How to datalog your car for tuning

A tune built without a log is a guess. A datalog is your car telling you what it actually does at full throttle — how much boost it really makes, whether it's pulling timing, and how close the injectors are to running out. Here's how to record one properly, and how to read it.

Updated 18 September 2026 · 10 min read

Why a log matters more than the map

Two identical cars can respond very differently to the same file. One has a clogged MAF, another has tired injectors, a third is heat-soaking on its third pull. None of that shows up in the map — it only shows up in the data. Logging before and after a tune turns tuning from guesswork into a measurement.

What to log — the channels that matter

You don't need fifty channels. You need the handful that decide whether a tune is safe:

• RPM & throttle position — proves it was a genuine full-throttle pull.
• Boost: requested vs actual — the single most useful pair. A gap between them means underboost, overboost or a boost leak.
• Ignition timing & knock retard (per cylinder) — the safety channel. Timing being pulled is the engine protecting itself.
• Lambda / AFR — going lean at high rpm under load is the classic way engines get hurt.
• Injector duty cycle — near 100% means you are out of fuel system, not out of tune.
• Intake air temp & coolant — tells you whether the pull was even valid.

Where the numbers come from

A VAG ECU exposes its live data as numbered measuring blocks (VCDS calls them groups), each a set of four values. A datalogger asks for the blocks you choose, as fast as the bus allows, and writes each reply with a timestamp. That "as fast as the bus allows" is the catch:

• K-line cars (ME7, EDC15, early EDC16 — everything before about 2008) answer a few times a second. Log four blocks and each one updates about once a second. A 3rd-gear pull lasts eight to ten seconds, so you get eight to ten points per channel. Enough to see a trend; not enough to see a single knock event. Log fewer blocks for more points.
• CAN cars (MED9, MED17, EDC17, Simos) answer many times faster, and the tool can ask for individual values rather than whole blocks, so you get a proper trace.

On an ME7 car the blocks worth logging are 115 (boost requested vs actual), 020 (knock retard per cylinder), 031 (lambda requested vs actual), 003 (rpm, load, air mass, timing) and 032 (fuel trims). Pick three per pull on K-line and rotate them across pulls rather than logging all five slowly.

Diesel: different channels, same idea

A TDI has no knock and no lambda target to watch. What decides whether a diesel tune is safe is:

• Boost requested vs actual — the same gap logic as petrol; a VNT turbo that overshoots and then falls away is the classic sign of a sticking actuator.
• Injection quantity (mg/stroke) vs what the map asks for — if the ECU is limiting, one of its limiters (torque, smoke, EGT model, rail pressure) has decided the request is too much.
• MAF — a tired MAF reads low, the ECU sees "not enough air" and the smoke limiter cuts fuel; the car feels flat and the tune gets the blame.
• Rail pressure requested vs actual on common-rail — a pump that can't hold pressure at high rpm is out of fuel system, like a petrol injector at 100%.
• EGT, if the car has a sensor or you fit one — the diesel equivalent of knock. Sustained high exhaust temperatures are what melt pistons and turbines on an over-fuelled TDI.

How to do a proper pull

1. Warm the car up. Drive gently for 10–15 minutes. Coolant at normal temperature, oil around 80–90 °C. A cold pull is wasted data and hard on the engine.
2. Somewhere safe and legal. A track or dyno is best. On the road: flat, straight, empty, dry, with clear visibility, and enough room to slow down again.
3. Use 3rd gear (or 4th on a 7-speed DSG). Never 1st or 2nd — they're over in a second and just spin the tyres. On DSG/auto use Manual/Tiptronic and don't press through the kickdown click.
4. Settle, then floor it. Let it settle to about 1500–2000 rpm, go to 100% throttle in one smooth movement, and hold it flat to redline — no lifting, no feathering.
5. Do 2–3 clean pulls, letting intake temps recover between them. Discard any pull with wheelspin, part throttle or a power cut.

How to read what you recorded

What "healthy" looks like on a turbo petrol

Rules of thumb for a 3rd-gear pull on a warm engine, stock or Stage 1. Every engine differs; these are the ranges where nobody worries.

The full loop: log → tune → log again

The workflow that produces a safe, genuinely faster car is a loop, not a one-shot:

1. Baseline log on the stock file, so you know your starting point.
2. Read your ECU out and keep the original — see how to back up your ECU.
3. Build the tune around what the log showed, not a generic file — see what each stage actually changes.
4. Flash it, then log the same pull again.
5. Compare. More power with clean knock and safe lambda means there may be more on the table; lean or knocking means stop and fix.

Common mistakes

• Logging a cold car — the numbers mean nothing.
• Pulling in too low a gear, so the pull is over before it's readable.
• Lifting mid-pull, then reading the result as if it were full throttle.
• Back-to-back pulls with no cool-down, then blaming the tune for heat-soak.
• Chasing a peak number instead of watching knock and lambda across the whole range.

Common questions