1.8T M3.8.3 tuning

[Stuart B]

No I dont know him.,, but I know that his few points his made helped me alot.,, and made me understand my ecu alot better

Sent from my HUAWEI CRR-L09 using Tapatalk

The only reason I asked - is you only joined ASN yesterday a day after Thomas's thread, although I "lurked" on these forums for a while awaiting some inspiration too.

I certainly am no expert in the ECUs but my little understanding is ME7.5 controls the components to achieve the "numbers requested" and M3.8 reacts to the values from components to try and work out what the numbers are - making any meaningful hobby modifications an "expensive" <-- not in money but time and fruitless (dangerous) experimentation.

When I had my 2003 Kia Sedona I bought essentially a variable resistor which turned the diesel pressure up and that made a V6 3.0 CRDi 15% more powerful than before? but the other day I told my ME7.5 I wanted a maximum of "210" (whatever's) @ 5500RPM @ 100% load which equals about 21PSI and said I want the car to try and get 0.83 Lamda - and that is after a couple of hours of reading? 3 additional settings later I have set a 2 step launch control, I was told don't touch a PID as I will not know what to set properly - this is why people are not fans of "difficult" / "black art" tuning.

 

[mogsalie0010]

So we on the same page.,,im too very green when it comes to tuning ecu,, eventhou I would love to know how my ecu works and understand how to tune my ecu(m3.8) and other ecu if needs be,, I dont mind the hours struggling,, but getting it right at the end of the day,,that feeling of accomplishment is priceless,,

Sent from my HUAWEI CRR-L09 using Tapatalk

 

[mogsalie0010]

I bought a mpps v16 software and cable to try and read and write and familiarize myself with my car ecu but was unable to read I then heard that the kwp 2000 plus tool will work om my car and I wasted no time and went to purchase it,, but yet I wasnt able to even read my ecu.,,,lol

Sent from my HUAWEI CRR-L09 using Tapatalk

 

[Broken Byzan]

m383 cannot be read over obd, only flashed.

 

[Stuart B]

don't you need to "fake the k-line" using the extra hard wire - Prawn raised earlier? even the AMK S3 Me7.5 has problems with using ME7Logger because of the connectivity.

"cars are s h i t e"

 

[antwan64og]

The eprom can only be read if you desolder it and read it with something like monoscan, but you don’t need to do that.

Other people before you have done it and the files are available. Some ecu flashes end 018R, 018AQ etc. Just so add to the confusion for each ecu although the numbers match there appear to be variations and not all match up on a definition file even if it was made for the same ecu number.

When it gets warmer I hope to be able to test more things on the car and release a basic tuning guide for the ecu on nefarious, with a stock bin and fairly comprehensive xdf explaining how the maps interact.

The main holdup is the two key constants KFHM and FGATO are not fully understood and they cannot be worked out mathematically. What people do effectively is trial and error; put in new injectors and reduce FGATO until it idles ok, adjusting TEMIN (min injection), TLST (start fuel ms), TVUB (injector voltage correction) then massage the KHFM figure until it keeps max load below 12.75ms (theoretical max load threshold). Which is a BS nightmare to tune before you even get to the actual tuning, the correction tables and cold/hot start.

Yes you can have baseline numbers for 310cc Yellows, 415cc Genesis, 440cc Bosch but that only gets you in the ballpark. The ecu depends on the hardware and everyone has a different configuration of parts once they start modifying so will always need tweaking.

Ideally you’d need a 29F200/28F200 emulator and byte swap board for real time tuning but at the cheapest end that’s a $500 investment.

Rtech used to take a week to custom tune the cars and he worked with the ecu a lot, had a dyno and an emulator.

Sounds like a fun ecu to tune doesn’t it Bill....

 

[Stuart B]

it sounds like a £2000 invoice to map a £700 car?

how much is a K6 emerald? -- or whatever Prawn uses!

