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
Simos 18
Applies to
Simos 18.1
Tool groups
None
Manufacturers
Volkswagen Group

Correction versus learned adaptation

Fuel trim is feedback about corrections made to the fuel calculation. Short-term correction responds to current lambda feedback; long-term adaptation records learned corrections under the conditions permitted by the ECU. The ECU may have several operating regions or injection paths, so a single displayed LTFT value does not necessarily describe every cell. The lambda-control guide covers targets and controller operation separately.

For conventional percentage displays, positive trim generally indicates fuel added and negative trim fuel removed. Always verify the actual tool, channel, units and sign: a factor and a percentage are different representations. COBB's MQB monitor list describes its LTFT as stored correction and gives 1.0 as neutral and 1.1 as 10% added. Its negative example is printed inconsistently; check the installed logger definition rather than assuming all tools use the same scale. Do not blindly add STFT and LTFT without knowing how the channel is constructed and applied.

Factory-related adaptation maps

The following are examples in the community SC8S5031_C_OEM Simos 18.1 LACO adaptation reference. They are adaptation calibration limits, weights or conditions, not the learned STFT/LTFT readings. Verify ASW, CAL, A2L, active combustion mode, array selection and scaling. A factor may be applied or interpolated differently by another version.

  • id_mff_sp_bol_add_rng_lam_ad[1] and id_mff_sp_tol_add_rng_lam_ad[1] define lower and upper fuel-mass-setpoint bounds for offset adaptation versus rpm; their outputs are mg/stroke.
  • ip_fac_mpl_lam_ad[3] weights the multipoint multiplicative adaptation against rpm and a fuel-mass-setpoint axis. The latter is labelled NO_INPUT_QUANTITY in the reference metadata, so confirm the real lookup and array selector in the matching software.
  • ip_fac_n_fac_lam_ad weights the multiplicative adaptation versus rpm. ip_fac_n_mff_add_lam_ad weights additive adaptation against rpm and a fuel-mass-related axis whose input is likewise not identified in the axis metadata. The weights are not the learned correction values.
  • ip_fac_wup_fac_lam_ad and ip_fac_wup_mff_add_lam_ad weight multiplicative and additive adaptation during warm-up. Both use coolant temperature for one axis; their second axes are respectively a percentage and an additive fuel-mass offset in mg/stroke, with inputs marked NO_INPUT_QUANTITY. Confirm their actual runtime sources. The warm-up adaptation reference also lists ip_fac_wup_fac_lam_ad_wup as a copy used by the warm-up path and says its values must match ip_fac_wup_fac_lam_ad in this example. That is an image-specific consistency instruction, not proof every version has the same maps. Do not assume warm learning and cold operation share identical conditions.
  • c_fac_min_mpl_lam_ad and c_fac_max_mpl_lam_ad bound multiplicative adaptation; c_mff_min_add_rng_lam_ad and c_mff_max_add_rng_lam_ad bound additive mass offset in the stated units. c_tco_min_lam_ad is a normal-learning coolant threshold. These bounds and gates are not desired trim values or a substitute for correcting an underlying fault.

Downstream dynamic trim

The same software reference also uses the term dynamic fuel trim for a downstream-sensor controller. ip_vls_sp_lam_adj[0] is its downstream voltage setpoint in volts versus cylinder air mass maf_hb and engine rpm; ip_lamb_delta_p_lam_adj and ip_lamb_delta_i_lam_adj are proportional and integral correction curves with voltage-related axes marked NO_INPUT_QUANTITY. This downstream controller is not automatically the scan tool's generic STFT, and its voltage setpoint is not the upstream commanded lambda. Preserve catalyst and sensor monitoring.

Signals and active conditions

In the same LACO reference, mff_add_lam_ad_ipt[0] is interpolated additive adaptation in mg/stroke, with mff_add_lam_ad_ipt_hpdi[0] and mff_add_lam_ad_ipt_mpi[0] identifying injection-path examples. fac_lam_ad_hpdi[0] and fac_lam_ad_mpi[0] are adaptation factors. lv_lam_ad_cdn_add[0], lv_lam_ad_cdn_fac[0] and lv_lam_ad_stop[0] indicate learning conditions or a stopped adaptation cycle. They are runtime signals, not maps; confirm availability and active mode.

