The Mystery of my BMW Phenomenon Unravels - Thank you Andre and HPA!

Dear HPA Community, I first want to thank Andre and his amazing team for this incredible place where people like me can gather with their thoughts and questions. I would not have made it this far in this field that I love so dearly if it wasn't for Andre and his team. And by this far I mean just attempting to tune my own car. The work you have all done is the greatest service I have ever received. My gratitude is endless and I just want to give all the support I can.

This journey has been nothing short of insane and has cost me untold amounts of time and agony but there is finally some form of success. This is condensed but detailed write up of my work.

Subject: 1999 BMW 328iS (e36), Siemens MS41 ECU, M52 engine with an M50 Manifold swap, AFE Throttle Body Spacer, Larger Throttle Body, Traction Control Throttle Body Delete with New Silicone Boot and finally a Short Ram Air Conical Intake that I made myself with a heat shield. There are also Underdrive Pulleys and a Clutch Fan delete.

Tools: Alientech Kess V2 Master Read/ Write Module, Alientech ECM Titanium Driver Editing Software, Innovate Motorsports LM-2 Data Logger and wide band oxygen sensor with LogWorks3 Software.

Tests: Steady State/ Increasing RPM/ WOT Ramp Run. My car does something strange when it comes to the Increasing RPM test, I have added this write up to the bottom of the post.

Maps: MAF Scaling (3D in Volts)/ Injection at full load (2D in Milliseconds). Current changes and values shown in my Excel Spreadsheet.

Data Logs:

Concerns = Innovate Motorsports has a data logging software called LogWorks3 which is what I used. It has its challenges as accurate scaling of the data is required by way of voltage modification in the channel configuration settings. This can cause serious issues if done wrong and I might have fallen victim to this. The graphs shown are not scaled and are all WOT Ramp Runs.

12161900 = No Tune. The grossly rich top end is indicative of high amounts of un-metered air from the much larger intake manifold and other Post-MAF upgrades..

12211900 = Injection at full load: Entire Map Scaled to .90 Lambda using a Correction Factor calculation. This was done to set a static Lambda value in order to accurately scale the MAF Sensor. I also want to keep it here at .90 Lambda after I am done with the MAF tuning.

Concerns: Possibly inaccurate correction factor calculation and/or MAF Scaling as I will explain later.

12241900 = MAF Scaling: My best scaled map with near 0 STFT and no drive-ability issues. I tuned the map by rows and interpolated by percent to keep things as smooth as possible.

Concerns = The shape of the Lambda readout curve changed and became very bumpy. Also, the voltage readout seems to have decreased from a max of 1.4 volts (12161900) to a max of only 1.2 volts (12241900) in the unedited MAF readout. Why would the voltage decrease and is this the shape that represents the Injection at full load map?

12241901 = MAF Scaling: Attempt to fine tune the MAF Scaling and make sure I wasn't leaving anything undone. No positive changes present and STFT went significantly negative.

12241902 = Injection at full load: Entire Map Scaled to .90 Lambda using another iteration of a Correction Factor.

Concerns = Why would I need another Correction Factor calculation to scale the injectors to .90 Lambda again? Was this not correctly done the first time? After doing many Correction Factor tunes later the Lambda readout curve was only gradually flattening out. I found out that the first Correction Factor calculation corrects about 3% and each subsequent one is 1%. Why is the Correction Factor an iterative process? Do I need to multiply by 6 as the car has 6 cylinders?

12251900 = Injection at full load: Entire Map Scaled to .90 Lambda using Pure Guessing. A much better result was present with a massive anomaly. This is where the VANOS System kicks in. Could this spike be just that?

12251901 = Injection at full load: Entire Map Scaled to .90 Lambda using an Educated Guess by applying a Correction Factor not as a percent, but as an absolute change to the map values. The spike is still present but much shorter. A hill has appeared just downstream of this spike.

12271900 = Injection at full load: Entire Map Scaled to .90 Lambda using another Educated Guess from above. The spike is shorter still and the hill is still present albeit shorter as well. This is the most current log.

Moving Forward: I am of the mindset that more fine tuning needs to be done to get this Lambda readout flatter as there are 2 major lean areas present. I do not want to keep revving my car without knowing what changes are affecting what in my tuning. I will upload my results for your viewing pleasure as this was a complete mystery to me as to how my car's ECU literally seems to have a mind of its own. Any feedback will be greatly appreciated and please free to use this information if you find it relevant. Thank you all for your time.

Another Mystery

Request Lambda Map = ECM Titanium has a Requested Lambda map that I asked Alientech to update and they did as their customer service is excellent. While trying to use that map, I found it would not change the Lambda. My meter was always showing 1. I was puzzled and still am. It did however make my exhaust sound amazing with a backfire when set rich. Testing the map, I set it to .90 Lambda in its entirety and drove around and my meter was always showing 1.

Do I have to change another map along with this one to get it work? I did a trial of this method to no avail.

Is there some sort of background compensation going on that I am not aware of? There is a Lambda Correction Factor map but the slightest change to this and car will not operate under driving conditions. This is the main calculation table from that the MAF Sensor calculates fueling from based on air flow. I believe this is not meant to be tuned.

I even set this Request Lambda map to Max and then in another trial to Min. The car drove just barely at Max with terrible performance but was always reading a Lambda of 1. The car almost didn't move at Min.

Another interesting Lambda related situation is that this car does not go into Open Loop based on RPM. There is a map that shows it is TPS% dependent. I have done a test where I have revved the engine up to 5000 RPM under partial load and throttle with a Lambda of 1. I stopped immediately to avoid damage but this car seems to love Lambda 1 and nothing else and I cannot change that with this Requested Lambda map. This test did come in very useful however when checking the STFT. It was refreshing to see that my MAF Scaling was well accepted with an STFT average of 0.

I do not know what this map is but it would have been easier to use if it worked or if I knew how to use it. It does seem to do something but what that is, I do not know.