CAN & CANopen Studio
Universal CAN & CANopen Protocol Analyzer, Telemetry Plotter, and Transmit Station
CAN & CANopen Studio is a modern, modular, cross-platform protocol engineering suite designed for testing, reverse-engineering, diagnosing, and interacting with CAN bus networks and CANopen automation devices.
Initially created for electric kart telemetry (SEVCON Gen4 inverter & De Haardt safety transponders), it has been completely generalized into a universal tool compatible with industry standard hardware adapters, CiA 301 / CiA 402 drives, J1939 commercial vehicles, and arbitrary CAN traffic.
Key Highlights
- Multi-Adapter Hardware Support: Direct plug-and-play support for SLCAN (Lawicel CANUSB, USBtin, CANable), PEAK-System PCAN, Kvaser, Vector XL, IXXAT, gs_usb (Candlelight), Linux SocketCAN, and built-in Virtual simulation loop.
- Protocol & Device Decoders: Extensible decoding engine for generic CANopen (CiA 301 / CiA 402), SEVCON Gen4 inverters, De Haardt track safety transponders, 29-bit J1939 frames, and Raw CAN.
- Multi-Trace Telemetry Oscilloscope: Real-time simultaneous time-series plotting of all 8 payload bytes (B0 to B7) or decoded physical values (RPM, torque, temperatures, voltage).
- Network Node Monitor & Gauges: Live visual tracking of node NMT states (Operational, Pre-Operational, Stopped), message counters, and real-time RPM / torque dials.
- Full Transmission Console: Generic NMT Master controller, high-precision SYNC clock pulse generator (50 Hz), arbitrary frame transmitter, and pre-configured frame templates library.
- SDO Object Dictionary Explorer: Expedited dictionary reader and writer for inspecting and calibrating any CANopen node.
- OBD-II Vehicle Diagnostics (SAE J1979): Live parameters, trouble codes and VIN over an ELM327 (USB, Bluetooth SPP, Wi-Fi) or straight over a native CAN adapter via ISO-TP. Supported PIDs are discovered from the vehicle rather than assumed, vehicle profiles inherit from a generic J1979 base, and everything is read-only by default.
- Native Rust Engine: Frame parsing, CANopen classification, ISO-TP reassembly and the virtual simulator run as compiled code shared by every front end — see the measured results.
- Wireshark and PulseView Export: Capture to PCAP-NG, live through a named pipe or into a file, and reconstruct the waveform a frame would have made as VCD for sigrok — one capture, the protocol in one tool and the bits in the other.
- AI Agent Integration: Built-in MCP and A2A servers let an AI agent read the live trace, inspect nodes and telemetry, transmit frames, drive the NMT state machine and control the network bridge.
- Real Bus Network Bridge: Mirror a physical CAN bus — a drive, an inverter, a live harness — onto a UDP multicast group so remote machines observe the real traffic as if they were wired to it.
- Two Front Ends, One Engine: The full-featured Python studio, and a lean native front end built with Slint, both driving the same protocol core.
- Integrated In-App Updater: Automated background checks against GitHub Releases, direct 1-click Windows installer upgrade, and Git update automation.
Quick Navigation
Getting started
- Installation Guide — standalone installer, 1-click script, or Python package.
- Quick Start — connect hardware or launch the virtual simulator in seconds.
- Screenshots Gallery — a tour of the graphical studio.
Using the bus
- Hardware & Interfaces — supported USB-to-CAN converters, UDP multicast and the real-bus bridge.
- Protocol Decoders — how CAN packets and devices are decoded, and how to add your own.
- Wireshark Integration — capture files, live pipes,
Decode Asand display filters. - Logic Analyzer & Signals — probing the wire itself, one layer below the adapter.
- Telemetry & Oscilloscope — multi-trace plotting and signal reverse engineering.
- Transmission Console — NMT master, SYNC generator and frame templates.
- SDO & NMT Master — reading and writing the object dictionary.
- OBD-II Vehicle Diagnostics — live parameters, trouble codes and VIN from a vehicle.
Under the hood
- Architecture — one engine, two front ends, and where each piece lives.
- Native Engine & Front End — what the Rust core accelerates, and how to build it.
- Command-Line Tools — the Python sniffer and the native
canopen-cli. - Performance Benchmarks — measured latency, jitter and throughput.
Automation and project
- AI Integration — drive the bus from an AI agent over MCP or A2A.
- In-App Updates — how the studio updates itself.
- Roadmap and Changelog.