NefMoto

Hi Everyone,
I don't know about everyone else, but every time I've needed to fix FKKVS, I've never been able to find a good solution that works easily every time. I started a project last night that I hope will fix that issue.
Please download and test out the software. Feedback and suggestions are greatly appreciated.

https://bitbucket.org/guitar24t/fkkvs-fixer/overview

Thank you,
Rob

Here's a demo screenshot from the program using an all 1s FKKVS input and a sample log.

(http://hiltontuning.com/FKKVSDemo.png)

Panda Antivir reports a virus is in your file and deletes it :(

I can guarantee that my file does not contain a virus.
Here is a scan report from virustotal.com
Every antivirus including Panda reports that it is clean.
I recommend scanning your computer with MalwareBytes AntiMalware
https://www.malwarebytes.org/

the program shuts down after you click "process fkkvs" if the log file doesn't have the right variables logged.

I'll add some error handling to inform you the log doesn't contain the correct info.

On a more problematic note, I've identified an algorithm error for computing the new FKKVS map I didn't think of at 4am when I was coding this. :) I'll correct it in the next commit

Do you have plans for a KFKHFM fixer?

I ask because you generally correct that map first, then make corrections to FKKVS in areas where KFKHFM is still 1. That's what I gather from the FR anyway..

I didn't have plans for a KFKHFM fixer but I suppose I can add it in if someone wants to detail their procedure for doing it or can refer me to a link that describes how. It was my understanding that corrections could be applied to FKKVS to compensate for both maps (which may not be the Bosch proper way to do things, but it's still effective).

Next commit has been posted. The algorithm should be corrected and the application now informs you if the log files do not contain the correct information.
Here's another screenshot with an input of a non-1s FKKVS map.
(http://hiltontuning.com/FKKVSDemo2.png)

Tested this fixer today, only one time.

The idea it's to run only once or should I do one new run ??

you can do it as many time as you'd like.  The program applies your changes to a base FKKVS that you supply.

I usually do several iterations of 30min driving sessions of all different kinds of driving. How many times you "need" to use the program really depends on the fuel system and the injectors. I also usually process each file with 3 or 4 smoothing passes and roughly 75% smoothing factor. I find that that seems to give decent results.

guitar24t  - would you plz check my settings as i i am not able to get this working on my file?

Hi Mocke,

Your FKKVS is transposed and was missing the corner zero. Pulse width column has to be vertical on the left, and RPM has to be horizontal across the top. I have uploaded the correct format FKKVS file.

Today I've done two more runs... it's almost perferct. The problematic rpm range is when turbo spools... it gets leaner and it can't correct.... but the oe pump sould be the problem...

lamfa?

Thank you for taking your time.
Am i logging wright values, the rest of files are OK?

Here's sunday tune...

MAF scaling correct?  I've seen similar results when I've used incorrect MAF scaling.  In my case, I thought I had 225tt MAF sensor from a WB, and it turned out I had one from a narrowband.  Two different sensors, slightly different MAF scalings, resulting in a bumpy AFR.

It's a narrowband to wideband conversion and the maf it's the nb... the file has the nb maf calibration...

The MAF readings with the wideband ecu and and wb maf data were very impressive.... but not realistic

But that flutuation was allways there....

Is this the correct way to log: set NOLRA to 7 and log nmot_w, te_w, fr_w ?

I was under impresson that we should log


tevfakge_w       ;{}                                 ; {Adressierung Kennfeld FKKVS mit effektiver Einspritzzeit vor Feinabgleich}

te_w             ;{EffInjectionTime}                 ; {effektive Einspritzzeit (word)}

nmot_w           ;{EngineSpeed}                      ; {Motordrehzahl}

NOLRA 7 or 6

I used NOLRA 6 and logged nmot, fr_w and te_w

NOLRA=7

What's the difference between nolra 6 ou 7???

@Mocke,
You need to log tevfakge_w, nmot_w, and fr_w. If tevfakge_w is not available, use tevfa_w. The log I looked at had the correct variables in it.

@pedrosousa
NOLRA=7 is required, otherwise the idle trims will be compromised since the program will be adjusting the idle trims with LTFT idle enabled.

You must log tevfa_w or tevfakge_w. They are more accurate than te_w for this purpose.

it's all used for bit instructions iirc 00x6 = 0b110 0x7 = 111 (= 1 = true)

I have tried to use this exe today, but it totally messes up the FKKVS map.
Now, I did 2 log runs logging the right variables but the outcome remains the same. Even if I use the sample csv's, the result looks like my own below. What's wrong? I used raw csv, parsed csv, different variables etc. Also tested on 2 laptops (first x64 win10 and later on x86 winxp)

Question: are you trying to fix fuel system non-linearity, or problems with MAF calibration?