An additive correction in mg/stroke is not directly comparable to a multiplicative percentage, and it must not be converted by simple addition without knowing the ECU's calculation. A neutral reading may mean learning is inactive or was reset, rather than that fuelling is correct. Sensor state, temperature, transients, fuel cut, purge and fuel-quality conditions can affect whether feedback or learning is valid.

Practical diagnosis

  1. Record the vehicle, software, fuel and blend, existing codes, active DI/MPI mode and the trim channel's units before changing or resetting anything. Establish steady warm idle and light-load observations as well as relevant transition and load logs.
  2. Compare measured versus target lambda, short-term correction, learned adaptation and any readiness/stop flags with air charge, fuel-pressure target and actual, injector pulse widths, temperature and purge state where available. Look for a repeatable condition-dependent pattern, not a single instantaneous number.
  3. Investigate intake/PCV leaks, exhaust leaks ahead of the sensor, injector scaling and operation, fuel delivery and pressure, sensor condition, fuel composition and air-charge modelling as appropriate. Idle-only and load-dependent patterns can suggest different causes but do not uniquely identify one.
  4. Make a focused correction, then repeat matched observations and allow the relevant adaptation conditions to be met. Some flashing procedures reset learned values; COBB's MQB tuning guide discusses this for its workflow. A reset can hide the original evidence and is not itself a repair.

Do not infer that every rich/full-load event is open loop or that every logged correction remains active during fuel cut or transients. Compare the actual controller state, target and pressure at the operating point. Closed-loop correction can conceal a base fuelling error until its available authority is exhausted.

Vapour and ventilation guides: EVAP purge · crankcase ventilation.

Worked examples

Rich full-load lambda with continuing correction (supplier-reported road logs)

CaracalTech's project-car report covers a 2017 Golf 2.0 TSI with Simos 18.1, ECU part number 5G0906259L and software SC800S5000000. It compares reported road logs on the original file and on the supplier's Stage 2 version V3. On the Stage 2 pull the lambda target was 0.83–0.85 and measured lambda 0.80–0.86 under steady load, with short-term correction ranging from zero to +13.3% over the logged pull; correction continued at the rich target. Long-term trim was 0.0% in every sample on both files, and rail pressure was close to target above 5,500 rpm.

This is a supplier-reported example, not raw data independently reanalysed here, and not a universal target. Ambient and coolant temperatures differed between runs, and the cause of the zero long-term reading and the reason for the correction were not established. A displayed zero does not prove that learning was reset or that no adaptation is applied elsewhere. Verify the logger's definitions and controller state, repeat comparable observations after the applicable learning conditions, and investigate air, fuel, pressure and sensor paths before changing a map. ECU-calculated torque in the report is not a measured dyno gain.

Lean trim after refuelling (owner report)

A 2019 Mk7.5 GTI with an APR Stage 1 93-octane tune was reported to show P0171, rough idle and a displayed trim of approximately 20–30%, followed by severe hesitation and a stall as boost began. Symptoms began immediately after refuelling. A smoke test revealed no leak, the HPFP had recently been replaced and a low-side pressure test appeared normal. The type and conditions of the trim and pressure readings were not fully documented, ECU software was not identified, and the cause remained unresolved in the accessible report. Refuelling timing, a negative smoke test or a prior pump replacement alone does not establish or exclude the cause. The useful practice is to preserve the actual trim channel and conditions, verify pressure at the relevant operating point, and investigate the fuel, air, purge and sensor paths using the applicable diagnostic procedure before revising the calibration.

Negative trim under load (owner report)

A 2015 Mk7 GTI with IE APEX Stage 2 and an upgraded HPFP was reported to show an STFT reading around −19 at full throttle, approximately zero at idle, and a P0131 code. Fuel pressure and lambda were reported to be consistent with or to follow requested values; the readings came from an IE POWERlink log. The logger scaling, active feedback state, exact ECU software and cause were not independently verified. A reported target-tracking lambda channel does not establish how much correction was active, and either negative trim or P0131 alone does not identify a failed sensor or fuel component. Compare the channel definitions and feedback state with matched pressure, lambda, fuel and air-system evidence before making a calibration change.

Related: lambda control · MPI injector calibration · direct injection and fuel pressure · cold start · acceleration enrichment. References reviewed 17 November 2026.

For cylinder-balancing adaptation, roughness-based correction and diagnostic limitations, see Simos 18 cylinder balancing and roughness adaptation.