A sensor and a control state are different things
This guide concerns VAG ME7.1 and ME7.5 examples. Identify the ECU software, fitted sensor type, bank and active calibration. A wideband sensor can measure richer mixtures over a specified range, but its presence does not prove that feedback control is continuously active at every load, temperature and operating state. A narrowband sensor has different useful measurement limits.
A forum reply describes wideband control as always active. In a separate ME7.5 discussion, contributors describe feedback stopping under certain conditions and disagree about the relationship with component-protection enrichment. One cited log had lambts_w different from 1 while the correction was changing, but its author had not logged the B_lambts status. Those statements do not establish one universal rule or prove the two reported conditions were identical. Log the actual conditions and consult the matched function description.
Read the signals together
| Example in the cited ME7Logger alias list | Question to ask |
|---|---|
B_lr / B_lr2 | Pre-catalyst lambda-regulation condition for bank 1 / bank 2. Is the relevant condition present, and is the flag correctly defined for this software? |
fr_w / fr2_w | Lambda-regulator output for each bank. Is it changing, neutral or held? Check scaling and sign; a value at its neutral point is not by itself proof that regulation is disabled. |
fra_w, additive adaptations | Learned mixture corrections are distinct from the immediate regulator output. Record them and their learning state rather than treating one as the other. |
| Final requested and measured lambda | Is the sensor ready and within its validated range, and is delivered mixture following the final request? Match bank, units and time delay. |
lambts_w, B_lambts, where defined | Inspect the protection request and its status as separate signals; a request value alone is not a substitute for the status flag. |
These labels are examples from the ME7Logger alias list, not guaranteed addresses or logic for another file. Where fr_w is a factor with 1.0 as neutral, the forum trim discussion illustrates percentage conversion as (fr − 1) × 100: 0.95 corresponds to −5% in that representation. Check the actual channel before comparing it with a diagnostic tool's percentage.
When control drops out
Include the time before and after the transition. Record load, RPM, pedal, final lambda request, sensor reading, regulation condition, correction, adaptations, coolant and intake temperature, fault codes and relevant protection status. Check sensor readiness, range and fault state. A display at its lower limit can be a range or scaling problem; verify the actual sensor and controller rather than assuming the mixture equals the displayed bound.
In the 018CH case, the owner reported resolving a modelled-manifold-pressure issue after undoing an unrelated airflow-limit change, while the lambda-control question remained. In another 1.8T case, correcting an over-scaled MAF was reported to improve the feedback duration, but the contributor still observed control stopping later. Neither case proves a universal cause.
The M-box narrowband discussion describes stoichiometric closed-loop operation in its 2.7T example. Use suitable independent wideband measurement for rich mixtures when the fitted feedback cannot measure them. Do not suppress component protection to make a feedback graph look continuous; investigate its actual request and conditions.
Related: MAF and lambda log review, S4 exhaust temperature and enrichment and mixture tuning: AFR and lambda. Forum observations have not been independently vehicle-tested by this wiki.