The maf and mlhfm are stock (not for my engine, but corresponding) but on a large custom turbo inlet, and the fuel system is a custom return system on 550cc. I am trying to finetune the fueling department. I know you are sceptical about this type of tuning, and it is possible that my kfkhfm should be adressed too but without the exact stock (corresponding to the B5.5 A4) fueling or a flowbench tester its not really easy to realize. I did my best to get the maf reading stable, its on a large K&N with 40cm of straight 80mm tube between the maf and the filter:
(http://i284.photobucket.com/albums/ll32/tijnvanesch/IMG-20150610-WA0000_zps0wwrcup2.jpeg) (http://s284.photobucket.com/user/tijnvanesch/media/IMG-20150610-WA0000_zps0wwrcup2.jpeg.html)

If the result is like that even for sample CSVs - something is wrong with your PC. Most likely - locale settings. Either just set it to "US" or verify that the following are chosen:
Decimal symbol: .
Digit grouping symbol: <blank>
List separation: ,

EDIT: After looking at your screenshot closer, I'm 100% sure that it's your locale. Check how it parses 0.85 as 85 and ignores decimals in next lines as well.

Yes! Didnt know that location settings could affect the software. I'm located in the Netherlands.
Thanks a lot, looks much better like this  ;D Now time to parse my own fkkvs map (its not all 1's in my software)

Very nice!

Glad to see people are still using this thing. Does anyone have any feedback about feature requests or improvements/changes?

Well, if it is possible to make the program location setting independant (comma or dot seperation) this would be more accessible for foreign users regardless of computer settings. I got this program working after a tip, but its not very practical to switch this location settings from European to American any time you want to use the program. The program itself works very well, I got my fuel trims back to ~2% ;D

I just published an update to the utility that force en-US localization in the software. If you want to return your localization settings back to normal (with the , instead of the .) and try the software, I'll be interested to know if everything works.

Thanks!

I just tested it with the sample files and it works great

thanks  ;)

Guitar24t Hello! your program really works!
Logs actual factor correct fuel became almost follow the request, but there are still a few problems that I would like to fix it!
The first and perhaps the most serious is the problem with copying and pasting parameters in the program window Tuner Pro  map FKKVS.
When you try to not put anything happening ,but if you copy columns, the bottom margin for some reason is not copied into the map

I just can't seem to get this to work properly?

I'm logging tevfakge_w, nmot_w, and fr_w with NOLRA=7 and FKKVS all 1 on a 30 min drive(3 times now) and it generates a complete correction map.

But when I write the corrections to the car it's just all over the place with the fuel. STFT range from -25 to +25 in some cases, idle is a mess too. Although LTFT are -1.5 to -2.5 which is pretty decent. And when I log a WOT pull it's hunting hard and fr_w is getting a good workout...

Have I missed an important step? Or are my injector just complete garbage?

P.s. I'm dialing in some 386cc 210HP injectors.

post logs, and screenshot from the corrected smoothed table

here you go :)

Have you started with a clean all 1's fkkvs table and the file for this provided by guitar?

this is my outcome with 4 passes

Yeah, all 1's :)

My ms and rpm axis is different, could this be my problem?
Does FKKVS share axis with other maps that could skew them?

Change the axis in the "input all 1's csv file" and the output should reflect your axis.

yeah already did that :)

but does the map in the .bin file share axis with other maps?

i've just reverted to stock axis in .bin file and made a new fix map from logs, going to try and see if that changes anything.

Are you running MAFless by any chance?

So that made no difference changing the map axis back to stock on the flash, same rollercoaster trims :-\


No, stock MAF :)

Is TVUB correct for these injectors?
Pressure-tested the intake?

I mean, something else is definitely going on if your STFTs are all over the place & idle is effed up.

EDIT: also, regarding your question about FKKVS axis - didn't check your specific binary, but usually FKKVS has its own axis. You can check yourself - if both axis precede the map itself, they're almost always exclusive to this specific map.

Pressure pipes tested to 2.5 bar :)

I have thought of TVUB, but it should be right since its taken from a ecu file that would have been in the car, but could be I got it wrong?

Injectors are Bosch 0 280 155 892 EV12 06A906031J  386cc(393cc stated by Bosch)
The car is a European Audi TT 8N AMU 8N0906018K

Too bad it does not work for med9 :(

so since everything looks to be correct(except for FKKVS), i think that the only conclusion can be that the injectors are inconsistent aka junk  :-\

time to go ID725 or Bosch EV14 SxT 630cc :P

If FKKVS is not all 1s to start with stock, you should start with stock FKKVS. FKKVS is meant to correct for returnless fuel systems. If you have a returnless fuel system, and you set FKKVS all to 1s, it's going to be completely messed up. Starting with the stock calibrations should make things much better.

Is FKKVS per car or per injectors?