 

[antwan64og]

Its £25 if you source a cable checksum plugin.
Then a £130 for a wideband.
Then hours of your life hoping to find the holy grail that Bosch motronic logic will one day make sense, lol.
For example Max load threshold is 12.75ms in the LDR boost maps, but max injection shown in VCDS is 16.32ms (presumably due to sampling rate), you have to calculate duty cycle above that point.
WTF!

You can get it running ok(ish) quite quickly, to get it perfect is what takes the time.

After a quick look it looks like Max air flow you can put in the MAF table is 7991.89 Kg/hr of air flow (according to the graph, I didn’t check what 255 equates to in the hex editor) which equates to approximately 2750 hp but the resolution would be awful, lol.

Even If a smart guy put in the time to make a nice M3.8.3 tune for specific hardware like the K380 kit, 3.5” downpipe to 3” exhaust with Wellycooler, N/A cams, etc I doubt they would freely share their knowledge afterwards as they lost hours of their life doing so and it would need tweaking as soon as someone went overbore due to scores in cylinders.

So this is why no one ever really bothered with the ecu

 

[SHUk]

All your input is greatly received...

Thank you!

 

[mogsalie0010]

Isnt that supose to reason enough for some1to actually get to know that ecu,,I would actually love to get to know that ecu and what makes it work,,,

Sent from my HUAWEI CRR-L09 using Tapatalk

 

[jimmcgee]

I love posts like this , excellent info all round

 

[badger5]

With Thomas knowledge he would have more traction and positive feedback from
http://nefariousmotorsports.com/forum/index.php that's a tuning forum, full of hobby tuners and experts?

Do you know Thomas? People have complianed about the ECU not the posters interest in it.

full of primdonna ********* more like..
all you need to know there is read the fcuking function diagram. Thats all they will tell you 99% of the time. A "few" helpful soles in there but dayum otherwise, its a vipers nest.. each wanting to "out smart" the other in knowledge etc... Sadly..

 

[mogsalie0010]

What would you guys recommend,, diktator 60-2 stand alone ecu or having ecu(agu) remapped.

Sent from my HUAWEI CRR-L09 using Tapatalk

 

[badger5]

MLHFM is just the fairly linear MAF table with volts against kg/hr of Air.

M3.8 isn’t really comparable in anyway to ME7.
Since M3.8 is a physically based system, (that’s how Bosch describe it) how it works is wholly dependent on the sensor readings it gets.
ME7 being torque based has theoretically calculated what everything should be from the throttle angle/rpm/altitude/injector size etc so it knows as soon as anything is wrong.

M3.8 doesn’t seem to precalculate much beyond being hardcoded to achieve lambda 1 when criteria is met. So if the hardware goes wrong/leaks/sensor starts fails you have a tough time tracking it down with M3.8

what this non linear "linear thing"

 

[badger5]

KHFM - MAF/Injector constant

KFMLDMX - MAF limit im Kg/HR

FGATO - injector constant

TLST - start fuel ms

TEMIN - min injection time

TVUB - injector voltage correction

MLHFM - rescale for new kg/h 0-5v

KFTLWS - base fuel from TPS, pre injection, in ms for 1 Lambda

KFLFLAV - map for setting 1 lambda

KFLF - adjust fuel 0-1.99

KFLDS - Boost control ; ms

KFLDTV - boost/load start to prevent overshoot

KFLUL - load/boost limit ; measured in ms - 12.75ms max

LDSMXN - Load max ; measured in ms

KFZW - Base Ignition timing


NMAXDV - soft cut limiter

NMAX - Hard cut limit


Fuelling calculated by (KFTLWS x KFLFLAV) x KFLF = fuel injected ms


Iginition and fuelling corrections also for IAT, coolant temp, barometric

MLHFM in M3.8?

 

[badger5]

MLHFM is just the fairly linear MAF table with volts against kg/hr of Air.

