A quick and easy way to street tune your CSL conversion for drivability.

[heinzboehmer]

Thanks a lot for this Pavlo! Just finished the warm tune and my car drives much better now. My car needed fuel in a lot of places (much like Bry5on's):



Rev matching is more responsive (have yet to get the engine to stumble like it sometimes did before) and the car feels a little livelier. Won't go as far as saying that it feels more powerful, but it does feel like it struggles less with around town driving. But by far, the most noticeable change is the engine behavior at very low throttle inputs, like when pulling away from a stop or changing gears. Before, the rpms would oscillate quite a bit and make things pretty jerky (sidenote: hadn't realized I was driving around this until now), but that is completely gone now. Feels sooo much more refined. No more weird stumbling when pulling away from a light.

Next up is applying the same process to the warm up map. It's painfully obvious when the car switches over from that to the warm map now. For now, I interpolated the final scaling factors (what's in the table above), as the Relative Opening and RPM axes are slightly different, and applied them to the warm up map. Will probably data log a couple drives, but I'm betting it will be pretty close.

Also, I took some notes while tuning and figured I'd share as they might be useful to others:

• Testo
  • Under "MSS54 (SERIAL)", open "STATUS_MOTOR_1" and "STATUS_MOTOR_2"
  • Log "relativer Oeffnungsquerschnitt", "RPM", "Lambdaintegrator 1" and "Lambdaintegrator 2"
  • Logging "Motor Temp." is useful for the warm up tuning
• Megalogviewer
  • Need premium version to apply settings file
  • Can replicate settings file with free version, but it's more work
  • Did not bother adding any filters to the data, just made sure to datalog when warm
• Integrator Data Interpretation
  • Initially averaged out integrators for both banks and used that, but bank 1 consistently wanted more fuel than bank 2, so ended up just using bank 1
  • Seemed to be off by a fixed offset, so decided long term trims could take care of it
  • Would have gotten there eventually with average, just would have taken longer for the same result
  • Kept iterating until I saw most integrator values oscillating around 1
• Linked XDF
  • Truncates decimals to int in both warm up and warm VE tables
  • Table stores values as decimals with 1 decimal point, but XDF makes tunerpro display them as ints and thus truncates any decimals (does not round) -> affects next iteration calculations
  • To fix, in parameter tree, right click "Density Correction_VOLUMETRIC EFFICIENCY (RF)", go to "Edit Parameter XDF Info" and change "Output Type" to "Floating Point"
    • Do the same for warm up map
• Flap map
  • I do not have a flap, so I zeroed out the entire flap map
  • Map contains additive compensation factors, so it's effectively not doing anything now
• Logging strategy
  • Mostly drove around normally, to try and capture as many data points as possible when I spend the most time
  • After a few iterations, there were some areas where engine was clearly still stumbling, so I focused on those combinations of throttle/rpm for a couple iterations
    • Found hills with no traffic to load up engine and be able to slowly climb up the rev range at a fixed throttle input
    • Think of replicating how the car would behave on a dyno

Lastly, here's my version of Pavlo's spreadsheet. Not the most readable thing in the world, but has some useful things like a (very hacky) way to autofill 1s and the warm up interpolation strategy: https://docs.google.com/spreadsheets...it?usp=sharing

[Bry5on]

Update: Bank 1 long term fuel trim is 1.01 and bank 2 is 1.00 after turning adaptations back on. That’s as good as it gets!

[Bry5on]

Wideband for this-- should we put it pre or post cat?

I run my wideband pre-cat, it’s most useful for the open loop WOT tuning

[Bry5on]

Do you use map sensor as kassel kit? When it installed on intake, or kit when it is installed near to ecu box? Or original csl tube for map?

I use a map sensor in the ebox area, connected to the brake booster hose

[Obioban]

Wideband for this-- should we put it pre or post cat?

[pnvhome]

For the record, here’s the difference between the default CSL tune and the 5 runs in version using this method on my car. Super clever, thank you for sharing!


Throttle tip in is so much smoother and the motor seems much happier. I agree that a good amount of the rasp is gone in that mid-2k RPM range. Looks like those CSL cams and the SS headers really do make a difference. With these headers, I needed fuel almost everywhere, and particularly in two islands.

