How to Compare WOT Datalogs: Why Two Pulls Can Look Different

You make a wide-open-throttle pull and open the datalog. Boost appears to follow its usual pattern, fueling looks relatively consistent, and nothing immediately stands out.

Ten minutes later, you make another pull with the same vehicle, map, hardware, and fuel. This time the log looks different.

Boost is slightly lower. Intake air temperature is higher. Timing behavior has changed. Fuel trims have moved. Fuel pressure follows a slightly different path.

Nothing was intentionally changed—so why did the datalog change?

Because a datalog records how the vehicle behaved under the exact conditions present during that particular pull. It is not a permanent fingerprint of the vehicle or a promise that the next pull will produce identical numbers.

A single datalog is a snapshot. Multiple comparable datalogs provide context.

One log can show what happened during one pull. Additional comparable logs can help show whether that behavior repeated, improved, worsened, or was simply part of normal pull-to-pull variation.

This article is about interpreting and comparing telemetry. It does not provide tuning instructions, prescribe calibration changes, or establish universal safe and unsafe thresholds.

A datalog is a snapshot

Every WOT pull takes place under a particular combination of conditions, including:

• Intake air temperature
• Ambient temperature
• Engine and coolant temperature
• Transmission temperature
• Starting RPM
• Gear
• Vehicle speed
• Road gradient and load
• Fuel composition
• ECU adaptations
• Accelerator and throttle behavior
• Traction or torque intervention
• The thermal effects of earlier driving or previous pulls

Even when the tune and hardware remain unchanged, differences in those conditions can influence the telemetry recorded during a pull.

This is not unique to enthusiast datalogging. Controlled vehicle testing pays close attention to test conditions because repeatability depends on controlling relevant variables. One SAE study of vehicle testing found that measured results were affected by ambient conditions, engine load and speed, and the speed-and-acceleration profile of the test. Although that research concerned emissions testing rather than modified-vehicle WOT logs, the underlying measurement lesson is relevant: results are easier to compare when the operating conditions are comparable.¹

Two logs therefore do not need to be numerically identical for both to represent normal operation. The first question should not be, “Are these numbers exactly the same?” It should be, “Were these pulls comparable, and did the important signals behave in a meaningfully different way?”

Heat and intake air temperature

A vehicle may begin its first pull with a relatively cool intake system and engine bay. After idling, driving slowly, sitting after a pull, or completing repeated WOT pulls, the vehicle may retain additional heat.

That retained heat can change the conditions at the beginning of the next pull. The intake tract, intercooler, cooling system, engine bay, and surrounding components may no longer be in the same thermal state.

This is why intake air temperature, or IAT, should be part of a WOT datalog comparison rather than treated as background information.

Useful questions include:

• What was IAT at the beginning of each comparable WOT section?
• How did IAT move through the RPM range?
• Did one pull begin after more idling or low-speed driving?
• Was one pull made shortly after an earlier pull?
• Were ambient conditions similar?
• Did other telemetry change at the same time?

Temperature context can help explain why two pulls occurred under different operating conditions. It does not, by itself, prove why another signal changed.

For example, a warmer pull that also contains more timing-correction activity does not establish that the IAT increase caused the timing behavior. RPM, load, fuel, control strategy, measurement differences, and other factors may also be relevant. The log establishes what occurred together; a causal conclusion requires stronger evidence.

Starting RPM, gear, and load

Two logs are not automatically comparable simply because the accelerator reached 100 percent in both.

Consider these two pulls:

These pulls expose the engine to different operating conditions. The gear changes the relationship between engine speed, vehicle acceleration, and the time spent under load. The higher starting RPM also removes part of the lower-RPM response from the second log.

Comparing peak values from these two pulls without considering those differences can be misleading.

A more useful comparison uses reasonably similar:

• Gear
• Starting RPM
• Ending RPM
• RPM range
• Accelerator demand
• Throttle behavior
• Vehicle configuration
• Fuel
• Thermal and environmental conditions

Perfect laboratory control is rarely possible on the road. The goal is not to make every condition identical; it is to make the conditions similar enough that observed differences have useful context.

Throttle and traction intervention

A pull that feels uninterrupted from the driver’s seat may not be uninterrupted in the telemetry.

Depending on the vehicle and logging platform, a WOT log may contain:

• A brief throttle closure
• A reduction in accelerator input
• Traction-control activity
• Torque intervention
• A gear change
• A short interruption in load
• Wheel slip
• A logging gap or irregular sample interval

Electronic throttle control and traction control are distinct ECU capabilities, and a logger may record accelerator position, throttle angle, wheel speeds, or related control channels separately.² Driver demand and throttle-plate position should therefore not be assumed to be the same signal.

A brief interruption can affect boost, airflow, fueling, timing, and fuel-pressure behavior at the same point in the pull. If one log includes an interruption and the other does not, the two sections may not represent equivalent WOT operation.

