What is MDF4? A guide to ASAM MDF4 for automotive engineers

What is MDF4?

MDF4 — or ASAM MDF version 4 — is the binary file format used across the automotive industry to store measurement data captured during vehicle tests, dyno runs, calibration sessions, and road campaigns. The format is defined and maintained by ASAM e.V. (Association for Standardisation of Automation and Measuring Systems).

An MDF4 file (extension .mf4) holds time-series signal data alongside rich metadata: signal names and descriptions, physical units, conversion rules (e.g. voltage → temperature), channel groups, and attachments such as CAN databases or calibration files. A single MF4 from a highway test might contain thousands of simultaneously logged signals — CAN bus messages, ECU parameters, analog sensor readings, GPS coordinates, and accelerometer data — at varying sample rates, all time-aligned in one file.

This makes MDF4 the de facto data exchange format between OEMs, Tier-1 suppliers, test equipment vendors, and calibration tool providers across the industry.

MDF4 file structure

MDF4 uses a block-based binary structure. Understanding the main blocks helps when debugging signal extraction or building tooling around the format:

MDF3 vs MDF4 — what changed

MDF3 (.mdf) is still produced by older data loggers and some legacy measurement systems. Here is what MDF4 adds:

How to analyze MDF4 files

There are several paths for working with MDF4 data, depending on whether you need a graphical interface, scripted access, or production pipeline integration.

Graphical analysis tools

For interactive signal exploration, overlay plotting, and oscilloscope-style inspection, a dedicated MDF4 analyzer is the fastest way to work. MDX by Vehicle Lab is a professional MDF/MF4 data lab with sub-second loading for GB-sized files, a high-resolution oscilloscope with dual cursors, a variable explorer with cross-DBC signal matching, and offline GPS route mapping. It includes an AI diagnostics layer (DiagAI) for natural-language queries over signal data.

Other graphical tools include ETAS INCA (calibration-focused), Vector CANalyzer (CAN protocol analysis), and neoVI (Intrepid Control Systems).

Python — asammdf

The open-source asammdf library is the standard Python interface for MDF3 and MDF4 files:

pip install asammdf

from asammdf import MDF

mdf = MDF("test_run_highway.mf4")

# Get all channel names
print(mdf.channels_db.keys())

# Read a signal as a pandas Series
rpm = mdf.get("engine_speed_n")
print(rpm.samples[:10], rpm.timestamps[:10])

# Export selected channels to CSV
mdf.export(fmt="csv", channels=["engine_speed_n", "lambda_1", "coolant_temp"])

asammdf handles MDF3 (.mdf) and MDF4 (.mf4), DZ compression, VLSD channels, and bus-logging data transparently. It is the dependency used internally by MDX's backend for file parsing.

MDX Python SDK and REST API

For production automation — batch processing of test campaigns, signal anomaly detection, WLTP analysis across fleets of MF4 files — MDX exposes a Python SDK (mdx_client) and REST API on top of its DuckDB time-series engine:

from mdx_client import MDXClient

client = MDXClient("http://localhost:5001")

# Import a file and get all signals indexed
file_id = client.files.import_file("run_42.mf4")

# Query a signal with LTTB downsampling
data = client.signals.get(file_id, "engine_speed_n", max_points=2000)

# Run misfire analysis
result = client.analysis.misfire(file_id)
print(result["events"])

The REST API allows scripted integration into CI/CD pipelines, test automation systems, or data warehouses — keeping the heavy data processing local while surfacing results via API.

MDF4 vs CSV for vehicle data

CSV is frequently used as a data exchange format in automotive test, especially when teams want something universally readable. But MDF4 has significant structural advantages for measurement data:

• →Multi-rate signals. A single MF4 can store a 1 Hz GPS signal alongside a 10 kHz combustion pressure signal, time-aligned, in one file. CSV requires either resampling (lossy) or separate files per rate.
• →Built-in metadata. Units, conversion rules, channel descriptions, and signal provenance are in the file. CSV has no standard metadata convention.
• →Size. A compressed MF4 is typically 3–5× smaller than the equivalent CSV for floating-point data. A 10-minute highway run at 1000 signals × 100 Hz = 6 GB as CSV, ~1.5 GB as compressed MF4.
• →Random access. MDF4 tools can seek to specific time ranges without reading the full file. CSV requires a full sequential parse.

CSV is reasonable for small exports or sharing isolated signals with external parties. For anything at test-campaign scale, MDF4 is the right format.

Common MDF4 tooling questions