TroubleshootingN54 N55 S55 B58 S58

My tuner asked for a datalog: the fifteen-minute answer

You got the message — 'send me a log of it happening.' What your tuner needs, the pull that carries evidence, what makes a log worthless, and how the upload actually works.

4 min readUpdated 2026-10-20Rapid Racing

The message arrives at the worst moment — "something's not right, can you send me a log of it happening?" — and it lands like homework. It isn't. A datalog request is the tuner's version of a doctor ordering bloodwork: it's how the conversation moves from symptoms to numbers, and it takes about fifteen minutes to do properly. This guide is the runbook for the moment the request lands. For the tooling reference, see the MHD datalogging guide; for how to read what comes out, reading your datalog. This page is just: your tuner asked, tonight, now what.

What your tuner is actually asking for

A single logged pull that shows the problem happening — the same stretch of road, one clean sweep, with the channels that carry the story. When the request names a symptom ("boost falls off up top," "it pulls timing after the shift"), the pull needs to include that moment. The log is evidence, not a general health snapshot: a file of puttering around town answers nothing about a top-end problem.

The pre-flight (five minutes, saves the whole evening)

**Monitor license active, adapter connected** — the same adapter that flashed the car reads the channels; see the [adapter guide](../mhd-adapter-guide) if connection is flaky. A log session is also the cheapest adapter health check there is.
**Units: PSI, not BAR.** Set it once. Half the back-and-forth in tuning threads is unit conversion, and your tuner reads PSI.
**Auto-logging armed** (start on pedal threshold) so the pull can't outrun the record button.
**Full pull of fuel, same fuel as the complaint.** If the symptom involves [E85 or a mix](../why-e85-changes-your-boost-targets), the tank in the car during the log must match the tank during the complaint — flex content is a channel, not a footnote.
**Fully warm — oil, not just coolant.** Cold-engine logs read like lies: enrichment, closed-loop lag, timing retard that vanishes at temperature.

The pull itself

The protocol, compressed — the fuller version lives in the datalogging guide:

Safe, legal, familiar road. **DSC fully off** — a stability event mid-pull corrupts the data.
Stabilize around **2000 rpm** in the long gear (3rd on 6/7-speed manuals, 4th on 8-speed autos).
**Wide-open throttle, smoothly, to redline**, then hold a couple of seconds before lifting.
One pull. Let it cool if you need another.

The log-killers (how pulls go worthless)

The failure modes we see in uploads, ranked by frequency:

**The drive-around.** Ten minutes of town driving with one half-throttle moment. No sweep, no story.
**The lift.** Panicked pedal work when the car surges — the log shows exactly that instead of the symptom.
**The shift mid-pull.** Autos hunting gears or a manual shift resets every trace's story right where it gets interesting. The long gear exists to prevent this.
**Heat-soak stacking.** Pull five back-to-back and IAT dominates everything — the log now describes the intercooler, not the complaint.
**The screenshot.** A photo of a gauge cluster or one zoomed graph frame. It's not a datalog; it's a souvenir. Send the **raw CSV**.

Getting it to us

Logs are CSVs on the phone — get the file out (email it to yourself, or use a log host), then upload it through your account on the platform. That's the part most people don't know: the upload isn't email-to-a-human, it's the front door of the revision loop. The platform verifies your VIN, the log lands against your tune's history, and the next revision gets built against evidence instead of adjectives. When the symptom is one of the classic families — boost behavior, fueling, ignition corrections — the numbers in that file map straight onto the guides: 30FF underboost stories read one way in a log, fuel-delivery sag reads another, and the tune-or-car differentiator starts from exactly this kind of file.

When the log isn't the answer

Some complaints don't live in a pull — cold-start behavior, in-town surging, a noise, a leak. Say that in the reply instead of forcing a highway log; the tuner will tell you which of the other guides fits (misfires start here, boost sag here, no-starts here). The datalog is the tool for things a pull can show — using it for everything is how bad logs happen.

Questions

Do I need to buy anything to send a datalog?

If you flashed with MHD, you need the Monitor license — that's it. The adapter you already own reads the channels; the phone you already own records them. No dyno, no laptop, no extra hardware.

Can I just send a photo of my dashboard gauges or a screenshot of a graph?

No. A screenshot carries one moment and one zoom level; the raw CSV carries every channel at full resolution for the whole pull. Your tuner will think of questions overnight that the screenshot can't answer — the raw file already does.

How many logs should I send?

One clean pull that shows the problem, plus one baseline pull in the same conditions if you can. Five messy pulls are five times the noise and none of the signal — one good one is the entire deliverable.