miszul.

Thanks for the clarification! :)

i think it was just a coincidence that it stopped at 191 load.. since my last value was ~195.. so ill give it another go, im sure ill manage to add some more boost to this thing now  :)


to ensure not going above miszul, and to avoid interventions, can i raise the values in KFMIZUFIL?


i think we went off-topic by far with this thred, but this has been really helpfull, and perhaps it helps others thinking they need to underscale things.
im still curious about how to do it, since i will eventually hit that 2550mbar limit, and i know people have been running higher boost, just without the PID-control, but perhaps the 5120 method is the way to go from now on..

You will get Level 2 intervention unless you mirror the changes in the UM maps.

With a little bit of effort, you can balance the maps without having to touch KFMIZUFIL, the UM maps, and subjecting yourself to a runaway car.

I dont get what you mean by runaway car?

I have ran mine every way possible, with every limiter, monitor, nanny map turned off it feels fine, part throttle is actually not distinguishable from normal TQ monitoring files.  Civil when you want, violent if you want.  How many here have actually tried it with everything off?  Curious to their experiences as I have only done it on my single car and I have a huge turbo.  My peak boost is just before 4000 and i can easily modulate everything under and above that.

 ???

I'm not saying that it's bound to happen, but the purpose of the watchdog is to have a second level of protection in case of corruption of some sort to ensure that nothing catastrophic happens.

I don't have everything disabled, but it is tuned without compromise with ZERO conflict in the torque model.

Maybe when its on it just isnt holding me back anyway because i edited everything.

So.. i ended up raising the KFMIRL last column to specify more cylinder filling and raising the KFMIOP last axis value to ~200, plus raising the values in the two last columns by a few % to allow higher torque.

this now requested 1.5bar boost, from previous 1.3.

KFLDIMX had to be tweaked in the last column(1200mbar)for it to reach target pressure, but once this was done 1.5bar boost was realized, tapering off to 1.15 at revlimit.. talk about pushing a k04.. :P

engine runs soo smooth yet aggresive and there is no hesitations whatsoever, no interventions on the timing, throttle or WGdc from what i can tell, and it pulls like an OX..!  ;D
this on E85 using Genesis II 500cc injectors at 4bar rail pressure(these really have insane linearity and atomization and i cannot even begin to explain the improvement they did for the e85 tune and fuel economy, in short, amazing)

You are run on E85 on stock fuel pump ? By simply filling E85 to Tank ? what are the other mods ?
Did you dynoed actual setup ?  my friend is runing k04 1.5bar boost making ca 270-280ps, and thinking to run on e85 or metanol to make more power :)

That isn't what you are doing by raising the last two columns of IOP... if you RAISE the values, you are increasing the actual calculated torque at those operating points, making the ecu think you are making more % torque at that load.

Apples to apples (if you didn't alter the axis), for a given load operating point, you actually want to tell the ecu you have LESS % torque.

i.e. if 150 was 100% torque, and now 210 is 100% torque, and 150 is... something a lot lower than 100% torque.

IOP IS NOT A TORQUE LIMIT!

if you scale the load axis ONLY, and do not alter the torque values look at what you are doing: you are LOWERING the %torque output of the map for a given load operating point.

So if the 150 load cell was 100 percent in IOP before, and your load now is 200, you'd want the the 150 cell of the axis to be 75?

Great question. There are more things here to consider.

The easiest thing to do (assuming your KFMIRL stays stock), is to just re-interpolate IOP based on the new axis scaling.

Not sure where that puts the 150 cell.. it may be near 75.

If you adjusted your KFMIRL, you can just scale the IOP load axis by a %, and not re-interpolate.

But that isn't quite right, unless you added the same % to every req torque point in KFMIRL...

If you adjusted just the TOP load of your KFMIRL, you can just scale the top part of the IOP load axis..

The FR says IOP is the inverse of IRL.... so you could start with that. Take whatever you have in IRL and run it  through masterj's excel template... but if you do that, you quickly notice the stock IOP isn't the strict inverse of the stock IRL *especially in low load regions*. And when I tried this, it did bad things to idle and very low load regions.

So. What to do? I'm not sure :)

My approach was to leave IRL stock pretty much everywhere except where it flattens out at high torque request.. i basically extrapolated it up so it hits where I want.

Then, i upscaled the last 3 axis values of IOP.. and did some mild interpolation, but not much, since all you REALLY need is to scale the IOP load axis!

IMO if you don't change IRL much, you can probably leave IOP alone *including not scaling the load axis*. As others have said, you'll hit the end of the table and calc torque will stay low... which is what you want!

