NefMoto

I am starting to write my first stage 1 file for my 2003 audi a4 1.8t.  I picked up an XDF off of this board, but it does not contain a map for KFMIOP.

Am I able to adjust the KFMIRL map without editing the kfmiop aswell?  Any info would be great (please don't mind this noob question)

Thanks,
Mike

Hi, Mike,
You will have to adjust KFMIOP otherwise you will have torque intervention and that will result in poor drivability (use my me7wizzard to generate kfmiop from kfmirl). Also you can disable torque management, but that will require more maps to be defined, so it would be easier just to find kfmiop ;) cheers

After doing some more research, I realize that the KMFIOP is the inverse of the KFMIRL.  I am trying to figure out how to locate the map with WINOLS, but i am having a hell of a time.  any pointers on how to "discover" it? 

post your original.

This is what i am working with.  I picked up both off of the forums.  i am going to pull a .bin off of my ecu soon, but I am holding off until I finish an exam tomorrow.  I don't want to get too into this before I finish that.  I know if I pull it off,  I wont study.

Also, kompressed... your from AZ right?  Nice to see you over here.

kfmiop: 17292 11x16
factor 0.001526

That was quick.  Mind letting me know how you found that?  That way I dont have to ask again :P

Experience :)

Load up a defined file that is similar to yours in winols, look at the map of interest in 2d mode and then look for a similar pattern in your own bin. Thats the easiest way.

A pattern search in IDA is quicker ;)
The quickest way is of course having already worked on the file before.

Ofcourse, but certainly NOT the easiest way :p

Please explain why, where, and how.

As for being the inverese, everyone interprets that way too literally.

IOP is torque from load
IRL is load from torque

Thats the inverse, the actual map values are not inverse.

I got it working so far.  Just wondering how I link the lables, or if they even matter.  Right now I have all 0's for the axis.

Without changing KFMIOP load axis I couldn't boost more than ~1bar of boost... But this problem might be also connected to other maps (*_UM maps in particular). Anyway, after changing KFMIOP (also all the maps that share same load axis) AND these http://nefariousmotorsports.com/forum/index.php?topic=1518.msg14433#msg14433 (http://nefariousmotorsports.com/forum/index.php?topic=1518.msg14433#msg14433) I had no more problems

Could I use manually entered values for the axis or do I have to find the internal hex lables?

I don't know of any path where the results of KFMIOP can interfere with requested boost. IIRC it can only affect timing intervention.

Am i missing something?

mifa_w will be limited to what is in the last column of KFMIOP.
So if the axis ends early (before your req. load) then it means that you will never reach req. load, since it won't know how to request all of it.
If you simply compensate for this by raising the KFMIOP values, you will get torque intervention.

The UM maps will never interfere unless you mess with the maps from MDZUL.

The only reason you would have for that would be if you did something crazy with KFPED or KFMIOP.

milsol comes from MDFAW -> MDKOL -> MDFUE

This is the torque input for KFMIRL

mimax_w is KFMIOP from rlmax_w (LDRLMX - LDRXN corrected)

mivbeg_w is mrfa_w(mimax_w - mimin_w) + mimin_w + dmllri_w limited by MDIMX

mivbeb_w is min(mivbeg_w, mimax_w)

mivbeb_w then goes through a slew of potential interventions and finally gives mifa_w.

mifa_w then goes through MDKOL and is subject again to numerous interventions finally giving milsol_w for KFMIRL.

KFMIRL will request the air charge you set for a given torque request.

is anyone able to tell me the location for the axis definitions?

Am i able to use the same ones as in KFMIRL?

Normally it isn't same as KFMIRL...

Yes... but I did like my layman's terms description ;)

But you can compensate in KFMIRL.

FYI for everyone modifying the load axis, it is shared with KFZWOP(2) and KFMDS.

So, no one can pinpoint the exact condition that is causing the interventions?

I have gone through the code a million times and logged almost the entire path, but I don't have interventions. I have suspicions, but I want solid logs capturing it. I guess I could sabotage my file. I just don't have time to play...

This map is my pet peeve though. I hate how people just regurgitate crap without any actual information.

Can I label the axis to match what I have seen online, or do I need to find the hex address for the internal axis parameters.

think this is right:
13b44, .25, rpm
173f4, 0.023438, %

What do you mean by that?

If KFMIOP and KFMIRL no longer more-or-less match the ECU will not be able to request a given torque anymore.
If you want to request more load than the last column in KFMIOP you will have to change the axis.

One side effect of getting this wrong, is an ESP system that is completely useless.

Match? It doesn't matter.

KFMIRL can give you whatever desired load you want. If milsol is limited you only lose resolution in the map (map becomes smaller with lower torque request).

How does esp become useless? The torque request from ESP during intervention is extremely low, the map would have to be really screwed up for the driver's torque request to be lower.

I need to take a day and do some logging to settle this for good.

Remember, KFMIOP for mimax_w uses rlmax_w. I think alot of people overlook this.

Not the case on my car when I have smaller scale traction problems in 3rd gear pulls on the road.
There is intervention, but very little of it. If the top end of the map is messed up, I get hard cuts. In fact the intervention is almost not noticeable unless you eyeball the ESP light.

And the reason is simple. If your KFMIOP axis ends at 150 load, but you request 200 load, then anything over 75% will be the same value.
Yes, you can request enough load, by just filling 75% to 100% in KFMIRL with 200, but for example the ESP will not be able to finely control the torque request.

It will see during the first request that it has no effect, and during the second request you will get a hard cut because mivbeg_w will be very low.
Basically the function is supposed to be a pretty literal inverse. And it is that on all the factory cars I've seen, give or take a percent.

Otherwise the torque can not be dialed back and forth correctly. Not a problem if you are regulating it with your pedal, but a huge problem if it is needed for ESP.
The load axis of KFMIOP should represent a good distribution of actual up to rlmax. It should never top out on the last cell during the operation, as it is no longer possible to adjust torque past that point (due to the hacks you will need to do to KFMIRL to make it work).

IIRC in my logs the interventions cut torque down to 40 or lower.

FWIW, I'm not saying not to rescale the axis.

Time permitting I will get some logs this weekend.

K0mpresd, I don't think these are the right values.  They don't seem to be working. 

they worked for me. 16bit.

Please see attached for my results.

(http://i205.photobucket.com/albums/bb240/Rockfordca/KMFIOP.jpg)

Correct for me, too.

16 bit, address step 2, LSB First, significant digits for rows not more than 2, not signed.

dont know. heres mine using those values. attached at bottom.

I'd get you some logs of mine, but the weather won't permit this for a few more months.

Is the RPM supposed to go along the X or Y axis? 

In Tunerpro rpm is row (Y, left) axis. Load is columns (X, top).  If you post pictures of the XDF row and column set-up the problem will probably pop right out at us.

Thanks for doing this, by the way.  I expect it will help when I start work on my daughter's '04 Passat 4Motion 1.8T.

ive got a few maps sorted in an 04 passat file. 018dq i think it is. enough for a basic tune anyways.

DQ it is.  Cool.

nevermind. the one i have is da. its all very similar either way.

Thanks for checking all the same.  I'm sure you're right.

got it working.... thanks guys!