M3.8 isn’t really comparable in anyway to ME7.
Since M3.8 is a physically based system, (that’s how Bosch describe it) how it works is wholly dependent on the sensor readings it gets.
ME7 being torque based has theoretically calculated what everything should be from the throttle angle/rpm/altitude/injector size etc so it knows as soon as anything is wrong.

M3.8 doesn’t seem to precalculate much beyond being hardcoded to achieve lambda 1 when criteria is met. So if the hardware goes wrong/leaks/sensor starts fails you have a tough time tracking it down with M3.8

I know.. lol
but you posted up a list of map names... MLHFM one of them, in the context of M3.8?

 

[badger5]

Sounds like a fun ecu to tune doesn’t it Bill....

yea but no.. lol
but I do have an ols300 emulator and an m3.8 ecu with headers soldered onto it.. but never got round (/could be bothered) to play mroe with them, when I can better serve the wide range of turbo/engine/ builds by stand alone ecus and their respective pnp adaotors.
They interest me but only up to the point of way too busy to actually "get into them" as it were.

 

[Thomas_20VT]

I do know NefMoto forum but they are ME7 or newer addicts and I don't like the atmosphere that much to be honest... even if there are people who do really know what they are talking about !

I will prepare an update for boost regulation and monitoring

BOOST regulation and monitoring

Boost is controlled using MAF signal and N75 valve. It is a PID-regulator using a pre-control map KFLDTV. This one has 2 axis which are throttle angle and engine speed. The values given are % of PWM duty cycle that controls the N75 valve

I will not into too much details as it might not be so relevant for tuning unless going mental. For the regulation, there are additional curves for Gain (P-part), Integrator (I-part) and Derivative (D-part) with windows for fast or slow regulation and so on...

For people not willing to spend a lot of time on doing the things right and like the italian tuning, you could use only waste-gate duty cycle to control boost by setting all the PID parameters to '0'

For monitoring purposes and avoiding that you either kill your engine or break your turbocharger in case for example the pipe on the waste-gate would have fallen down and give maximum boost with no opening of the waste-gate. There is a map (KFMLDMX) defining the max load that can be expected without a defect in your turbocharger regulation system. When MAF sees values greater that what is in this MAP, it will go into limp-mode.
Axis are throttle angle and engine speed with values filled in the map being airflow in kg/hr => at 6000rpm, 704kg/hr is roughly 1,0-1,1bar

Badger5, there IS a MLHFM curve in M3.8, same as in ME7

 

[antwan64og]

Difference is you probably have the official super mappack ols or kp file for the ecu so if they call it something different in there I don’t know.

I just have what was gleamed/cross referenced from the M3.8.2 Ols and the various xdf and ols from diy enthusiasts and everyone uses MLHFM to describe the MAF table 0-4.98v. Previously it was well known rescaling the MAF kicks in other limiters so its left and the ecu keeps timing and fuel the same as the last cell even though it’s off the map. So with a 2.75” VR6 MAF saturating at 320hp with VCDS Reading 187G/s you could run nicely upto 400hp even though the ecu was blind and unscaled.

Some of the ECU Files I have from a DIY tuner in Austria has successfully rescaled MLHFM, I haven’t trawled through the files and compared the threshold limits he had to change to make it work though.

 

[badger5]

interesting stuff

 

[Broken Byzan]

Difference is you probably have the official super mappack ols or kp file for the ecu so if they call it something different in there I don’t know.

I just have what was gleamed/cross referenced from the M3.8.2 Ols and the various xdf and ols from diy enthusiasts and everyone uses MLHFM to describe the MAF table 0-4.98v. Previously it was well known rescaling the MAF kicks in other limiters so its left and the ecu keeps timing and fuel the same as the last cell even though it’s off the map. So with a 2.75” VR6 MAF saturating at 320hp with VCDS Reading 187G/s you could run nicely upto 400hp even though the ecu was blind and unscaled.

Some of the ECU Files I have from a DIY tuner in Austria has successfully rescaled MLHFM, I haven’t trawled through the files and compared the threshold limits he had to change to make it work though.

