NefMoto

Hi dears,
Just sharing here my project, it is a MB C230 SportCoupe V6, I just did an engine swap from the original M272E25 to a M272E35 (V6 NA 3.5L,that was originally from an ML350).

So far everything looks pretty good, initially did a light stage 1 without anything else than limiting the lambda at 0.85, increased the noise tolerance of the knock level recordings, and maxing the airflow restrictor. With only these modifications everything felt way more reactive than before, not based on anything else than buttometer, but I think I was objective.

Now, I made a second version as I will be swapping the manifold to a fixed manifold (from SLK350 R171), BMC air filters, and changing the cats to some magnaflow sport cats I had from a previous project never used.

I'm all open to suggestions, as for me this is just a hobby, my first job was at a tuning company but after a year pivoted to be developer / devops / automation engineer, so I kept it as a hobby for years and did my things mostly by comparison, with damos for patterns, and with other tunes to undestand approaches, this year decided to pay actually a training to understand better the concepts, get better at identifying patterns, and to kick start my journey on reverse engineering which is something I tried a few times and never got the idea on how to approach, now at least I understand a few methods to get to the places I want, to manipulate the code and etc, I think I needed just a kick there.

These are all the modifications I made:

KFKE*: Raised the level +25% (not by percentage actually, but I did it by +10 raw value), I understand that increasing the noise / characteristics tolerance will give me around 3 ~ 4 degrees of tolerance for knock detection
KFKRFKN (Knock retard, accumulative): Reduced the steps from the original -3 degrees to -0.750 as doing this will accumulate smaller steps back when detecting knock
LAMFA (Lambda on driver's wish): Just made the high load section a little richer, originally 0.92 and I moved it to 0.88
KLRLNMXN: Maxed to top, so basically removed this airflow restrictor
KFMIRL (Desired charged, VE): I basically flatted the decrease after the peak on high load, on this map I'm actually not completely sure if the effects will affect on anything on this NA car
KFMLDMX: I understand this one is just for diagnosis to not avoid the ECU detecting something is off, again on this one just increased the high part
KFMIOP (Max engine moment, OptTQ): Raised high load on +6%
KFZW* (Timing maps): Advanced the timing between 3 and 1.5 degrees, only at high section

I think my main doubts are around KFMIOP and KFMIRL, like most people does anyways as I've seen around.
So, doubts, on my training I understood that the idea is that to get an actual increase the objective is to to increase VE (KFMIRL) and reduce OptTQ (KFMIOP), as far as axis do not get modified to allow extrapolation do the magic, now if you check on my file I actually increased both, by different %, but I'm 100% honest, the file I'm working on is 100% original as I'm working on an ECU I flashed and virginized when doing the swap, the tune is 80% my work, and on this two particular maps (KFMIOP and KFMIRL) I made by comparison with another tune.

I have not tested this yet on the car as I'm waiting for the manifold to arrive, and will work on those set of physical changes by the end of this year hopefully.

Aside the doubt on KFMIP and KFMIRL, can you share a general opinion / recommendation / correction on the approach?

Sorry, forgot the second attachment, here the tuned file. If it helps I can also attach the winols project to share the definitions

If you don't care about emissions and efficiency, go for Lambda 0,85 as soon as you tip the pedal ;)
It will give you maximum torque right away.

Watch knock closely. With only 0,75 Degree knock retard per event the knock control reacts more slowly on knock. If heavy knock occurs out of the sudden, it might be too little retard per time.

Could work in moderate temperatures with good octane fuel.
But could blow on a hot day with medium octane.
Or when somehow some oil is entering combustion chamber.

I would raise it again and lower Lambda below 0,85 on spots where knock occurs. Typicaly the engine has one spot for a certain rpm range, depending on mods.

Good luck.

Thanks Daniel for the feedback!
Well, I modified this:
KFKRFKN (Knock retard, accumulative): back to stock -3
LAMFA (Lambda on driver's wish): looks like this now.