I tried with stock FKKVS  from the Audi TT file and it was no good.
Haven't tried with the stock FKKVS from my own car since I thought that wouldn't work well.
So stared "fresh" thinking it would be best? :)

You should start with FKKVS from the car, if it is a returnless fuel system

Stock return system, "same as the Audi :)

I just tried copying the FKKVS from the Audi, axis and all.
It's more reasonable, idle is good and LTFT is -3.9% but still hunting a bit but not as much :)

Not ideal, but sort of acceptable for the time being.

is there anything else that can interfere with the calibration? :)
wall-wetting, overrun, knock enrichment?

you have to disable the fuel trims to properly calibrate FKKVS, have you considered that?

Thank you very much for this tool, made a huge step change in driveability for my car. 1.8t with large MAF.

After 3 iterations it´s really good, will continue to use it, but cannot be much better than what it´s today.

Many thanks!

So on the surface it seem pretty solid after 3 consecutive 30 min runs with a new FKKVS calculated from the previous.

LTFT is 0% on normal driving, and with a heavy foot it's -0.8% :)

But I still see it hunting a bit in some situations.

Does the software take the whole log and average out the data points, Or just a set number of data points?

Right, NOLRA=7, I didn't notice it before. I am in the same boat now working on FKKVS.
In addition I log also B_lr, then edit the log for the rows where B_lr=1 and use only the rest of the data points. It is detailed in the other thread for FKKVS fixer.
Need also high load/ high rpm logs to adjust the high load component.
Lastly the fluctuations that still show, I understood are due to KFBAKL/KFVAKL and need refinement also.

I've seen hunting with wrong TVUB and/or changed and incorrectly calibrated MAF.

Might try with that B_lr thing and see if I can eliminate wall wetting and disable EVAP.

TVUB should be correct for the injectors(taken from ECU from car they came from) and MAF is stock

Hello, I would like to use this tool, I have prepared the file for me7logger, and located nolra in my .bin
Of origin is: 00000100 = 04

How do I change it to disable the correction? Should I use 07? Or 05 as commented in another post? Thank you

is a me7.5 032rp ibiza cupra

NOLRA=7 so idle adaption doesn't interfere.

0000 0111 = 07 is correct?. thanks

Yup.

Nolra 7 works for me on a BAM CB file. The tool itselfs is amazing. After one run my trims are already close to 0.

Funny thing. I just drove through a tank of gas with fuel system/injector cleaner mixed in.
After a short while my LTFT went to -3.9 and -4.7 when I poke it.
My first thought was that it was the cleaner that messed with the trims.

But now after 150 km on fresh gas, the trims doesn't go back to what they were before.(0&-0.8 )

So maybe the cleaner actually worked and my injector(s) were clogged/dirty, and that was why I got so inconsistent FKKVS fix maps and lambda hunting.

Or maybe it's just some other expensive thing gone wrong  ;)

hi i want use this on my 1.8t 4B0906018CG    but on stock file i have norla 36 it its possible? change this for 7 right?

possibly wrong defined in your winols

can have a look later

What specific cleaner did you use that actually yelded some results? My limited experience has been its all snake oil.

That has been my experience all this time too.
But got it in a service kit so I thought what the the hell, why not try.
It might have been a placebo effect, but it did something since my trims permanently changed :)

It's called belladd
http://www.belladd.dk/da/produkter/bell-add/servicerens/servicerens-1b-1-12-detail.html

With WBAFR checkbox do you need to include WBAFR in the log? Sorry if it's a stupid question just wanted to make sure. With or without WBAFR column included I keep getting an error that parts are missing in the log file. Any help would be greatly appreciated :)

D

Are you using me7logger for logging or vcds?

Great share

Then make a log configuration that logs only the variables mentioned in the quote I posted before :)

:)

Do it a bank at a time, but I'm not sure the software can figure it out when there's bank designation in the data (1&2)
Also I seem to remember that all 2.7TT was narrowband?

Hi,

I need help with something that's happening.

I use NOLRA=7
I do the logs with me7 and the correct variables
I'm using the latest version of the fixer

Last year I used this program and worked great, recently I'm getting this kind of output and can't understand what I'm doing wrong.

The process is :
I copy my current fkkvs, charge the log, and must give me the recalculated fkkvs is not it?

thanks

Should I enter the new fkkvs in each pass? or charge the same fkkvs and I'm going through the different logs?

Enter your current FKKVS, log, run fixer. Copy output FKKVS to your tune, flash. Enter the NEW FKKVS that you just copied into your tune and fresh log into fixer, generate even more better new FKKVS.... repeat until the changes become minimal.

