Tuning Wiki
Applies to

Check the hardware and software limits stated in this article.

What changes when an axis changes?

A calibration table relates an input operating point to a stored output. The ECU uses the input axes and the table values together; where and how it interpolates must be established for the specific function. Changing a breakpoint can therefore change the output at an existing operating point even when no data cell is edited.

Illustration, not a tune: suppose a linearly interpolated ignition row has relative-load breakpoints 60%, 100% and 140%, with outputs 15°, 12° and 9°. At 120% load, interpolation gives 10.5°. If the breakpoints alone become 60, 120 and 180 while the three outputs remain unchanged, the output at 120% load becomes 12°. The same numbers now describe different operating points.

Reported definition error. In a forum-reported 1.8T throttle-response case, the XDF for KFVPDKSD/E was found to have swapped axes. The definition issue was later reported resolved, while some pressure-ratio interpretation remained open. This is a reported case, not proof that every similar symptom has the same cause. Check each axis quantity, orientation and units against the matching software and definition before editing the cells.

A forum-reported 1.8T ECU conversion found a 12-by-16 KFMIRL in one image and 16-by-16 in the other. These are file-specific claims, not a rule for every narrowband or wideband ECU. Different dimensions require checking the input axes, scale, shared storage and interpolation before any transfer; an identically named map is not automatically a compatible matrix.

ME7 map relationships

Function in the cited VAG exampleReview question
KFMIOP: relative charge/load to modelled relative torqueWhat load and RPM axes does the definition actually use, and which other functions use those axes?
KFMIRL: torque request to requested charge/loadAre its input and output units, operating range and relationship to the torque model understood?
KFZWOP / KFZWOP2: optimum-timing referencesHas a shared axis change also changed the torque/ignition-efficiency model?
KFZW / KFZW2: base ignition paths where fittedWhich table and cam state are active, and how are corrections and interventions added?

S4Wiki’s documented M-box uses shared load and speed axes for KFMIOP, KFZWOP, KFZWOP2 and KFMDS. That is a fact about the cited image, not proof that every similarly named table in another ECU shares the same storage. A definition can also show equal-looking axes that are stored separately. Check addresses, dimensions and conversions before assuming either arrangement.

The ME7 map names and definition checks compares KFMIOP, LDRXN, KFLDIMX and KFWDKMSN with the Bosch Audi ME7.1 function description. It also identifies conflicting descriptions in the Tuning Database overview, so a reader can check function and units before relying on an editor's label.

About inverse maps and extrapolation

KFMIOP and KFMIRL describe related conversions in opposite directions. MxT offers a map inversion workflow, but the actual request path also contains limits, losses and efficiency terms. “Inverse” here is a relationship between calibrated lookups, not a guarantee that a mathematical matrix inverse is appropriate or that every point in the full ECU calculation returns exactly to its starting value. Check the relevant function, interpolation, range and monotonicity for the file.

Values generated beyond the measured or characterised axis range are estimates. Review the generated cells and the effect on neighbouring, previously covered operating points; compare modelled torque, base timing, actual ignition, knock and interventions under relevant conditions. Extending an axis does not supply evidence for safe ignition or available airflow at the new edge.

Reported torque-display change. In a forum-reported ME7 case, diagnostic block 120 fell from about 300 to 270 Nm after KFMIOP/KFMIRL and axis edits, while actual torque was believed unchanged. No independent torque measurement establishes the true output. A reply in that discussion describes a calculation involving modelled torque, losses and normalisation, and points out a shared axis. Treat the displayed number as a software-derived diagnostic quantity, check the active model and shared axes, and use an appropriate independent comparison if actual output is the question. The exact calculation and diagnostic presentation require verification for the software in hand.

The high-load representation and map-tracing guide separates runtime signal width from axis storage, editor display and downstream limits. Its NefMoto examples include an explicitly untested patch proposal and a different ME7.1 application reported to have a wider axis.

Sources and attribution

Adapted and expanded from S4Wiki contributors’ Tuning, revision 5324, Creative Commons Attribution 4.0, and TracqiTechnology/MxT contributors’ ME7 Calibration Guide, Apache License 2.0, full text. The numeric axis example on this page is invented for illustration. See ME7.1 S4 load, torque and ignition examples, ME7 ignition and knock reference, and ME7 getting started for definition checks.

Continue the conversation

If this topic interests you, consider joining the following forums. They host material referenced in this article.