Before comparing the outcome, identify the relevant interval:

1. Find the beginning and end of the intended WOT section.
2. Confirm that accelerator demand remains reasonably consistent.
3. Inspect actual throttle behavior where that channel is available.
4. Look for gear changes, traction events, or other interruptions.
5. Compare the overlapping RPM and load range rather than unrelated portions of the files.

Fueling behavior can vary

Fueling telemetry does not necessarily reproduce identical values during every pull.

Depending on the platform and available channels, relevant signals may include:

• Actual AFR or lambda
• Commanded or target AFR/lambda
• Short-term and long-term fuel trims
• High-pressure fuel-pump pressure
• Low-pressure fuel-pump pressure
• Compatible fuel-pressure targets
• Ethanol content, where measured
• Documented fuel-enrichment or auxiliary-fueling channels

Channel meaning matters. A field should only be treated as commanded fueling when its documented meaning is a commanded or target AFR/lambda value. A vendor-specific fuel or enrichment field must not automatically be substituted for an AFR/lambda target.

When compatible target and actual channels are available, compare how the actual measurement followed the command throughout the relevant section. When a target is absent, the actual signal is still useful: its stability, movement, persistence, and relationship with load can still be compared across equivalent pulls. The missing target only prevents a target-error calculation; it does not invalidate the rest of the log.

Instead of focusing on one isolated value, ask:

• Did actual AFR or lambda follow a similar shape?
• If a compatible target exists, did target-versus-actual behavior change?
• Did fuel trims follow approximately the same pattern?
• Did actual fuel pressure remain stable through the comparable interval?
• If a compatible pressure target exists, did tracking change?
• Did pressure movement persist or occur only at one sample?
• Did the behavior repeat in another comparable pull?
• Did several related fueling signals change together?

Fuel composition and learned control behavior also deserve context. An EPA vehicle test program documented that vehicles used learned fuel adjustments and required operating time across different speed and load modes for fuel-control behavior to stabilize after a fuel change. The same program used repeated tests because individual test results contained inherent variability.³

That does not imply that every trim change is adaptation or that a specific amount of adaptation time applies to every vehicle. It demonstrates why fuel history and repeat observations can matter when comparing logs.

Boost behavior is more than peak boost

Suppose two pulls produce these peak values:

• Pull 1: 18.7 psi
• Pull 2: 18.3 psi

The 0.4 psi difference alone does not establish that the vehicle improved, deteriorated, or developed a problem.

Peak boost is one point in a time-series dataset. It does not describe how quickly boost developed, how long it was sustained, whether the throttle remained open, or how actual boost related to the requested target.

A stronger comparison considers:

• Target or requested boost, when available and correctly identified
• Actual boost
• Target-versus-actual deviation
• Spool or response behavior
• Midrange behavior
• High-RPM behavior
• RPM and load
• Accelerator and throttle position
• Wastegate-related telemetry, where available and correctly normalized
• Whether any intervention occurred

When a compatible target exists, the useful question is:

How did actual boost behave relative to the requested target throughout each comparable WOT section?

If a target is not logged, do not invent one. Actual boost can still be compared for response, stability, persistence, and high-RPM behavior alongside RPM, throttle, load, and wastegate-related signals. That is an actual-only comparison, not a target-tracking analysis.

Timing behavior should be evaluated across the pull

Timing deserves the same treatment. One isolated value should not be used to declare an entire log good or bad.

First, confirm what the channels mean. Ignition advance and timing correction or retard are not interchangeable. A channel should only be analyzed as correction when the logger or platform documentation identifies it that way.

Then compare:

• Where timing changes occurred
• The RPM and load at those points
• Whether the behavior was brief or persistent
• Whether it appeared in one cylinder or several
• Whether it repeated in another comparable pull
• Whether throttle, boost, fueling, or temperature behavior changed at the same time
• Whether the later pull became better, worse, or simply different

Timing behavior is platform- and calibration-dependent. The purpose of comparison is to identify patterns and changes, not to assign a universal threshold or diagnose a component from one event.

Normal pull-to-pull variation

Not every telemetry difference represents a problem.

Sensors fluctuate. Environmental conditions change. Load changes. ECU control behavior adapts. Driver input varies. Sampling times may not line up perfectly between files.

Measurement science distinguishes repeatability from perfect numerical identity. NIST describes variability as the tendency of a measurement process to produce slightly different measurements even on the same test item, and notes that temperature, time, handling, and other conditions can contribute to the variation.⁴

For a WOT datalog comparison, this means small differences should not automatically be interpreted as improvement or deterioration.

A meaningful comparison asks whether the difference is:

• Large enough to alter the overall behavior
• Persistent rather than momentary
• Present across a meaningful RPM/load interval
• Supported by related signals
• Repeatable in another comparable log
• Explainable by a known difference in test conditions

The objective is not to eliminate every difference. It is to separate ordinary variation from meaningful, repeatable change.

