TL;DR
- Tuning is changing targets and limits so the ECU can control torque safely under your real conditions.
- Modern ECUs are torque-based: they decide airflow, fueling, and spark to hit a torque request.
- Validation beats screenshots: consistent conditions and repeat runs are the only way to trust changes.
A mental model (text diagram)
Driver intent → Torque request → ECU torque model → Airflow (boost/throttle) + Fuel + Spark
↑ ↓
Sensors (air, temps, knock, fuel) ← Outcome (torque delivered)What makes a pull “clean”
- Stable temperatures (or at least recorded and similar)
- Same road and direction (grade and wind)
- Same gear and RPM range
- No traction events (no spin, no shifts mid-measurement)
What makes a pull “dirty”
- Heat soak differences
- Grade/wind differences
- Traction changes
- Driver input differences
Validation plan for this topic
Start with one written hypothesis from this page: A “dirty” pull adds uncontrolled variables; a clean pull controls the environment so changes reflect the car - not the conditions. Do not begin the test until the vehicle is mechanically healthy, the fuel is known, and the operator can use a legal controlled environment. Keep gear, engine-speed window, load request, and starting temperature as consistent as the platform allows.
For Clean pull vs dirty pull: why comparisons fail, align these signals on the same time axis:
- starting temperatures - establishes what the system was asked to do or the condition entering the event.
- throttle continuity - shows the primary response rather than a dashboard summary.
- fuel pressure and lambda - provides the safety or control context that can explain an apparently good or bad result.
- correction and intervention events - tests whether another system, not the headline variable, actually set the limit.
Mark the start and end of the relevant event before interpreting it. Look for sequence: request, response, deviation, intervention, and recovery. A value that changes after throttle lift cannot explain the event that happened before it. Compare at least two clean repetitions and retain the original file, calibration identifier, fuel, ambient condition, and any warning or driver note.
What counts as a useful result
A useful result either supports the hypothesis under comparable conditions or disproves it clearly enough to change the next action. It is also useful to learn that maintenance, sensor quality, traction, or temperature made the test invalid. Do not average invalid and valid attempts together, and do not hide a protection event because the final headline number improved.
Stop and investigate
Reduce load and investigate persistent correction, misfire, unexpected lean operation, falling fuel pressure, uncontrolled temperature rise, fluid leakage, new mechanical noise, or an intervention that was not present in the baseline. A generic internet threshold is not a substitute for the platform’s factory documentation, the calibration provider’s limits, and review by a qualified professional.
For the full control-of-variables method, use Repeatability and consistency. For signal ordering and context, use How to read a datalog.
Common mistakes
- Using a single “hero run” as proof.
- Changing multiple mods and doing one pull.
Diagnostics / what logs tell you (high level)
| Signal | What it usually means | What to check |
|---|---|---|
| RPM | Context for everything else | Compare in the same gear and load range |
| Throttle angle / torque intervention | How the ECU is controlling torque | Look for closures that explain boost drop |
| Boost target vs actual | Control quality and limits | Oscillation can be hardware or control strategy |
| IAT / temps | Margin and repeatability | Heat soak changes results dramatically |
| Knock / timing correction | Combustion safety response | Sustained corrections = reduce load/verify fuel/temps |
FAQ
How many runs is enough?
More than one. Look for repeatability under similar conditions, not the best number.
Related guides
- Hub: Tuning hub
- More in this pillar: - /academy/tuning/repeatability-and-consistency/
- Related (other pillars): - /academy/mods/tires-for-performance-driving/
- Reference: Glossary
- Brands: Brand pages