This was what i was trying to say, however badly

 

[antwan64og]

I missed a key point reading back and just wanted to clear it up.


If you leave MLHFM as stock with a max reading of 673 Kg/hr, you can add bigger injectors, adjust FGATO, TEMIN, TLST, TVUB and get it running to map it.

Then using a bigger MAF housing in proportion to injectors means the load has shifted and the timing especially will really need lowering for everything above 2000rpm but fuelling should be close. The MAF readings will max out at 187 G/S and not be an accurate reading.


If you change MLHFM so that the MAF reads correctly to say 915 Kg/hr (253 G/S), then the Load (Ld) measured in ms will exceed 12.75ms (the maximum).


Load calculated as

Ld = K * (MLHFM/KHFM)


To effectively keep the load below 12.75ms when you raise MLHFM to be accurate; you need to change KHFM. KHFM is effectively a load constant (here’s the kicker) which means that a change here means every map using load needs rewriting.


So choose to unscale the MAF and have some work to do or scale the MAF and have a hell of a lot of work to do.

 

[Thomas_20VT]

What is easier when you hit 12,75ms load limit is to multiply KHFM by a factor so that load remains with numbers around 10ms and scale FGATO with the same factor so that fueling MAPs are 95% right and you take care of timing MAPs with some logging comparison before and after this mod and checking timing pull at the same time


I don’t even mention KLHFM as I don’t see the point in mis-calibrating this parameter

 

[mogsalie0010]

Hi guys.,,, im in search of a second opinion.,, I rebuilt my golf 4 20vt engine replaced rings,bearings etc and new turbo even, and had my cylinder head by engineers,for valve stem seals and reseating of my valves. Now all of a sudden when I wanted to start my car today I have got no compression on cylinder 2,3 and 4..,? Any suggestions.,,?

Sent from my HUAWEI CRR-L09 using Tapatalk

 

[Stuart B]

You should start your own thread really. This thread is about mapping A3 150/180 ECUs

 

[mogsalie0010]

My bad.,,, sorry abt that

Sent from my HUAWEI CRR-L09 using Tapatalk

 

[Stuart B]

Forum experts in these engines will be able to advise you properly in your own thread.

 

[mogsalie0010]

Thanks

Sent from my HUAWEI CRR-L09 using Tapatalk

 

[Thomas_20VT]

Happy new year to all !

Chapter: LOAD calculation

As there seem to be one main fear about M3.8 leading to people saying that this ECU is sh*t and should be replaced by either standalone ECU (which can't do half of what a stock ECU does!) or switch to ME7.5 (way better than M3.8 but require money and electrical skills)

I decided to investigate a bit with my stock AGU to figure out if there may be a way to tune those M3.8 ECUs for big HP figures.

The main worry was the side effects caused by scaling load by KHFM as all the MAPs using load as an axis will be wrong and must be redone. This can be a very time consuming process and the reason why most people don't bother with this ECU

Here is data from VCDS logs on my own car:

I am able to back-calculate theoretical Load purely from MAF (g/s) read by VAGCOM and using KHFM in the following formula:

Load [ms] = (4/2) x 3,6 x MAF x KHFM /1000 / RPM

where: 4/2 is the number of cylinders divided by the number of cylinders being ignited per revolution (4-stroke engine) and the other factors are only for unit conversion.

How to proceed from here:

First, scale your KHFM factor (43860 as stock value) so that you keep load in a safe area with enough marging against the 12,75ms hard limit.
Then, you scale FGATO (injector constant) by the same factor as you used for KHFM so that all the injection maps using load will still be correct.

The work you need to make on your own is then for timing advance where you should recalculate your load axis supporting points so that they match with the scale you made on KHFM.

This way, all should be functional at once and require only slight fine-tuning but as you would do anyway with any kind of ECU.