69.9982 79.9988 94.0002 100.0000 110.0006 120.0012

1000.0   1400.0   1800.0   2200.0   2600.0   3400.0   3800.0   4200.0   4600.0   5000.0   5400.0   5800.0   6000.0   6200.0   6400.0 1.0000   1.0000   1.0000   1.0000   1.0000   1.0000   1.0000   1.0000   1.0000   1.0000   1.0000   1.0000   1.0000   1.0000   1.0000 1.0000   1.0000   1.0000   1.0000   1.0000   1.0000   1.0000   1.0000   0.9062   0.9062   0.9062   0.9062   0.9062   0.9062   0.9062 0.8828   0.8828   0.8828   0.8828   0.8828   0.8828   0.8828   0.8828   0.8828   0.8828   0.8828   0.8828   0.8828   0.8828   0.8828 0.8828   0.8828   0.8828   0.8828   0.8828   0.8828   0.8828   0.8828   0.8828   0.8828   0.8828   0.8828   0.8828   0.8828   0.8828 0.8516   0.8516   0.8516   0.8516   0.8516   0.8516   0.8516   0.8516   0.8516   0.8516   0.8516   0.8516   0.8516   0.8516   0.8516 0.8516   0.8516   0.8516   0.8516   0.8516   0.8516   0.8516   0.8516   0.8516   0.8516   0.8516   0.8516   0.8516   0.8516   0.8516

If I judge the engine based on the knock recordings, hardest noise looks to be around 2400 RPM, is it correct to use those cylinder noise recordings as a guidance?
For example cylinder 1:

39.000 50.250 65.250 80.250

800.0   1200.0   1600.0   2000.0   2400.0   2800.0   3200.0   3600.0   4000.0   4400.0   4800.0   5200.0   5600.0   5920.0   6200.0   6520.0 2.8125   2.8750   2.9375   3.2500   3.1250   3.1250   3.0625   3.2500   3.1875   3.0625   3.2500   3.0000   2.7500   2.5625   2.5000   2.5000 2.8125   2.8750   2.9375   3.2500   3.1250   3.1250   3.0625   3.2500   3.1875   3.0625   3.2500   3.0000   2.7500   2.5625   2.5000   2.5000 3.0625   3.1875   3.2500   3.3750   3.2500   3.2500   3.1250   3.1875   3.1875   2.9375   3.0625   2.8750   2.7500   2.5625   2.5000   2.5000 3.0625   3.2500   3.3750   3.3750   3.2500   3.2500   3.1250   3.1875   3.1875   2.9375   3.0000   2.8750   2.7500   2.5625   2.5000   2.5000

1. You need to start logging instead of making random changes.
2. Doing anything but 1.00 lambda below 2000 rpm is really stupid. It will only result in lots of black soot and horrible fuel consumption.
3. You need to log the axis on the LAMFA map, because how do you know what is 110 or 120?
4. Leave the knock detection maps alone. You have no idea what you are doing - increasing the integrator delta will not give you any more power. You will damage the engine and LOSE power doing it.

The only thing you can change is mixture, timing and camshaft adjustment.
For mixture and cams you need a dyno. This 0.85 suggestion in this thread is retarded, these engines don't make peak power at 0.85.

The only thing you can change without a dyno is timing by adjusting it closer to the knock limit on your fuel (if there is anything to adjust).
So unfuck all the maps you messed up, get a logger that can show you rpm&load and per-cylinder knock retard, and advance the timing until you find the optimal one.

That's it, you're done. Unless you can get to a dyno and then see if there is anything to gain with cams and find the optimal mixture profile.
But we're talking max 5 hp here for a ton of money. I wouldn't bother, and I have my own dyno.

Or just run the stock software, because I guarantee your changes up to this point have made it worse.

Thanks for the constructive feedback.

1. You need to start logging instead of making random changes.
- I agree logging is important, I got the plugin for OLS to use vehical but still not paying the subscription as mentioned earlier I won't be running this tuning till December when I get the parts to do physical changes first. The actual file I'm running just removes the KLRLNMXN restrictor, and limited lambda at 0.85
2. Doing anything but 1.00 lambda below 2000 rpm is really stupid. It will only result in lots of black soot and horrible fuel consumption.
- Yes, I get it, the map is navigated in diagonal, the rest of the values are adjusted just to keep coherency, as the stock values are also there set kind of linearly despite 2 extremes will never be used.
3. You need to log the axis on the LAMFA map, because how do you know what is 110 or 120?
- Same as on one, I agree logging is correctly. However I understand that % (mrfa_w in this case) refers to the torque requested through the pedal (so KFWAZU vs KFMIOP and consequently by KFMIRL) so as it determines de delta % of actual moment vs requested. As said, for me this is a learning path, so checking if my line of thought is correct.
4. Leave the knock detection maps alone. You have no idea what you are doing - increasing the integrator delta will not give you any more power. You will damage the engine and LOSE power doing it.
- Ok

Just got vehical installed while writing the post, will try getting it to work with OLS sometime this week and revisit after I have some logging to analyze the maps with valid data.

