DAMGood AGENTS.md

FA20 Engine Tuning Agents

Quick Start

# Initialize a new tuning project
./scripts/setup_tuning_project.sh ~/my-tuning-project

# Analyze datalogs
python scripts/analyze_datalog.py --wot path/to/wot.csv --cruise path/to/cruise.csv

Commands

• Analyze datalog: python scripts/analyze_datalog.py <file.csv>
• Setup project: ./scripts/setup_tuning_project.sh <project_dir>
• Build workflow: See .github/workflows/fa20-analyze.yml

Documentation

Reference docs under docs/ for: fuel, ignition, airflow, avcs, engine, throttle, sensors, transmission

---

Specialized Agents

Fuel Agent

Expert in fuel trims (STFT/LTFT), AFR targets, MAF scaling, open/closed loop transitions, and HPFP timing.

Thresholds:

Parameter	Green	Yellow	Red
STFT	±5%	±5-10%	>±10%
LTFT	±5%	±5-10%	>±10%

Ignition Agent

Expert in knock detection, DAM, timing tables, Fine Knock Learn, and per-cylinder knock thresholds.

Thresholds:

Parameter	Green	Yellow	Red
DAM	≥0.95	0.75-0.95	<0.75
Feedback Knock	0°	-1° to -3°	<-3°
Fine Knock Learn	0°	-1° to -2°	<-2°

Boost/Airflow Agent

Expert in wastegate duty, PI control, boost targets, MAF VE corrections, and IAT compensation.

AVCS Agent

Expert in intake/exhaust cam timing, barometric multipliers, TGV states, and cam advance targets.

Datalog Analyst

Interprets CSV datalogs, correlates fuel/ignition/boost behavior, identifies issues by load range.

---

Datalog Analysis Workflow

When analyzing FA20 datalogs, follow this exact process to generate comprehensive tuning reports:

Step 1: Load and Parse Data

import pandas as pd
import csv

# Load CSV datalogs
with open('wot.csv', 'r') as f:
    wot_rows = list(csv.DictReader(f))
with open('cruise.csv', 'r') as f:
    cruise_rows = list(csv.DictReader(f))

df_wot = pd.DataFrame(wot_rows)
df_cruise = pd.DataFrame(cruise_rows)
df_all = pd.concat([df_wot, df_cruise], ignore_index=True)

# Convert numeric columns
numeric_cols = [
    'Engine - RPM',
    'Fuel - Command - Corrections - AF Correction STFT',
    'Fuel - Command - Corrections - AF Learn 1 (LTFT)',
    'Ignition - Dynamic Advance Multiplier',
    'Ignition - Feedback Knock',
    'Ignition - Fine Knock Learn',
    'Analytical - Boost Pressure',
    'Sensors - AF Ratio 1',
    'PIDs - (F410) Mass Air Flow',
    'Engine - Calculated Load',
    'Throttle - Requested Torque - Main Accelerator Position',
    'Airflow - Turbo - Boost - Boost Target Final (Absolute)',
    'Airflow - Turbo - Boost - Manifold Absolute Pressure',
    'Airflow - Turbo - Wastegate - Duty Cycle Commanded'
]
for col in numeric_cols:
    df_all[col] = pd.to_numeric(df_all[col], errors='coerce')