WRX ATLAS Datalogging Guide

How to record a log worth reading

Technique matters more than parameter count. **Use third or fourth gear for full throttle pulls.** First and second pass through the rev range too quickly to produce useful resolution, and they load the drivetrain harder for no diagnostic benefit. A third gear pull from low rpm to redline is the standard because it spends enough time at each load point to show what is happening. **Get the car properly warm first.** A log taken on cold oil tells you about a cold engine, not about your tune. Wait for oil temperature to come up rather than trusting the coolant gauge. **Log normal driving too.** This is the advice most often ignored. Drivability problems live at part throttle and cruise - hesitation, surging, poor tip-in response - and a log consisting only of wide open throttle pulls contains none of that. If you are working with a remote tuner, part-throttle logs are frequently what they need most. **Note the conditions.** Fuel and octane, ambient temperature, and anything unusual. The same map behaves differently on a cold morning and a hot afternoon, and a log without context is harder to interpret. **Change one thing at a time.** If a map revision, a new intake and a different fuel all arrive together, a subsequent observation has three possible causes and tells you nothing. ATLAS also provides an always-on logger maintaining a rolling buffer, which is genuinely useful for catching transient events you were not deliberately recording.

What logs commonly reveal

Reading knock data without panicking

Knock monitors cause more unnecessary alarm than any other channel, largely because owners apply advice written for older Subarus. Subaru's knock detection deliberately errs toward caution, which means false knock is common. It shows up predictably when accelerating from a stop while releasing the clutch, during gear shifts and especially if you grunt one, accelerating at low rpm in a high gear, cruising on the motorway, and during abrupt throttle changes. Corrections clustered in those situations are a car hearing noise, not a car in trouble. Feedback Knock Correction of roughly -2.5 degrees or less is generally considered normal and not a major concern. The combination that warrants investigation is a Dynamic Advance Multiplier below ideal together with corrections beyond that, and it is especially worth attention if it occurs at full boost and wide open throttle rather than at cruise. The important VB-specific point: on direct injection Subarus the DAM strategy is deliberately more reactive than on older non-DIT cars. Corrections that would have appeared as Fine Knock Learning on an older car can show up as a DAM drop here, and a drop is not automatically an alarm. How long DAM takes to return to its maximum says nothing about how serious the event was, and it only climbs back after driving in specific load and rpm ranges without knock. So read patterns, not single numbers. Isolated corrections in classic false-knock situations are noise. Repeated corrections in the same rpm and load area under boost are a message worth acting on.

Gauges versus logs

ATLAS supports configurable gauge sets for at-a-glance monitoring while driving, alongside detailed logging for review afterward. They serve different purposes and it is worth using both deliberately. Gauges are for awareness in the moment - oil temperature so you know when the car is ready, boost so you can feel what you are seeing, and a knock indicator so you notice something unusual as it happens. Keep the set small, because a screen full of numbers is a distraction rather than information while driving. Logs are for analysis afterward, where you can scroll, zoom and compare without a road in front of you. This is where the real understanding comes from. The mistake is trying to analyse while driving. Watching a knock channel at full throttle is both unsafe and useless - you cannot interpret a transient reading in real time, and you are not looking where you should be. Record it, park, then read it. If you are working with a remote tuner, they will want logs rather than descriptions. Learning to record clean, well-labelled logs of the specific conditions they ask for is the single most useful skill for getting a good map out of an e-tune, and it is entirely a matter of process rather than expertise.