(on the graph, load calculation error range is very narrow between -0,7% to 0,9%. Considering there is already a 0,7% error only from VCDS accuracy at 7ms with only 0,05ms steps, I think that this formula is the right one)

Any opinion ? or data from some other vehicles ?

Next time, I will try to have a better understanding of fueling strategies as there are still some few parameters which I need to figure out first

 

[antwan64og]

Someone was using 1300cc injectors on nefarious with E85. I didn’t ever see that it was running nicely though.

You can rescale the load axis if desired, one configuration I have has 12ms instead of 10ms, some resolution is lost but it depends what you want.

 

[David Hook]

Hi,

I just discovered this forum and post today! I'm very interested in what you have written as I have been doing a lot of research over the last few weeks since buying a Golf with an AGU engine. The previously owner had it mapped after fitting bigger injectors (original MAF housing) and the car ran rich when cold. In the meantime I have fitted a standard ECU and orginal injectors which means I have a spare ECU to play with! I work with engine software as a day job but I'm learning quickly how limited the information is for the M3.8 in the aftermarket world!

When I get a chance I'll go through the thread in detail and see if I have any extra information.

Many thanks

David

 

[David Hook]

I missed a key point reading back and just wanted to clear it up.


If you leave MLHFM as stock with a max reading of 673 Kg/hr, you can add bigger injectors, adjust FGATO, TEMIN, TLST, TVUB and get it running to map it.

Then using a bigger MAF housing in proportion to injectors means the load has shifted and the timing especially will really need lowering for everything above 2000rpm but fuelling should be close. The MAF readings will max out at 187 G/S and not be an accurate reading.


If you change MLHFM so that the MAF reads correctly to say 915 Kg/hr (253 G/S), then the Load (Ld) measured in ms will exceed 12.75ms (the maximum).


Load calculated as

Ld = K * (MLHFM/KHFM)


To effectively keep the load below 12.75ms when you raise MLHFM to be accurate; you need to change KHFM. KHFM is effectively a load constant (here’s the kicker) which means that a change here means every map using load needs rewriting.


So choose to unscale the MAF and have some work to do or scale the MAF and have a hell of a lot of work to do.

Apologies if I'm asking a dumb question but I'm just getting up to speed with this. You mention the stock max data for MLHFM is 673 kg/hr (186.9g/s). I've seen the approximation that maximum power =MAF/0.8 therefore with am I right that the maximum power would be limited to approximately 233BHP unless the complete MAF or MAF housing are changed? As you mentioned, adding higher flow injectors only requires changing FGATO, TEMIN, TLST, TVUB if you don't need more than 186g/s air to meet your power requirement?

 

[Stuart B]

I don't think the Ko3s can make 187 g/s - so people add the inlet circuit from an S3 (FILTER, MAF, TURBO, FMIC, THROTTLE BODY, INJECTORS
I think if you go that far you just add some of the wiring loom to wideband and use an S3 ECU - or standalone

 

[antwan64og]

Apologies if I'm asking a dumb question but I'm just getting up to speed with this. You mention the stock max data for MLHFM is 673 kg/hr (186.9g/s). I've seen the approximation that maximum power =MAF/0.8 therefore with am I right that the maximum power would be limited to approximately 233BHP unless the complete MAF or MAF housing are changed? As you mentioned, adding higher flow injectors only requires changing FGATO, TEMIN, TLST, TVUB if you don't need more than 186g/s air to meet your power requirement?

Correct but stock injectors max out at the same time as the MAF saturates so you don’t even have to change injectors.
The stock ko3 turbo and tiny intercooler limit power before the injector/MAF combination does.

 

[gldgti]

Thanks for starting this great thread. I have recently bought a cheap 98 A4 quattro with AJL, and I'm also interested in learning how this ECU works. In the interests of contributing, I notice that Thomas_20VT asked if anyone wanted to supply some more data - I would be glad to contribute in this fashion.

I have had lots of VW's and audi's but this is my first 1.8T. Previously I have mostly tuned diesels, 1.9TD, TDI, 2.0TDI and 2.5TDI V6, my favorite. But the 1.8T is legendary, I had to have one to play with.

