Tuning Wiki

This is a historical revision. Read the current page or return to its history.

Page organisation in this revision
Page type
Technical article
Subject
Tuning
Parent
Bosch MED9.1
Applies to
Bosch MED9.1
Tool groups
None
Manufacturers
Audi, SEAT, Volkswagen

Application and starting point

Volkswagen's SSP 337, pages 6 and 20–21 identifies MED9.1 on the AXX 2.0 FSI Turbo: a four-cylinder, four-valve-per-cylinder engine with airflow meter G70, charge-pressure sender G31, boost-control solenoid N75 and direct injection. These are AXX example details, not a complete list of MED9 hardware. Identify the engine, vehicle, hardware and software before selecting definitions; the MED9.1 software matching guide explains what to record.

Save and verify the original read, establish the vehicle's fault status and mechanical condition, and log a repeatable baseline before changing calibration. Keep fuel type, intake and coolant temperatures, gear and ambient conditions with the run. Use suitable load and speed conditions for the vehicle and test environment.

Follow the request through the model

The following names and roles are taken from Aliant Auto's published MED9.1 SEAT Leon Cupra example. They are signposts for finding and checking functions, not a universal map list. The guide's title says “20V”; VW identifies its AXX 2.0 FSI Turbo example as four valves per cylinder. Verify vehicle details, axes, variants, units and actual control logic for the exact software.

Map labels in the exampleRole to investigateUseful check
KFPEDDriver torque request from pedal position and engine speed.More pedal sensitivity is not the same as more available torque.
KFMIRL and KFMIOPRequested torque to load and the related load-to-torque model.Check that the paired models, axes and limiters remain consistent over the used range.
LDRXNOne engine-load ceiling in the example.A request above a limit can be clipped; other limits and interventions may also apply.
KFLDIMX and KFLDRLBoost-control integral limit and duty linearisation.Compare requested and actual pressure and actuator behaviour; do not assume a controller setting is the cause of a leak or mechanical fault.
LAMFA and KFLBTSExample lambda request and component-protection enrichment functions.Identify which function and operating state are active when comparing target and measured lambda.
KFZW and KFZWOP familiesIgnition maps and an optimum-timing reference used by the model.Distinguish actual spark control from the model reference; verify timing and knock data in logs.

As a conceptual example, if a torque-to-load request were 160% and an applicable load ceiling were 140%, that ceiling could cap the request at 140% at that operating point. These are illustrative numbers, not recommended values; the real outcome depends on all active limits, model scaling and operating conditions.

Log and interpret the result

  1. Compare pedal and throttle, requested and actual load and boost, engine speed, airflow, intake temperature, and available torque intervention or limit flags. Confirm whether pressure is absolute or gauge and use consistent units.
  2. Log requested and actual lambda, high-pressure rail request and actual, low-pressure supply where available, ignition timing and per-cylinder knock correction or misfires. A repeatable pressure or mixture error needs diagnosis before further load is added. See MED9.1 fuel-pressure and logging checks.
  3. Change only verified functions in measured steps. Recheck part load as well as the area changed, and retain each version and its log. Validate required checksums with a tool supported for the exact ECU before writing. Keep thermal, knock and other protective strategies operational; a missing intervention in a short log is not proof of a safe calibration.

SSP 337 describes homogeneous operation for the AXX and a separate cold-start catalyst-heating strategy; record operating state when interpreting lambda. Sources describe particular implementations and have not been vehicle-tested by this wiki.

For intake-related changes in air mass, load and ignition correction, see MED9.1 intake comparison and fuel-trim diagnosis. The separate MAF signal and intake-correction reference identifies the sensor-conversion, correction and diagnostic functions.

MED9.1 boost control: integral limit and actuator duty

The Bosch MED9.1 function document for 41W038_PQ35, 20 August 2004, LDRPID 25.40.1, pages 931–939, shows KFLDIMX in the integral-limit path and KFLDRL after the controller sum. Consequently, an integral-limit percentage need not equal final N75 duty. This source describes that program's implementation, not every MED9 software revision.

prj's 7 March 2014 reply in “MED 9.1 basics” specifically corrects copying logged wastegate duty directly into KFLDIMX and explains the need to account for KFLDRL. Basano's follow-up recognises that proportional and derivative contributions also matter during spool-up. Treat settled duty, integral limit and transient response as separate quantities when interpreting a run.

  1. Verify both axes and scaling, not just the map shape. In the 8P0907115AQ S3 case, 30 March 2016, the author reported correcting an incorrectly assigned KFLDIMX axis address. That correction belongs to the examined file and supplies no universal offset.
  2. Compare time-aligned requested and measured pressure, rpm, N75 command, throttle, temperature and available intervention states. Establish whether the log and sensor still represent the pressure range being tested. Check the fitted boost-control hardware and faults before attributing a mismatch to controller calibration.
  3. If using settled actuator duty to investigate the controller domain, account for the matching KFLDRL lookup at that speed. An inverse lookup alone does not identify transient gains or the instantaneous integral contribution. Verify relevant controller states and corrections in the target implementation.

Pressure-range conversions need MED9-specific proof

The linked May 2024 development thread is explicitly scoped by elias to 1K8907115F_0040. It reports a pressure-variable conversion, while prj's reply also identifies load calculation as an unresolved part of the work. Neither post verifies a complete conversion across MED9.1 software versions.

The 2026 NefMoto “MED9 5120 hack” thread starts from an 8P0907115K 040 A2L and an ME7 comparison. On 26 July 2026, the author reported that the version-three changes started the engine but it immediately stalled with P1297. The reviewed continuation through 4 October 2026 does not establish a successful, validated conversion. It is research evidence, not a ready-to-use MED9.1 recipe; an ME7 scaling list or a replacement sensor alone does not prove the complete pressure path is correct.

For mixture requests, coding variants and flap-dependent enrichment, see MED9.1 lambda requests and map selection. Additional sources accessed 16 November 2026.

For a gearbox-related torque constraint, signal normalisation and the meaning of KFMIOPS, continue with MED9.1 torque reporting and DSG intervention.