VehiCAL logger does not have advanced for your ECU, so that's fairly limited as well.
Better than nothing though.

I don't think there is any sense in tuning a NA engine unless you are going to run on E85.

I suggested 0,85 Lambda for peak torque.
That's what he wants, more omph when hitting the pedal.

No need to call anyone retarded when the goals are unclear.

In my opinion not peak power in first attempt when using the "Popometer" or "buttmeter" approach is used as mentioned.



Back to topic.

Go out and log knocking.
The Daimler with that engine i had access to, had the knock area more in the 3500..4000 rpm area. When i remember correctly 0,82 was what we used there. In the top rpm we went to Lambda 0,92 around for the mentioned peak power.
Car was happy and felt good through the whole rpm range with that approach. First peak torque, then adjusting Lambda for higher rpm. Never went to dyno because we didn't care :D

No guarantee that anything of those data works well on your car.

Lamfa Axis should be taken care of, correct.

As i'm from Germany, we need to stay at Lambda 1,00 for any load up to about 1800 rpm du to emission tests every two years. That's the point to start lowering Lambda for me, works usually well, not too much smoke :)

0.85 is not peak torque on every engine. Suggesting that is retarded. I didn't call you retarded, I called the suggestion retarded.
Have you ever tuned an AMG or MB NA engine? What about a Subaru? Or an Alfa NA V6? Do you know what they all have in common? None of them make best power at 0.85.

Hence you have no idea what you did.
I do, because I've had all of the above on my own dyno.

Unless you take that car to the dyno don't bother adjusting the fueling on it, the factory is quite good.
Set the timing to best that can be run on the fuel you have and job done. If you do some dyno time, then move the inlet camshaft, adjust the fueling, you can get the curve a tiny bit better, but not worth it imo.

Yeah ok, no bullying needed. ;)

Yes, i tuned that specific MB engine as i said. Lambda 0,85 at peak torque felt best and gave best maf readings there. Thatswhy i replied and suggested it.



I tuned MB NA V8, 0,85 was not best. Leaner needed, just as you said.
Volvo NA engines (5, 6, 8 Cyl Yamaha) work best with 0,85 in mid range too.

No Subaru or NA Alfa so far.


Let's wait what thread creator finds out about peak torque.

Thanks Daniel for the feedback again.

Yep, I don't have much mods right now on the ECU, so will try first getting some logs, will give a try to vehical later today, and if the information there (as PRJ said it may not have what I need) is not enough I will give a try to vediamo if I can find out how to log stuff with it as I know it has a section for data logging but never used it, just for coding so far.

PRJ, when you say to adjust the inlet camshaft is camshaft position?

Regarding the goals, I'm not tuning for power here, the car is at around 290hp right now which is more than fine for this 1200kg car, I do these things for learning and experience, just that, so while the whole goal is experiencing how the things I modify to affect the engine and in consequence either power or economy are part of those ffects, I actually don't care much about the results perse because my goal is not to turn this into a Ferrari that it is never going to be, the goal is just learning. So for answers like tuning a NA engine is not worth it, thanks, just that it is an answer to something I never asked.

Regarding the peak power, it is clear at least for me 0.85 is not for peak power, if theory is fine 1.05 is peak economy and 0.86 is peak power, but that is just a guide, I understand that is a per engine definition and not something universal. I just made it richer on the top end because I only adjusted the top ends, I don't see a reason to adjust the mid to lower section of the car if I'm actually not doing anything interesting in there. And as timing was moved by 3 degrees (base timing, not optimum timing), made it a little richer just there at high load high revs, from 0.92 to 0.85

Elaborate some more on the connection between MAF readings and mixture target. Go on.
Keep digging that grave.

If anything my decade long stint into calibration taught me is that anyone who uses "feel" or "smooth" when describing engine tuning, does not know shit.

Don't touch the OEM mixture. It's more than likely not running what is in LAMFA anyway, on WOT it will be BTS depending on ignition angle efficiency.
Only do this if you have a dyno and can see the curve and the differences between the adjustment.

And anyway, don't touch anything before you logged the car stock. If you already have knock at stock timing then advancing it will lose power.

But hey, do whatever you want. I've had lots of these butt engine dyno tunes on my dyno which made less power than stock.
And what is described here - +timing +fuel is 99% of the time exactly what was done.

Yes, but again, without a dyno you can not see if it makes more power or not.
Looking at MAF numbers does not really tell you anything, because adjusting it too far into overlap won't necessarily increase cylinder filling, but sure the MAF number will be higher.