Illustrative example: two WOT pulls

Illustrative example — all values and descriptions below were created for educational purposes. They are simulated and are not taken from an actual vehicle, customer log, or CSV file.

This example does not show that the higher IAT caused the change in timing behavior. It only demonstrates how two pulls from the same hypothetical vehicle can contain different conditions and different telemetry.

A reasonable interpretation would be:

• The second pull occurred under a warmer measured intake-air condition.
• Peak boost differed slightly.
• The described fuel-pressure and trim patterns remained broadly similar.
• More timing-correction activity appeared in the second pull.
• Throttle remained open in both pulls.

A reasonable next question would be whether the timing behavior repeats in another pull made under comparable conditions. It would not be reasonable to diagnose a failed component, declare a particular temperature unsafe, or prescribe a tuning change from this simulated comparison.

Three types of comparison

1. Normal variation

Pull 1 and Pull 2 contain small numerical differences, but their overall telemetry behavior remains similar.

For example:

• Boost follows a similar shape
• Target tracking, where available, is comparable
• Fueling behavior remains consistent
• Fuel pressure follows a similar pattern
• Timing behavior does not show a meaningful repeatable change
• Throttle remains comparable

Lesson: Different numbers do not automatically mean the vehicle changed in a meaningful way.

2. Something became worse

The second comparable pull contains a regression in one or more measured signals.

Examples might include:

• A larger or more persistent boost-tracking deviation
• Increased timing-correction activity in a similar RPM/load region
• Increased fuel trims through the comparable interval
• A more pronounced decline in actual fuel pressure
• A throttle interruption that was absent from the first pull

This comparison identifies what changed. It does not automatically identify why it changed.

A possible cause should only be presented as a possibility when the necessary supporting signals and platform context exist. Correlation between signals is evidence to investigate, not proof of causation.

Lesson: Multiple logs can reveal a change that may deserve further investigation.

3. Something improved

An earlier log contained a specific telemetry behavior, and a later comparable log shows that the behavior became less pronounced or is no longer present.

Examples might include:

• Actual HPFP behavior became more stable under a comparable load
• Fuel trims decreased during the same operating region
• Boost tracking became closer or more consistent
• Previous timing-correction activity did not repeat
• A previous throttle interruption was absent

The comparison should remain specific. “The previously observed pressure decline did not repeat” is more defensible than “the fuel system is fixed.” The latter would require broader evidence.

Lesson: A follow-up datalog can help verify whether a previously observed behavior actually improved.

Compare behavior, not just numbers

A datalog is a time-series dataset. Each row represents a point in time, and the relationships between signals are often more informative than a single maximum or minimum.

Do not compare only:

18.7 psi versus 18.3 psi

Compare:

• The shape of each signal
• Direction and rate of movement
• Stability and persistence
• Target versus actual, when compatible target data exists
• RPM and load context
• Accelerator and throttle behavior
• Related control signals
• Where changes begin and end
• Whether the same behavior repeats

It is also important to align comparable regions. If one pull starts earlier, compare the shared RPM range rather than matching rows by their position in the CSV. Different loggers and exports may also use different sample rates, channel names, and units, so semantic normalization matters before values are compared.

This is particularly important across JB4, MHD, bootmod3, and other logging ecosystems. Similar-looking names do not guarantee identical meanings. Vendor documentation for one platform, vehicle, or firmware should not be generalized to every log source. For example, the published JB4 N54 logging reference explicitly describes its parameter meanings as applying to that application and firmware context.⁵

When another log becomes valuable

Another comparable log can add useful context:

• After changing fuel
• After maintenance
• After replacing spark plugs or another service item
• After changing hardware
• After receiving a tune revision
• After changing maps
• After unusual behavior appears
• After environmental conditions change significantly
• When checking whether a previous observation is repeatable
• When checking whether a specific behavior improved
• When one pull contains an interruption or incomplete WOT section

The purpose of the follow-up log is observation and verification. It is not to make random changes until the graph looks different.

A good follow-up comparison documents what changed between tests and keeps other relevant conditions reasonably similar. If several conditions change at once, it becomes harder to determine which difference matters.

How PullScan fits into the comparison

PullScan’s comparison tools are designed around this idea: a second log becomes more useful when it can be viewed in the context of the first.

Instead of treating each automotive CSV as an isolated analysis, comparison can help users inspect how relevant telemetry changed between pulls. This can make differences in boost, fueling, timing, temperature, throttle, and other available signals easier to see than manually reviewing two CSV files independently.

PullScan remains an analysis and explanation tool. A comparison does not replace source documentation, platform-specific knowledge, professional inspection, or a qualified tuner when those are needed.

Context turns a snapshot into a pattern

One datalog can tell you what happened during one pull.

Multiple comparable datalogs can help show whether that behavior:

• Repeated
• Changed
• Improved
• Worsened
• Or was simply normal pull-to-pull variation

The goal is not to make every pull look identical. It is to understand what changed, whether the change is meaningful, and whether the behavior is repeatable.

Sources and further reading