Start by identifying the temperature
These notes concern the cited VAG ME7.1/ME7.5 examples. First identify the engine, software, exhaust hardware and which exhaust-temperature sensors are actually fitted and active. A logged value with a temperature label is not necessarily a physical sensor reading. On a different arrangement, the Audi 2.7T S4 may have separate sender and model paths; see S4 exhaust-temperature and enrichment.
The Bosch Audi 132 kW ME7.1 function description identifies modelled temperatures in ATM and a component-protection lambda function LAMBTS. Names and source selection below are examples from that description, not a promise that every ECU exposes the same variables.
| Example | Meaning in the cited function | Check |
|---|---|---|
tabgm_w | Modelled exhaust temperature before the catalyst | Location and model; it is not inherently a sensor. |
tabgkrm_w | Modelled exhaust-manifold temperature | It represents a different point from the pre-catalyst model. |
tabgbts_w | Temperature used within the component-protection calculation | Trace the selected source and operating configuration in the exact software. |
TABGBTS | Calibration threshold for exhaust-temperature component protection | Do not confuse the fixed threshold with the similarly named runtime value. |
B_tabgbts, lambts_w | One threshold condition and a component-protection lambda request | A condition, the request and final commanded lambda are separate observations. |
A model has thermal history
In the cited Bosch ATM example, a speed/load map supplies a steady-state basis; corrections include lambda and ignition efficiency, while filters model exhaust and catalyst thermal response. The starting state and preceding operation also affect the calculated temperatures. Therefore a short pull and a long loaded run at similar RPM do not necessarily produce the same modelled temperature. Other software may implement different inputs and corrections.
A forum-reported modified 1.8T AMB case raised the question of whether a model calibrated for the original exhaust would remain representative after hardware or fuel changes. No measured correction for those changes was provided. A model can be useful without being a direct measurement, and its absolute error for a changed setup must be established rather than assumed.
A threshold is not the final lambda target
In a forum-reported AUQ 1.8T case, a substantial difference was seen between two calculated temperature channels, and it was unclear which fed protection. The channels were identified as modelled exhaust temperatures at different locations, and source selection was discussed; the model was not validated against a matching independent measurement and no safe replacement threshold was established. Later replies also show that the enable condition and the resulting enrichment need separate interpretation.
In a separate forum-reported ME7.5 case, a protection flag was observed without the expected change in final lambda. Competing requests, factors, filtering and definition issues were raised, and the precise cause remained disputed. Log the relevant temperature input, threshold conditions, lambts_w, driver or other requests, final requested lambda and measured lambda at the same time. A threshold flag alone does not prove that KFLBTS directly determines the final request.
Keep the protective strategy functioning while investigating. A richer request can be an intended response to thermal conditions; delaying or suppressing it merely to improve a graph can remove protection. The model and lambda request also depend on the rest of the calibration and hardware, including air measurement, ignition, fuel supply and load.
Compare with an appropriate measurement
- Confirm the actual probe or sensor, calibration, measuring range, channel and location. A manifold or turbine-inlet probe and a pre-catalyst temperature are different measurements; the turbine changes the exhaust-gas temperature. Garrett describes monitoring EGT in the manifold or turbine housing.
- Compare matched locations and operating histories where practicable. Include RPM, load, lambda request and measurement, ignition, air and coolant temperatures, pressure, duration and previous heat soak. Allow for sensor response and logger alignment.
- Investigate disagreements over more than one operating condition. A single probe trace at a different location cannot independently establish a complete model, sensor failure or safe universal limit. Record component-specific limits and follow the equipment and vehicle documentation.
Related: ME7 fueling and MAF calibration, lambda control state during boost and ignition and knock reference. Forum observations are attributed reports, not tests independently reproduced by this wiki.