Hmm, don't see a problem here, neither any grave ;)

Raising ignition for better octane. Knocking somewhere, maf readings are dropping there.

Raising fuel, knocking prevented, maf readings raising.
More torque on that specific engine point. Not touching cams.


That worked in my two decades of non-pro calibration in my spare time for most of the cars including that with M272 3.5
Im talking for getting max torque, for which this engine is designed. Not others including the ones you mentioned.


In my opinion you could respect that there are other possibilities than a dyno to make a car owner happy. What if you have a nice looking curve on dyno but owner is not happy with the real feel?

I totally agree that you only make best graph and real numbers with a dyno. But if that is not the goal, then there are other possibilities for improvement.

Call that retarded, call it not knowing anything, call it what ever you want.

Why not wait what thread creator reports about richer or leaner Lambda? Sure he will not have good looking graph, maybe less power. But a smile each time he accelerates his car. Maybe some day he will go to a dyno to make it even better.

Knock or mixture does not affect air flow through the engine on N/A unless the exhaust is hugely restricted. You would have to create artificial conditions in a lab to see it.
It affects combustion efficiency or the power extracted from the same amount of air.
On a turbocharged engine retarding ignition and having knock will always increase airflow, because there is more energy left over for the turbocharger.
On a N/A engine it makes no difference unless you put a restrictor in the exhaust in a lab.

If you want minimum EGT (thus smallest possible volume of exhaust) -> theoretical maximum airflow on N/A through restrictive exhaust, then you should run lambda on rich combustion limit (0.65 or so), does that make max power?
Of course not.

To know what the optimal fueling is you need a power measurement, there is nothing to discuss here.

I respect facts.

You are pushing bullshit with no correlation with applied physics.

I have called out people like you on this forum on their bullshit for a very long time, and I will continue to call you out.
Push your snake oil somewhere else.

Just goes to show that you can play excel calculator for 20 years without having a clue of how the combustion process in the engine works, and what an air pump is.

See, you think that is the truth. I doubt that it is true under any circumstances. Two opinions.

Here is my contrary opinion:

Bad combustion efficiency due to knock retard leads to less energy moving the engine inertia. This meaning slower engine rpm increase under similar load conditions. resulting in less air in the NA engine at a given point in acceleration phase. since the airflow is correlated to engine speed on NA engine very much.

This results in less torque at a specific point during acceleration and the feeling of a weaker car/engine.

I mainly work on forced induction engines, right. And my expertise is embedded software development.

What i observed on non-pro calibration on NA engines was what i posted. My explanation is written above. Might be wrong explanation, but the observed and the approach derived from it stands from my side. Going richer with increased ignition using higher octane fuel in rational steps might help and helped me on that specific engine. Helped me on many Volvo B5254S, B6304S and B8444S. Helped on LS3... 0,85 is the point of highest possible torque. Need to discuss this too? Surprisingly on all these engines including M272 it was the value for best acceleration and MAF readings during acceleration.

That's why watching MAF readings during that process helped me to find the "sweet spot" of best acceleration.

Again, im not talking about highest power at high rpm. Not about the prettiest curve. im talking about peak torque in an acceleration run. I have the impression, you intentionally mix that up?

Don't know why. To call me out? Do that.
To justify your Dyno? No need, it has its eligibility.
Not spilling any poison here.
Giving information about what worked for me to others - that's what a forum is for. And can be for in parallel to your usage of it to advertise your business ;)

Luckily it is up to everyone by itself decide to use that approach, or go to your shop or do something completely different :D

There is no opinion. There is fact, and your fairytales.

I really recommend picking up a few books and improving your level of ignorance. What you said here is complete and utter bullshit...
You are trying to invent your own definition of an air pump and physics here.

There is nothing else to respond to it. It's like arguing with a flat-earther.

While Trying to sound cool throwing around random technical jargon might get you rep with your buddies in real life, on here there are people who actually know how the internals of the engine work.
Trying to cast facts as an "opinion" while talking complete nonsense is not going to fly here. Quit the bullshit.

Ok, as i said. My opinion / explanation for the observed might be wrong.

I double checked some recent logs. On an LS3 V8, not realy that close to an M272, i know, i had a rise of the mean airflow by 3,5% when dialing Lambda from 0,92 to 0,84..0,81 while increasing ignition some degrees.

How can that be explained? The process was two or more runs with same software, same gear, same evening, same people in car. Minutes after each run. Mean value of Maf readings over rpm, mean of the runs.

Same approach gave nearly no changed MAF values on an SLK 55 AMG NA, but still a faster time 0..100 Km/h measured with V-box.


