How to Read Your First Haltech NSP Data Log

Write the run context before opening charts

A data log without context is just a pile of traces. The first note should explain what the car was doing, what changed, and what decision this log is supposed to support. If the note is missing, reconstruct it from the file name, vehicle, session, tune revision, weather, fuel, and driver comments before drawing conclusions.

• Vehicle, engine, ECU family, and firmware/software generation. Elite ECUs running NSP are not the same workflow as older Elite ESP files, and Nexus vehicles may include more integrated power-management and logging context.
• Fuel type and target strategy: pump gas, E85, race fuel, boost-by-gear, closed-loop lambda, flex fuel, or conservative break-in map.
• Run type: first start, idle warmup, steady-state cruise, ramp-rate dyno pull, street sweep, autocross run, or track session.
• Mechanical changes since the last clean baseline: injectors, regulator, fuel pump, turbo, wastegate spring, intercooler, sensor wiring, ignition components, or exhaust changes.
• The one decision you are trying to make from this log: safe to continue, needs fuel correction, boost control is unstable, heat management is poor, or the log is inconclusive.

Check the logging setup itself

Haltech documents datalog setup for Elite ECUs in its Datalog Elite support article, including channel selection and sample rate. That matters because a missing target channel or a slow sample rate can make a log look cleaner than it is. Before you read the story, confirm the camera was pointed at the right thing.

• Confirm the log includes both actual and target channels when available: measured lambda plus target lambda, MAP plus boost target, fuel correction plus base fuel state, ignition timing plus knock or timing trim.
• Confirm the sample rate is appropriate. A slow rate may be fine for coolant temperature but poor for boost oscillation, throttle transients, gear shifts, or short knock events.
• Check whether the log captures the whole event. A dyno pull without the ramp-in, or a track session missing the out-lap and cooldown behavior, can hide heat soak and recovery problems.
• Look for obvious sensor dropouts, flatlines, impossible values, unit mismatches, or channels that stay zero because the input was not configured.

Do not tune from a dirty pull

A dirty pull is one where the data does not isolate the condition you care about: throttle lift, wheelspin, clutch slip, unstable fuel pressure, a missed shift, protection strategy intervention, or a driver abort. Keep the file, label it, and learn from it, but do not make a confident fuel or ignition change from a pull that never represented the intended operating point.

The minimum useful view

For most first-pass reviews, start with RPM, throttle position, manifold pressure or load, lambda, target lambda, closed-loop fuel correction, ignition timing, coolant temperature, intake air temperature, battery voltage, and any logged oil/fuel pressure channels. Haltech has separate support material for Target Lambda and O2 Control; those concepts are more useful when actual, target, and correction are visible together.

• RPM and throttle position tell you whether the driver actually requested the operating point you are reviewing.
• MAP/load and boost target show whether the engine reached the intended load or if boost control, wastegate spring pressure, traction, or throttle closure changed the test.
• Lambda, target lambda, and fuel correction show whether the commanded mixture and measured mixture agree, and whether closed-loop control is hiding a base-table error.
• Ignition timing, knock activity, and any timing trims show whether the ECU is running the timing you think it is running.
• Coolant, intake air, oil pressure, fuel pressure, and battery voltage tell you whether the operating environment invalidates the performance conclusion.

Use channel groups instead of chart soup

A clean first layout usually has four panes: driver demand, air/load, fuel/lambda, and safety/temperature. That lets you read from cause to effect. Throttle and RPM create demand; MAP and airflow respond; lambda and fuel correction show mixture control; temperature, pressure, voltage, knock, and protection channels tell you whether the result was safe enough to repeat.

Add advanced channels only when they answer the next question

Once the first view finds a symptom, add the explanatory channels. For boost instability, bring in wastegate duty, boost control mode, boost target, gear, and throttle. For lean drift, add fuel pressure, injector duty cycle, fuel temperature if available, battery voltage, and any flex-fuel or fuel composition channel. For ignition instability, add knock level, knock threshold, timing trim, misfire clues, and crank/cam sync status if logged.

Fuel and lambda checks

Start with lambda error, not a standalone AFR number. Lambda is cleaner across fuels because it describes mixture relative to stoichiometric for that fuel, while AFR values depend on the fuel scale being assumed. If a gasoline-scaled AFR gauge says one thing and the fuel is ethanol-blended, lambda plus target lambda is the better diagnostic language.

