Twin Turbo 3.7L V6 part throttle stumble under boost SOLVED!

[Boody78]

Finally after 2 years of trying every method I could find on this forum and the cookbook I have found a solution to my issue.

What I have deduced by reading lots of posts is that the way the torque control system works is this:

1. Driver presses the accelerator a certain %
2. The ecu looks up the pedal % in the driver demand table and determines how much torque the driver wants.
3. The ecu looks up the the torque requested in the torque inverse table and determines the value needed to achieve the desired torque.
4. The ecu uses the value from the inverse table and sensor data from the engine in a complex calculation(which I don't believe we have all the parameters needed to tune) in order to command all the necessary outputs needed to achieve the desired torque.
5. The ecu has no way to measure actual torque achieved so it preforms another complex calculation (which I don't believe we have all the parameters needed to tune) to calculate the achieved torque output.
6. The ecu compares the desired torque to the calculated torque it believes was actually achieved to make sure the engine is operating correctly.
7. If the ecu believes the engine is producing more torque than is requested it will close the throttle to prevent the engine from racing away which results in stumble.

What I believe the issue is:

1. The complex calculations the ecu uses to determine how to command the engine to make the correct torque and to then calculate how much was actually achieved do not expect to see boost and we are not able to adequately tune all the parameters the ecu uses in these equations.
2. This is compounded in the s550 3.7 because it does not have a MAF sensor and is calculating MAF based on MAP and SD calculations.
3. This causes the equations to fail under boost and calculate the achieved torque incorrectly higher than reality.
4. This causes the ECU to close the throttle and the engine to stumble.

This is unrelated to IPC errors which is why chasing them does not seem to help and the stumble continues even after maxing out the IPC max error.

The solution that finally worked for me:

We need to fudge the numbers in the torque table so that the ecu thinks the torque it requested matches the actual output.

Log the following parameters:

Engine RPM
Air load
Accelerator pedal position
Engine Brake Torque
Desired Brake Torque
All 14(and OP if you want to use it) mapped point weights

Create a user math parameter (Engine Brake Torque - Desired Brake Torque) This will tell us how much more torque the ecu thinks the engine is producing than what it asked for.

Make sure you scale the row axis on your torque tables to reflect the higher load you will achieve with boost.

Using the row and column values from your torque tables Create a graph for each mapped point with air load on the row axis and engine rpm on the column axis populated with the user math parameter created above. Add filters for accelerator pedal position > 0 and mapped point weight > 25

DO NOT ATTEMPT TO LOCK IN A SINGLE MAPPED POINT AND TUNE ONLY THAT TABLE!. The mapped points are designed to work together and tuning them individually as suggested by other methods caused me many headaches.

Drive around and datalog.

Only concern yourself with positive values in the graph. Using the average, any cells with values over 100 reduce the value in the corresponding torque table by 5%. You want to sneak up on it, don't go crazy. I found that as long as the values are bellow 100 the stumble went away for me.

DO NOT RECALCULATE THE INVERSE TABLE! Leave them stock. Remember we are trying to fudge the numbers if you recalculate the inverse tables you just replicate the issue with different values. We need the tables to be wrong for this to work.

Repeat the process until the stumble stops AND THEN STOP! don't try to make it perfect that will cause more issues.

Note: This assumes you scaled all your tables properly and have accurate fueling and map sensor data. THIS IS NOT RELATED TO IPC ERRORS that is a separate issue.

This method has worked for me and is the only method that ever completely removed all stumble from my tune.

Brake TQ Err.MathParameter.xml TT vs MP.Graphs.xml

[veeefour]

Create a user math parameter (Engine Brake Torque - Desired Brake Torque) This will tell us how much more torque the ecu thinks the engine is producing than what it asked for - wrong
DO NOT ATTEMPT TO LOCK IN A SINGLE MAPPED POINT AND TUNE ONLY THAT TABLE! - wrong
DO NOT RECALCULATE THE INVERSE TABLE!- wrong
THIS IS NOT RELATED TO IPC ERRORS that is a separate issue - wrong

[Mistic67]

Hey @veeefour, would you be able to explain how to tune for higher torque numbers? I'm still a novice at tuning my ecoboost. After 4 months or so, I only just finally leveled out my boost. I can get up to 500 engine HP, but my torque is only maxing out about 480-490 lb-ft. Would you mind providing some insight?