Modbus RTU faults sit at three layers: physical RS-485 wiring, serial settings and the Modbus application request. Diagnose in that order and keep a known-good request as a control.
Follow the workflow
Every device on one RTU segment must agree on baud rate, data bits, parity and stop bits. 9600 8-E-1 and 19200 8-N-1 are both common; neither is universal. Record the actual settings instead of cycling through guesses.
Do this in the product
Use the Modbus lab to establish the unit ID and application request independently from the serial transport.
Open the exerciseA two-wire trunk should be daisy chained with short stubs, consistent A/B polarity, biasing at one point and termination at the two physical ends when required. Star wiring and multiple bias networks create reflections and unstable idle states.
Do this in the product
Save one project for the expected register map and name it after the physical segment so wiring and protocol evidence stay connected.
Open the exerciseAn RTU request contains unit address, function, data and a two-byte CRC. There is no MBAP transaction ID. A silent interval separates frames, so overloaded gateways or poor timing can merge or truncate traffic.
Do this in the product
Run the holding-register challenge and use its PDU as the middle of the RTU frame; a real analyser would add unit address and CRC.
Open the exerciseNo response suggests wiring, serial settings, duplicate address, timing or a dead device. A valid exception response proves the physical and serial layers worked well enough to exchange a frame, narrowing the fault to function, address or value.
Do this in the product
Use the Pro illegal-address challenge to practise reading exception 02 rather than changing random serial settings.
Open the exerciseCore concepts
The simulator creates a repeatable result; these concepts make that result transferable to real vendor software and supervised practical work.
RS-485 receivers respond to the voltage difference between the two conductors, improving noise immunity over long industrial runs.
Each server on a segment needs a unique address, usually 1–247. Address 0 is reserved for broadcast requests that do not receive a reply.
The cyclic redundancy check detects corrupted frames. Repeated CRC errors are evidence to inspect cabling, grounding, topology and noise.
Continue in the workspace
Run the foundation exercise publicly, then use a subscription for advanced challenges, saved configurations, full attempt history, sharing, assigned paths and team reporting.
Related learning surfaces