Oh, on more thing i tried:

using masterj's excel template, i calculated the inverse of my new IRL using the new IOP axis (last three load cells scaled).

Then, I just cut and paste ONLY those last 3 rows into IOP.

That seemed to work pretty well as well.

In any case, what I found was the less stuff i did in IOP, and the closer it was to stock, the better.

This is what I do. Leave the rest of the axis alone, just scale the last 3 columns so as not to have too large a spread between them. It's just with all the talk about IOP, I've heard (and experienced) quite a few things that don't jive with my general understanding.

Nyet, thanks for pointing that out, i didnt raise em more then ~1-2% this time tho, but i raised the last columns of the axis ~5-10 points at the same time.. so that was really contra-productive.. if i LOWER the kfmiop or extrapolate only the axis further, im sure better results could be realised.

the tune is really nice now tho, it hits requested boost SPOT ON and follows all the way untill revlimit in all gears, so i dont think i will tinker any more with it. 

again, thanks for all advices here.. this kfmiop map and its relations to mirl etc keeps eluding me.. i know what the fr sais etc.. but i too also find that the closer to stock the more stable results.. so lets stick with it ;D

You guys are going to make me go gray or bald or something.

There is no mystical black magic behind this stuff.

Please read what I have written on this subject. I have poured over the code and logged it all. Don't take this as arrogance, I'm humbly trying to convey that there are a few simple facts that need to be considered.

KFMIOP DOES act as a limit for the torque request via mimax.

Why is this the one area that everyone accepts blind calculations/guesses?

Phila, do you have a log showing what torque intervention via mimax looks like? It would help me visualize what you are saying and help me understand where I went wrong..

mimax_w is the max indicated torque and basically caps mifa_w in MDFAW.

mifa_w then caps mifal_w before milsol_w in MDKOL.

In both paths, ignition angle (mifa_w) and air charge (milsol_w), the torque request is limited to mimax_w.

Torque intervention, as far as KFMIOP is concerned, is caused by misolv, mizsolv, or mibas exceeding miszul.

The post that I linked at the top of the previous page summarizes everything for the most part.

I think the only other consideration has to be miist which goes over canbus for use in calculating intervention by other modules. I don't know the logic applied in the other modules, but the torque request and interventions (including external) are all percentage based.

BTW here is what I am working with now with ME7L:

http://nefariousmotorsports.com/forum/index.php?topic=837.msg39120#msg39120

I would like to point you to my little thread here http://nefariousmotorsports.com/forum/index.php?topic=8617.msg82115#msg82115
I'm hoping to get more attention to find answers of my silly questions.

Im feeling not very confident with this but while diving through all the old threads i learned that all of you asked stupid things back in the days.
Maybe because there were no answers at this time but 99% because we all have to start early and have to learn.

Unless you're running one of the billet compressors from the FSI-TFSI K04's, there isn't much more airflow to be had from a K04 as 215 g/sec is about 270 CHP.

The largest of the cast compressors (2288 IIRC from the Mazdaspeed 3 and 6) is rated to 32 lb/min, but in the stock longitudinal turbine housing, you'll top out at about 28 lb/min or so of airflow.

Even with the largest of the billet K04 compressors, I don't know how much more airflow over 28 lb/min it will sustain. I know I've seen European dynos using calculated CHP that have reported 310-320 CHP on some of the FSI compressor K04 hybrids , I don't know how they arrived at that number. I would suspect that the WHP numbers would tell a different story....

Trust me, I've had the stock K03 / K04-15 / 2288 compressor Mazdaspeed K04 hybrid / GT2860 / 60-1 T3/T4 and a Comp 5566 setps all on my 1.8T. So I'm very very familiar with virtually all turbo combinations on those engines.

You say you're running 27 deg advance on this setup by the way, I'm taking it you are running water/meth injection? If not, I would strongly advise using it. Especially if you are trying to tune the stock ECU directly.

Water/meth is like a get out of jail free card when it comes to tuning mistakes. Trust me on that one lol. Lord knows I've made enough of them hahahaha.

lololol, after doing this for 22 years I still ask stupid questions and do stupid things on a regular basis, so don't feel bad about it.

I say you're only stupid if you don't learn after frying your 4th ECU ;) http://nefariousmotorsports.com/forum/index.php?topic=9030.0title=

The most i have seen from billet K04 wheel is around 270 g/s and they Dyno giving around 330bhp.  It's not the compressor that is the restriction though, it's the K04 turbine housing.

Rick