So yes, you'll input the newest version of you FKKVS every time, not the base file with all 1's

hello again, however much I try to improve using fkkvs fixer, I always get a worse result, exactly how the factor of that map works, above the higher value I enrich the mix? I have tried increasing the map 0.3% the map to observe the reaction, but I have only managed to make it worse and make it more unstable. I want to make more stable to 1 in idle and cruise. could you help me?

are you sure tvub are krkte are right? lambda control within 10%?

yes, krkte,tuvb is from s3 file

is "099 -> switch basic settings ON -> LR off" in vcds the same as setting NOLRA=7?

no
nolra switch off adaptation 32 channel
lambda control not

found NOLRA scalar in xdf, is there no easy way to change its value without reuploading a new .bin every time I want to log for this FKKVS fixer? like in vcds or something?

Who gave you the idea this would be easy?
This is what I've done, I have had some days where I would write to the ECU 30 times.
Even the LDRPID feed forward takes a minimum of 10 writes/logs.
FKKVS fixer could be done in 2 writes and 1 log.

But I remember some mention of using the unused channel in investing to change a scalar value.
But can't remember who, where or if it worked?

I wrote a switched NOLRA function once, embedded in a 3 set mapswitch file, but it is really a lot of work that can be shortcut by just flashing the ecu. I mean, it takes 2 minutes? Other than that, NOLRA is a value in flash and without custom coding not possible to alter on the fly. The closest you will get is with VCDS basic settings for LR, but you will have to keep the pc with VCDS connected.

Hi all
I want to fix my fkkvs because i have ltft between -12 -> -3 and stft between -8 -> +6
I have a 1.8t auq with big intercooler and 386cc injectors from audi tt bam engine (software is modified for them with all audi maps and values).
I have some question because i don't understand well , so after make nolra = 7
1.must make all fkkvs value to 1 or not before make logs?
2.how much must i drive (in minutes) for log will be ok?
3.can make the log (one log) in all gears or i must make log for every gear and idle too
4.must run at constant speed al a normal drive in a city with stop and go?

Many thanks

1. Not necessarily. You could also start with your stock/modified FKKVS, just remember to input you current FKKVS into the tool

2. The longer the better. But 30 min. Seems to work good. (For me at least.)

3. Just start the log when engine is warm and just cruise around town.

4. Just try and get as many different pulsewidths in as many different engine speeds as you can.

hello, first you must be sure that your krkte and tuvb is the right one, the black injectors do not carry the same values ​​as the blue ones. then I as the best result I have obtained is starting from the whole fkkvs to 1. I made a log of about 100km trying to get all kinds of situations, and keep in each rpm range as long as possible. You can help with the cruise control. use at least 20 samples per second. after this log it loads the information in the tool, and the map should come quite close to what is acceptable. if you want to tune more, you can make shorter log but trying to use all the ranges and in several throttle positions. it's about looking at it and once you get that the tool does not correct just, you should have the adaptations close to 0. I use 50% smoothing and 3 passes. I have always used this configuration.

Many thanks for yout answers.

i kep getting error " index was outside of the bounds of array"

i know its due to my fkkvs file just dunno whats wrong
help? thanks

Your rpm axis is not good. You probably hit enter  ;D
try this one.

Hello!

I get this msg error...


Those are the data..  Something is wrong...  ???

I believe your input CSV or log file is not in the correct format. Please check the sample files.

I have been playing around with this a bit and was wondering what exactly the WBAFR correction was?  Also what variable do I need to log in order to utilize it if I need to.  I have a 1.8T with Wideband sensor...

This is used for correction of Narrowband cars that do not have a built-in wideband sensor, and one was added to the logs.

Cripes Rob is back!!

Hi Thanks,

This is an excellent tool but... it's weird as I downloaded this and it complained that the temp file created if clicking on "before" is neutralised as a virus see Panda attachment?

Hi
I can’t figure out how to make a CSV file from my FKKVS card
I work in WINOLS

it works

this does not work

what could be wrong

Download Notepad++ and compare both files
2nd file does have some empty lines at the end and in the first line

thank
already figured out
it's a cool tool
I'm just in shock
!!!!!!!

Hi,

I have been wondering how to use WBAFR correction?
I have 2.7t with stock narrowband setup in operation. I have also wideband sensors connected to rear-lambda inputs in ecu so I can get wideband AFR/lambda information to logs.
What column name should I use for wideband and in what format to present wideband information to fixer?

Hi all and Happy new year!
Somebody can tell me how works that field in fixer "number of smoothing passes" and "smotthing factor". Witch are the best values for them ro run fixer?
Thanks.

how do you make input .csv from existing fkkvs map? possible with tunerpro? what do you use? i don't have access to winols if its needed.


Edit: ctrl+c in tuner pro for fkks map and ctrl+v to notepad and replace all " " with ,  :D and manually write x and y axis topics

Anyone done this on a master/slave arrangement?

Figured its simple, log each bank ecu and write bew fkkvs to the associated bank?

This tool took me longer to get working than I'd like to admit  :) - Only due to getting the right file formats.

Brilliant tool thanks for sharing.