So where is the error in my explanation? Engine speeded up by energy from combustion. The more energy, the faster speeding up per time. Is that agreed or an error already in your opinion?


I realy would like to understand instead of throwing stuff at each other.

As said above, on NA engine if exhaust is very restrictive, then anything you do to lower EGT (which lowers the volume of the exhaust, because cooler gas has less volume) will increase airflow through the engine.
The richer you set the mixture and the more you advance the ignition in that case, the higher the airflow will be. Does not tell you anything about the power whatsoever, because the highest airflow with a very restrictive exhaust will be at combustion limit, but certainly the engine will not make the most power there.

Another thing is confirmation bias, it can easily be that after a few runs the intake has cooled down and was less heat soaked.
This is why a conditioned dyno cell helps where the car is being cooled all the time and you can actually compare two runs.

Hell, on the street a change in wind affects MAF readings. The MAF absolute reading is actually completely irrelevant for tuning, especially fine tuning, most engines.

This is complete bullshit in this context, see above. Crankshaft acceleration rate itself does not have any impact on airflow. You can even do a reverse pull on the dyno, and it's all the same on NA.
What affects it is EGT, but best EGT is not best power, far from it.

You are observing changes in exhaust density and/or changes due to flawed measuring methodology. Then coming up with random things not rooted in reality to explain them.

Yes, that sounds like e better explanation which i didn't thought of so far. Thanks.

Makes even more sense since the exhaust of the LS3 Crate Engine seems to be very restrictive. Should have 525 Hp with factory Dataset, on dyno it only had around 470 Hp.

Same goes for the Volvo exhaust on 5 and 6 Cylinder.

SLK 55 AMG has obviously a good exhaust.

So maybe this method only leads to identify a very restrictive exhaust, which a dyno can do probably faster than driving around all night full accelerating (which is fun too). :D


What is left are sometimes (not always) faster acceleration times on some NA engines with Lambda around 0,85 in mid rpm range instead of 1,00. Probably in cases where factory Dataset is very "emission friendly".


I still need to find out why my explanation / opinion is so much "bullshit", but no need to flood this thread even more.

Thanks for sharing the info to both, prj, honestly I appreciate all your feedback because it is actually constructive, just that sometimes I have to filter out comments that feel you are throwing stuff at the others for no specific reason, at least in my case I have never challenge your knowledge or feedback, just shared what I did and asked for opinions as I think this forum is way more technically oriented than others I have read over the years, and basically that was the reason I started my thread just asking for opinions / feedback stating it was just for learning, not for mere power or tuning a NA which I get is a lot of a effort for probably unnoticeable power (probably is more about changing how the power is managed than making more power) to avoid getting the standard answer of not worth it, don't do it, etc. Anyways, thanks both.

So, back to the origin of this thread, I got the car back to stock as I has having some weird issues and as daniel my base come from development and automation, so wanted to go back and depart from a  known stable base, but nothing to do with tuning, feels like I'm having some wiring issue (or something around that area), ign coils not working, misfires and so on, thought it was a dead driver on one of the ME9.7 I had, so took my previous one that was a known working unit, and yet same issue. So solving that first, then getting back to this again, meanwhile bought the CFF plugin from EVC as I wanted to flash my ECU (with the tools I have) without opening, so KTAG was not an option, so learned how to do it with vediamo yesterday after a few sweat tries got it working.

will get back to the thread hopefully soon as I get that stuff sorted

Yes, 0.85 will probably make more power than 1.00 on most engines.
But maybe 0.9 will make more. Or in case of overstroke engines they often need 0.8 or lower.

You have no way to find that out without a well repeated power measurement. Certainly not by looking at measured airmass or anything else.

Just as a follow up, this was not abandoned, couple of things changed though in the meantime. ME97 broke (typical on MB because of high temp), fixed it as it was bad drivers. Returned it to stock as I want to do some maintenance on the engine before pushing it further. Also decided to get a dyno as recommended by prj before, so preparing a proper place to tune like it has to be done, will take me a few months though to prepare the whole setup (2 or 3 months at least).
Meanwhile I got my reverse engineering skills really polished, had a few trainings to be honest with reputable people, unlocked me immediately in lots of things I was missing, strategies and so on, learned a lot more on how to use A2L to reverse engineer and so on, so investing time there, mainly on doing a multimap for ME97 so I can have stock and tuned params/maps during my testing periods, honestly, while not really invested time in tuning perse I really went forward with stuff I was more interested on like reverse engineering