Hi Guys,
i'll have to toy around with an M3.82 during the next weeks and so i wanted to get some things clearer.
First thing i couldnt get the best info about was KHFM - Grundanpassungskonstante.
basically it seems to be quite identical to the krkte in me7.
the only formula i found for it was:

KHFMnew= Qstatnew*KHFMorig/Qstatorig  . Being someone who likes if not everyhtings based on a rule of three i started digging into it 3 hours ago. So i found a patent whith a rough description of how it works:
lets call it:
air mass (better said arithmetic mean of it) gets divided by rpm (identical to me7):

        mL          kg*min
mLn=-------------
        nmot       h

after that follows the KHFM division. The result has to be in ms being the injection time needed for lambda=1 based on that mL:

mLn
----- = x[ms]
KHFM
                                                                     kg*min
from that follows that KHFM has to have the unit: -----------
                                                                     h *ms

based on that i would say the correct way to calculate KHFM is:

         nCyl*14,7*Qstat[g/min]*0,0000167[ms/min]*60[min/h]
KHFM=---------------------------------------------------------
           1000 [g/kg]

with nCyl=number of cylinders
0,0000167, 60 and 1000 bein there for changing units of the input. Also remember from ME7 FRM that Qstatbenzin= 1,05*QstatHeptan.
so if i feed that calculation now with my mLn the resulting unit of the equation is ms.
I made a test equation:


1. The original AGU injectors (06A906031; 0280150464) are called to have 249,1cc/min and 179,2 g/min according to Stan Weiss: http://users.erols.com/srweiss/tableifc.htm
with Qstatbenzin= 1,05*179,2 g/min = 188,16 g/min.

I made a guess with nmot=5000 and mL=140g/s= mL=504kg/h --> mLn=0.1008 [kg*min/h]

  mLn                      0,1008*1000
-----------=--------------------------------= 9,09 ms      (time for one rotation @ 5000 rpm= 12ms)
  KHFM            4*14.7*60*188.16*0.0000167


with 440cc Bosch Injectors the calculation delivers 5.065ms


I think the resulting values appear to make sense.
Based on that i checked the decimal value in the 06A906018CG File which is: 43860. dividing KHFM for the original Injectors (0,0111085936) by 43860 i got my factor for that value which is: 0.000000253. Does it make sense?
Also another question i have in my mind is: is this the value which is also used as the load-values?

so my initial question should be: Is anyone able to tell me whether this is correct or proof me wrong?

Hi MIL_On take a look......

ok...i dont get what you want to tell me with the excel, because if you ask me that just shows that my guessing was correct?
the measured load vs. calculated load after my method:
yellow measured, orange calc (mLn/KHFM=Load [ms])

then divide your injection time by 2 (load is built by division mL/nmot and not combustion cycle which takes 720°!) and compare it to my calculated value mLn/KHFM. of course it doesnt fit perfect as for example the values i got from the internet (Qstat for the injectors for example) are different or bosch has changed the value slightly (red=inj. period/2, blue= mLn/KHFM [ms])

and then compare youre so called K against my KHFM... it all fits more or less? and that could or even should differ a little bit based on youre so called C, D, E


a logging file with a real driving condition would be interesting also because idling is always some kind of special situation...

The idea of the excel file is to show you my results from my calculations.


Anyway as you said comparing K & KHFM for your calculation, mine and logs, both are very close. That was my point with excel file. Also some good logs will awesome to test the calculations.

Also C,D,E factors are that makes final injection period real. From Ti = (Tl * [C,D,E...]) + Tv, Tv will be TVUB or injectors battery voltage compensation. So C,D,E will temperature, altitude, etc..... The main idea is isolate KHFM as much as possible from load from logs taken to validate it formula.

Different versions of a MAP are used for different conditions, like in an automatic transmission, VVT, failure conditions, etc..... A good example of VVT with M383 is the Barzilian VW Gol 1.0 Turbo --> http://www.lunaticfringe.org/vwfox/specs/VW-Gol-1000-16v-turbo.html


During my work with a M383 Damos sometimes I change the value in some constants (8 <--> 16 bits) looking for better interpretation. For example KHFM (84 @ 8bits / 43860 @ 16bits), FGAT0 (1.0001 @ 8bits / 13.0008 @ 16bits), NMAXF (60 @ 8bits / 7100 @ 16bits)......
About KHFM at the 18CG-File ( @16 bit  it starts at 6A68 and containing the value 54AB Hex) I agree with you.

Thanks for your comment PRJ. As I understand from your comment, experience play a huge role in how to deal with old Motronic ECUs in a real world. But without that experience or detailed information, manage these ECUs gets hard, so only way to investigate about it, make some assumptions and with trial and error obtain some ideas/formulas about how to tune it as smart as possible.

In the case of this topic, in my humble opinion, I agree with MIL_On that KHFM could be something similar to KRKTE. Then our idea is try to define and modify KHFM in the same way that was done from the factory, letting us tweaking related ECU maps in a less empirical way.

Elio...As I said to you in MP...

Don´t try undertand M3.8 going down from ME7... it make no sense...

FGATO => fuel multiplier

Try looking at chanell 099 in VCDS
    if you raise it => you get STFT going negative
    if you low down it => you get STFT going positive

KHFM => is used to convert g/s in to load

It´s kind of logic that´s the most simple way to equation injector time to load in ms 

When you hit 12.75 load you are going to have to rescale load anyway.
Otherwise it will just cap out and nothing will work.

Also make sure you never ever hit the MAF limit or really funny things will happen with N75 duty cycle as actual can not reach setpoint. And when I say funny, I mean molten engine stuff.

Basically to tune it for high outputs, scale down KHFM by a factor, so you never hit 12.75 load.
Then scale FGAT0 up by the same factor, so that they cancel each other out.
Then re-calibrate all the timing and fueling maps, as well as cold start maps and you're done.