Waveshare USB TO RS485 (C) reference profile

The first supported lab adapter is the Waveshare USB TO RS485 (C), SKU

  1. Waveshare documents an FT232RNL USB interface, an isolated RS-485 field side, A+/B-/GND terminals, hardware-automatic direction control, Linux support, and an onboard 120 Ω balancing resistor.

Consequences for this project

  • Use normal 8N1 serial mode. Do not enable Linux TIOCSRS485, RTS toggling, GPIO DE/RE, or another direction-control scheme at the same time.
  • Use the FTDI serial number through /dev/serial/by-id/...; /dev/ttyUSB0 may change after reboot or replug.
  • A+ connects to A/+, B- connects to B/-. