Audi S5 4.2 FSI (CAUA) Rough Idle — Diagnostic Workflow

The problem

"Rough idle" on an Audi S5 with the CAUA 4.2 FSI V8 is a felt vibration or unevenness at idle that may or may not be a logged misfire. Because the CAUA is a V8, a single-cylinder weakness can be subtle at idle and only become obvious under load — so the first job is to find out what kind of rough idle it is before chasing parts.

This page is the problem-specific path for a CAUA rough idle. It is narrower than the CAUA engine hub and deeper than the generic rough-idle symptom guide, which covers the symptom across all engines.

Vehicle-specific verified specification required for any adaptation or trim tolerance on this page.

Separate the kind of rough idle first

• Random roughness — no single cylinder dominant; a whole-engine issue (air leak, fuel pressure, MAF, mixture, timing).
• Cylinder-specific misfire — one cylinder's misfire counter high; that cylinder's coil, plug, injector, or compression.
• Mixture-related rough idle — fuel trims pushed to a limit (lean or rich) producing uneven combustion without a single failed part.
• Mechanical imbalance — one cylinder's compression or valve condition different from the rest; a vibration without a logged misfire.

Per-cylinder misfire counters at idle split the first two; fuel trims (STFT/LTFT) identify the third; relative compression identifies the fourth.

Differential

Build the differential across these categories and confirm each with its evidence before condemning a part:

• Ignition — coil, plug. A coil/plug swap between cylinders is the fast differential: if the misfire follows the part, the part is faulty.
• Injector / fuelling — injector balance and contribution; an individual injector can misfire at idle without a fuel-pressure code.
• Fuel pressure — requested vs actual rail pressure; a genuine low pressure at idle leans the mixture across all cylinders and looks random.
• Intake leakage — unmetered air after the MAF drives lean trims and random roughness; smoke or propane test around the intake.
• Air measurement — a MAF that over- or under-reads shifts the load calculation and the mixture; correlate MAF against load and throttle.
• Fuel trims — STFT/LTFT direction and magnitude: large positive trims = lean (air leak or fuel starvation); large negative = rich.
• Cylinder-specific mechanical condition — relative compression and cylinder leakage; a low cylinder identifies a mechanical cause.
• Compression — confirm mechanical condition only after ignition, fuel, and air are exonerated.
• Crank / cam correlation — timing/cam control as a hypothesis; both signals can look fine individually but correlate wrong; scope the pair.
• Sensor plausibility — a biased MAF, MAP, or temperature sensor shifts the mixture without a mechanical fault; check the 5V reference and sensor supply.

Teach the separation

• Random vs cylinder-specific — counters, first. Do not assume "coil" before reading which cylinder (or none) is dominant.
• Mixture vs mechanical — trims and relative compression, not vibration feel. A vibration can be a mount amplifying a normal idle; a rough idle can be mixture with no misfire logged.
• Air vs fuel — trims + a leak test split an air leak (lean) from fuel starvation (also lean, but with good trims response to enrichment).
• Sensor vs reality — correlate the sensor against a known-good reference; a biased sensor shifts the mixture without a real mixture fault.

Misleading assumptions

• "Carbon buildup is the cause." Carbon on intake valves is a real FSI concern, but it is a hypothesis requiring confirmation (trims, leak test, scope correlation), not a default answer to rough idle.
• "It's a coil." Only if the misfire follows a swap. A V8 with one weak coil is usually a single-cylinder counter, not random.
• "Bad injector." Only if balance/contribution isolates that cylinder; a random rough idle is not one injector.
• "PCV." A PCV fault can introduce unmetered air, but confirm with trim and leak evidence, not by assumption.

Do not claim carbon buildup, coils, injectors, or PCV as the cause without the evidence that differentiates them.

Scan and live-data interpretation

With VCDS (Ross-Tech), capture before clearing anything:

• Per-cylinder misfire counters at idle — splits single-cylinder from random.
• STFT / LTFT at idle — direction and magnitude reveal lean or rich.
• MAF / load / throttle consistency — confirms air measurement plausibility.
• Requested vs actual rail pressure at idle — rules out fuel pressure as the mixture cause.
• Freeze-frame for P0300 / P0171 / P0172 — the condition that set the code.

Do not publish proprietary or universal channel/group numbers.

Relevant tests

• Coil/plug swap between cylinders — isolates ignition.
• Smoke or propane intake test — finds unmetered air.
• Injector balance / contribution — isolates a fuelling cylinder.
• Relative compression / cylinder leakage — isolates a mechanical cylinder.
• Crank / cam correlation scope — confirms timing/cam control.
• 5V reference / sensor supply check — rules out a biased sensor.

Next-best-test logic

1. Read per-cylinder misfire counters — split single-cylinder from random.
2. Read fuel trims at idle — split mixture from mechanical.
3. If single-cylinder: swap test first, then injector balance, then compression.
4. If random: leak test and MAF plausibility, then fuel pressure, then crank/cam correlation.
5. Confirm a mechanical cause only after ignition, fuel, and air are exonerated.

Confirmation criteria

Confirm the cause only when the evidence isolates one category: the misfire pattern identified, trims interpreted, air and fuel exonerated or confirmed, and the rough idle reproducibly explained by one element.

Repair verification

After repair, clear fault memory, and idle the engine under the condition that failed (cold, hot, in gear, A/C on). The repair is verified when idle is smooth under the reproducing condition, misfire counters stay at zero, and trims return to a normal range.

CTA

Already have scan data or live values from this engine? MechanIQ can help organise the evidence, rank the diagnostic hypotheses, and identify the next best test.

Trademark disclaimer

MechanIQ is independent software. Audi, VAG, CAUA, FSI and ODIS are trademarks of Volkswagen AG. VCDS is a trademark of Ross-Tech, LLC. MechanIQ is not affiliated with or endorsed by any of these companies. All specifications are guidance only; vehicle-specific verified specification required.