Data Logging for VE Analysis

What makes a good log

VETuner's VE Analysis compares actual lambda to target lambda at each RPM/load cell and calculates the correction each cell needs. The quality of the result depends entirely on the quality of the data: noisy, sparse, or condition-contaminated logs produce poor corrections.

A good log has:

• Coverage: every cell you care about visited enough times to build statistical confidence
• Steady-state data: stable throttle and RPM so the sensor reading reflects the fuelling at the current operating point, not a transient
• Clean sensor readings: wideband warm, exhaust leak-free, no misfires
• Warm engine: coolant at operating temperature throughout

Pre-log checklist

Before starting a logging session:

1. Engine fully warm: at least 5 minutes of running, coolant above 80 deg C
2. Wideband sensor at operating temperature: allow 30-60 seconds after startup before trusting readings
3. No active fault codes that could affect fuelling (failed MAT sensor, coolant sensor out of range, etc.)
4. Fuel level adequate: running out of fuel mid-log contaminates data with lean transients
5. No recent cold start enrichment still bleeding off. If in doubt, drive gently for a few minutes before starting the analysis

Drive technique

Steady-state passes give the cleanest data. Pick a target RPM and load, hold them stable for 5-10 seconds, then move to the next operating point. On a public road, a dual carriageway at part throttle works well for the 2000-3500 RPM cruise region. A long gentle hill covers high-load low-RPM cells.

Avoid:

• Hard accelerations: transient enrichment (accel enrichment, open-loop AE) masks the underlying VE table fuelling
• Rapid throttle changes: the lambda sensor lags behind the fuelling event; reading it during a transient attributes the wrong correction to the wrong cell
• Cold operating conditions: warm-up enrichment and afterstart enrichment are active; any corrections made to the VE table while these are running will be wrong once the engine is warm

Map coverage

Divide the VE table into regions and work methodically:

High-load and WOT cells are the most safety-critical to get right. Visit these last, once you are confident the light-load and cruise cells are close to correct.

How many passes?

VETuner accumulates data statistically. More passes mean more confidence. As a rough guide:

• Light-load cruise cells: 3-5 steady passes per cell is usually enough
• High-load cells: aim for at least 3 clean passes per cell
• Idle: the engine spends a lot of time here naturally; idle cells typically fill quickly

The VE Analysis heat map shows cell confidence visually. Grey cells have insufficient data; coloured cells have enough to make a suggestion. Do not apply corrections to grey cells.

Sensor placement and exhaust leaks

The wideband sensor should be in the exhaust collector, after all cylinders have merged, far enough downstream to avoid scavenging effects from valve overlap. Sensors positioned too close to the head or to individual header primaries can read false-lean during scavenging events.

An exhaust leak upstream of the sensor dilutes exhaust gas with ambient air, making every reading appear lean. If your VE Analysis consistently suggests the whole map is lean when the engine seems to be running acceptably, check for exhaust leaks before adjusting anything.

Sessions and merging

Each VE Analysis session accumulates data independently. VETuner merges sessions statistically, so you can run multiple shorter logs rather than one long session. This is useful for:

• Building coverage gradually across different driving conditions
• Comparing before/after a hardware change
• Logging different temperature conditions separately if the engine behaves differently when very hot

After applying corrections from one session, reset and start a fresh session to validate the results rather than accumulating old corrections on top of new ones.

What to do with the results

Once the heat map shows reasonable coverage and the suggested corrections are small (within +/-5% on most cells), apply the changes to the VE table and take another validation log. The corrections should converge towards zero. If corrections remain large after two or three rounds, check:

• Wideband calibration: compare against a known-good sensor or calibration gas
• AFR target table: are the targets actually achievable and sensible?
• Injector dead time: systematic lean-at-idle / correct-at-load points to wrong dead time
• Fuel pressure: if pressure varies with load, injector effective flow rate changes

Large corrections that refuse to converge, or corrections that are large in one direction across the whole map, usually indicate a hardware or configuration issue rather than a VE table problem.