Aug 12-13, 2026
Day 4 — MLG Format Cracked, Cooling Bleed Begins
5 logs analyzed · 63 minutes of logged data · 0 stall/restart events detected
Infrastructure day
The binary .mlg format was fully reverse-engineered this session — record stride, channel descriptor table, and the header layout — which is what made direct, quantitative analysis of every subsequent log possible instead of relying on manual TunerStudio inspection. The coolant bleeding process also began in earnest, and a fan step-response test on Start-1 produced the first hard evidence that the idle solenoid polarity was inverted.
1649
Avg idle RPM (session-wide)
04Log-by-Log Breakdown
Session-wide statistics computed across each full log. These pool every idle-qualifying sample in the file, so they run higher than the narrow steady-state windows cited in the final project report — that's expected; this view shows total session variability, not just the best segment.
Integra-Start-4.mlgLOW CRANK VOLTAGE 0.0V
First binary MLG log received. The .mlg format (record stride, channel table, delta encoding) was reverse-engineered this session, enabling direct quantitative log analysis for the rest of the project.
Start-1-ColdStart-CoolantBleed.mlgLOW CRANK VOLTAGE 0.0V
Cold start during coolant bleeding. A fan step-response test here (+10 position -> -750 RPM) revealed the idle solenoid polarity was inverted.
Start-2-WarmStart-CoolantBleed2.mlgLOW CRANK VOLTAGE 0.0V
Warm bleed session, continued air purge.
Start-3-WarmStart-CoolantBleed3.mlgLOW CRANK VOLTAGE 0.0V
Warm bleed session, continued air purge.
Start-4-WarmStart-CoolantBleed4.mlgLOW CRANK VOLTAGE 0.0V
Warm bleed session, continued air purge.
* On stall/restart counting
Detected algorithmically as RPM dropping below 300 for a sustained period followed by a subsequent rise — a reasonable proxy for restart events, not a precise fault classifier. Treat as directionally accurate, not exact.