My motor is a 180k mile stock internal CSL conversion, with flap, SS v1 headers and matching catted section 1. This tuning was done with flap closed to the appropriate map.

Adaptations are now turned back on, so I’ll be watching what the motor does.

edit: I’d bet that the resonances and jitteriness in the 2k RPM range can be totally tuned by changing the interpolation motor RPMs. I did this for my WOT tune and was able to get a super flat AFR curve by targeting the resonant and nodal frequencies as the interpolation points.

Do you use map sensor as kassel kit? When it installed on intake, or kit when it is installed near to ecu box? Or original csl tube for map?

[Bry5on]

For the record, here’s the difference between the default CSL tune and the 5 runs in version using this method on my car. Super clever, thank you for sharing!


Throttle tip in is so much smoother and the motor seems much happier. I agree that a good amount of the rasp is gone in that mid-2k RPM range. Looks like those CSL cams and the SS headers really do make a difference. With these headers, I needed fuel almost everywhere, and particularly in two islands.

My motor is a 180k mile stock internal CSL conversion, with flap, SS v1 headers and matching catted section 1. This tuning was done with flap closed to the appropriate map.

Adaptations are now turned back on, so I’ll be watching what the motor does.

edit: I’d bet that the resonances and jitteriness in the 2k RPM range can be totally tuned by changing the interpolation motor RPMs. I did this for my WOT tune and was able to get a super flat AFR curve by targeting the resonant and nodal frequencies as the interpolation points.

[pnvhome]

I was absolutely dummy) Made about 5 runs, last one was almost absolutely ideal, also changed WOT table from the topic near this, and cleared FLAP table, i have turner intake. So now car runs almost perfectly just a little jigs on 2200rpm with 4 celcius outside, also there is no more bad resonations in exhaust. And car runs absolutely smooth even from 1800rpm
also i think problem in jigs becouse of map sensor with kit from kassel, on logs it shows that to many pulsations

[chapmans]

Any potential for a 'dummies' guide to this? I have edited tunes very basically but never been in the tables or anything.

[Bry5on]

Got two sessions in and it was super clear in the map that one small ‘island’ in the tune was sending my long term fuel trims to ~115%, causing every other spot to run rich. Thank you! Really clever solution you’ve come up with here

[SliM3]

Mind sharing the most up to date/accurate one?

I don't mind, but I need to clean it up and get rid of all the custom stuff.

I'll post when we get back home in a couple days.


Sent from my iPhone using Tapatalk

[Obioban]

I'm the author (Brian H.) and don't mind that it is shared, albeit I've probably made quite a few revisions to that XDF since it was posted.


Sent from my iPhone using Tapatalk

Mind sharing the most up to date/accurate one?

[SliM3]

Although you can remove all transients using a filter in Megalog Viewer (they have a tutorial just for this), Lambda Integrator correction is a closed loop system only, it disables itself otherwise and defaults to a 1. This really only tunes the drivability portions (closed loop) you are on your own for the rest.

And yes, you need to have a lot of data points and rely on the power of large data sets, you don't want to make a change with very few hit counts.

If the lambda regulator is constantly shutting down due to accel enrichment you can change the value at which they cut-off using K_LA_BA_OFF_POS (shuts down regulator if resulting accel enrich factor exceeds 1.04) and K_LA_BA_OFF_NEG (shuts down regulator if resulting accel enrich factor falls below 0.94).

[SliM3]

I forget who the author of this one is, it was on the old M3 forum MSS54 master thread. I did not make it, but since it was publicly posted in the past, I don't mind sharing it: https://drive.google.com/file/d/1nyV...ew?usp=sharing

If the author does not want this XDF shared, please let me know and I'll delete the link.

I'm the author (Brian H.) and don't mind that it is shared, albeit I've probably made quite a few revisions to that XDF since it was posted.


Sent from my iPhone using Tapatalk

[Pavlo]

How do you filter accel fuel because you must not correct your VE table when accel fuel is ON ? Reving up with constant pedal position can be a solution.

Although you can remove all transients using a filter in Megalog Viewer (they have a tutorial just for this), Lambda Integrator correction is a closed loop system only, it disables itself otherwise and defaults to a 1. This really only tunes the drivability portions (closed loop) you are on your own for the rest.

And yes, you need to have a lot of data points and rely on the power of large data sets, you don't want to make a change with very few hit counts.