At present my N75 is broken, so I have added a manual boost controller (I had one sitting around) and a boost gauge. I have set up the MBC to regulate max boost to about 8psi at the moment. I notice that boost response is much faster with the mbc than with a dead or dying N75 .

Anyway I hope this thread is not dead, and I hope we can work together to generate and share some knowledge about this ECU.

Regards
Aydan

 

[antwan64og]

Your A4’s seem to have 12v wired to the OBD port(mk4 Golfs do too, A3’s don’t) so can be flashed through there if the ecu has a flashable chip 29F200 versus the earlier 28F200.

You can swap ecu’s to get a flashable chip and either immo off them or adapt them to the cluster.

Search online for an A4 bin file and either make an xdf or find a well populated one and the cheksum plugin for $10 for tunerpro makes life easier.

You can use even use an immo off m3.8.3 ecu if you want

https://www.audi-sport.net/xf/threads/t28-powered-aeb.170183/

Then you can work with the 018CG bin file that a lot of people use.

2 main constraints seem to be that some parameters are hardcoded so adjustments are not infinite and the feedback loops are limited which means if there is a hardware failure you can be left chasing your tail to identify it.

If you want to get stuck into this Pm me your email address and i’ll send you some files to get a head start

 

[AGUMK4]

Also watching this thread, have been toying with the idea of fitting a megasquirt ECU I have laying around or going me7.5 instead of running my M3.8.3 but would prefer to keep it original. I have logging and flashing capability so I hope to follow along and contribute where I can. I have some bins and data sheets I've gleamed, every thing is from the public domain so I'll gladly share. I spent a little while looking into the early motronic m series but found with no previous mapping experience I was soon struggling with a mountain of information. Thanks for reigniting my interest, taking the time to teach what you have learnt and keeping the faith for the cable driven crew !!

 

[ben4866]

Also watching this thread, have been toying with the idea of fitting a megasquirt ECU I have laying around or going me7.5 instead of running my M3.8.3 but would prefer to keep it original. I have logging and flashing capability so I hope to follow along and contribute where I can. I have some bins and data sheets I've gleamed, every thing is from the public domain so I'll gladly share. I spent a little while looking into the early motronic m series but found with no previous mapping experience I was soon struggling with a mountain of information. Thanks for reigniting my interest, taking the time to teach what you have learnt and keeping the faith for the cable driven crew !!

I've just recently carried out the m3.8.3 to me7.5 conversion and to be honest despite some people claiming it's a big job, it really isn't at all.

The hardest part is mounting the dbw throttle pedal but doesn't take long to weld a bracket for it.

My car is an early 1997 agu and as such had the most wiring to alter but I just de-pinned all the scuttle plugs and re-pinned to the aum pinout,
Same with the 10 pin to 14 pin plug swap near the battery, although some pins have to be spliced due to differing pin sizes.

I only swapped to me7.5 due to wideband facility, it's a much more advanced management system,
However there is nothing wrong with m3.8, and more than enough info out there nowadays for self tuning if that's your aim.

Even the old motronic 2.5 that I cut my teeth on (self tuned) is still working well on my dad's supercharged Vauxhall Chevette and that has a serious amount of power!

 

[gldgti]

I've been super busy at work of late but just in the last few days been thinking on this again - thanks for the information guys.

Just bought an aftermarket BOV to fit as mine is leaking a bit it seems, once thats in and set up I will start doing a few baseline logs to see where the car is at. I suspect I might need to do plugs/coilpacks soon as I'm experiencing breakdown/misfiring at high throttle/high boost on occasion.

once ive sorted these issues I will be looking at tuning it. Logs to come...

 

[antwan64og]

I hope BOV was an error in your typing and you didn’t buy a blow off valve!

A blow off valve after the MAF will dump air that has been measured.....
And the ecu relies on the MAF reading being accurate