• If measured lambda goes leaner than target as RPM/load rises and closed-loop correction trends positive, suspect base fuel table, fuel pressure drop, injector characterization, or delivery limit.
• If measured lambda oscillates around target while correction swings hard both directions, suspect control instability, exhaust leak, sensor placement, or a poor transient area rather than a simple global fuel error.
• If lambda looks perfect but fuel correction is maxed out, the tune may be relying on closed-loop correction to hide a base error. That is not a healthy final state.
• If fuel pressure is not logged, avoid pretending you have proven the fuel system. You may have proven only that the wideband reported a mixture during this one condition.

Ignition and knock checks

Ignition review is not just “did knock happen?” Compare commanded timing, delivered timing, knock activity, and any timing trims or protection states. A log can look conservative because the ECU pulled timing, not because the base timing table is conservative.

• If knock activity rises with intake air temperature and timing trims pull timing, the next decision may be charge-air temperature control, octane, plug heat range, or timing shape rather than fuel alone.
• If RPM breaks up while lambda spikes lean and boost is steady, separate true lean operation from misfire or spark blowout. A misfire can put oxygen in the exhaust and make the wideband report lean even when fuel delivery is not the root cause.
• If timing suddenly changes without an obvious table reason, look for compensations: coolant, IAT, boost, flex fuel, traction, knock, or protection strategies.

Temperature, pressure, and voltage checks

Many bad tuning decisions come from reading a performance trace while ignoring the environment around it. Heat soak, unstable battery voltage, and weak pressure data can create symptoms that look like calibration problems.

• Coolant temperature should be stable enough for the test. A pull during warmup is not directly comparable to a pull after three heat-soaked dyno runs.
• Intake air temperature changes can explain power drop-off, knock sensitivity, timing trims, and lambda differences between runs.
• Battery voltage affects injector dead time, ignition energy, pumps, fans, and sensor behavior. A voltage sag during high load deserves attention before fine tuning.
• Oil and fuel pressure, when logged, should be reviewed as trends against RPM/load, not just minimum and maximum values.

Example: lean high-RPM pull

The symptom is measured lambda drifting lean above 6,500 rpm. The diagnostic path is to compare target lambda, fuel correction, injector duty cycle, fuel pressure, battery voltage, MAP, and throttle position in the same window. If target is stable, throttle is wide open, MAP continues rising, fuel correction trends positive, injector duty is high, and fuel pressure falls, the evidence points toward delivery limit or pressure control. If fuel pressure is stable but correction is positive and injector duty is modest, the base fuel model or injector data may be more likely.

Example: boost overshoot or creep

For boost behavior, put MAP, boost target, throttle, RPM, gear, wastegate duty, and boost control state together. Haltech’s Boost Control User’s Guide and closed-loop base duty cycle docs are useful references because boost diagnosis depends on whether the ECU is commanding a duty change and whether the hardware can respond. If duty drops but MAP keeps climbing, suspect mechanical creep, spring pressure, exhaust backpressure, or wastegate flow. If duty oscillates and MAP follows, suspect control strategy or base duty calibration.

Example: heat soak hiding as tune drift

A car that gets slower or knock-prone after repeated pulls may not need a broad fuel/timing rewrite. Compare IAT, coolant temperature, timing trims, lambda correction, boost, and power/acceleration shape across runs. If IAT climbs, timing trims increase, and acceleration falls while boost and lambda stay similar, the issue may be charge cooling, intercooler recovery, fan placement on the dyno, or staging time.

Use a short evidence format

A useful note has four parts: what the run was, what changed, what the data showed, and what happens next. This is the difference between collecting logs and building a development history.

• Run: third-gear dyno pull from 3,000 to 7,200 rpm, E70 fuel, 18 psi target, coolant 185–192°F, IAT 92–118°F.
• Observation: lambda was 0.03 leaner than target above 6,400 rpm, correction reached +9%, fuel pressure fell 6 psi relative to base plus boost, injector duty reached 86%.
• Decision: do not increase boost; inspect fuel delivery and regulator reference before another high-load pull.
• Follow-up log: repeat same gear/ramp after fuel pressure fix with the same channel group and file naming convention.

Label inconclusive logs instead of deleting them

A missed shift, aborted pull, bad sensor, or missing channel is still useful history if it is labeled correctly. Mark it as inconclusive, write why, and keep it out of baseline comparisons. That prevents the same bad file from becoming evidence later.

Turn the first review into a repeatable workflow

Once you have a clean first review, save the channel layout, naming convention, and checklist. The goal is not to become slower and more formal; the goal is to spend less time rediscovering the same context every time you open a Haltech log.