P0171 with rising long-term trims: vacuum leak, MAF or fuel delivery

Reading the trims

The ECU is adding fuel, so it sees more air than it has measured (or less fuel than it expected). The question is how the correction behaves with airflow.

• Idle: LTFT +24 %
• Cruise, 38 % load, 2 480 rpm: LTFT +21 %, STFT +18 %

The correction is proportionally worse at low airflow. That is the signature of unmetered air: a fixed leak mass matters most when the engine is only drawing a few grams per second at idle, and becomes a smaller share of the total as airflow rises.

• A MAF under-reading by a percentage would give roughly the same correction at idle and at cruise.
• A weak pump or restricted filter would be worst at high load and high fuel demand, the opposite of what we see. Fuel pressure of 3.8 bar is in spec anyway, though it should be re-checked under load.

The propane test confirms it: adding unmetered fuel at the manifold cuts trims by 8 % at idle but only 2 % at 2 500 rpm, so the excess air is concentrated at idle. MAP 34 kPa at idle against 101 kPa baro is normal, and 2.4 g/s MAF is inside spec. The idle whistle is the leak itself.

Verdict: a vacuum leak, not the MAF and not fuel delivery.

1. Smoke test into the intake

Feed smoke through the intake with the throttle closed. Look at the intake boot after the MAF, the manifold gasket line, injector seals, PCV hoses, brake booster check valve and hose, and the EVAP purge line and canister. Visible smoke is proof. Start here because it is safe, fast and finds small leaks the other methods miss.

2. Isolate the EVAP purge line

Clamp or cap the purge hose at the manifold and watch trims at idle. A stuck-open purge valve is a common unmetered-air path and the trims will fall by several percent within a minute. This is the single most commonly missed leak on this type of fault.

3. Pinch or plug test, then propane at specific joints

Pinch the brake booster hose and the PCV hose one at a time and watch LTFT and O2 response. Propane or a small amount of carb cleaner at suspect joints, watching trims and O2, localises the leak. Trade-off: aerosols near a hot exhaust are a fire risk, so use the minimum and keep a extinguisher at hand. A smoke test is safer and should have found it first.

Ruling it out: if smoke is clean, the purge is capped and trims stay at +20 % or more at idle, stop treating it as a leak and move to a MAF comparison (scope the signal against throttle snap) and an injector balance or flow test.

Why it takes 40 km

The code is not set at the first lean excursion. The ECU runs a two-trip monitor: it needs closed-loop operation, coolant up to temperature, a minimum run time and the fault present across a drive cycle before it commands the MIL. Long-term trim also has to adapt - the ECU learns in steps, and a small leak can take tens of kilometres of mixed driving before LTFT drifts past the threshold (commonly around +20–25 %).

The 40 km is therefore a measure of how long the adaption took and how the car was driven, not of how long the fault has existed. The leak may have been there for months.

What the freeze frame says

• 2 480 rpm, 62 km/h, 38 % load: the decision was made at a steady cruise, not at idle. That is typical - the monitor runs best under steady, moderate load with stable airflow.
• STFT +18 % and LTFT +21 % at the moment of setting: both are pushing fuel in, so the fault was active at confirmation, not a stored historical value.
• Coolant 89 °C: fully warm, closed loop, so the trims are meaningful.

Use the freeze frame to reproduce the condition on a test drive. If trims go leaner still when you recreate 2 500 rpm at light load, you are on the right track; if they normalise, look at the leak's behaviour with heat and vibration.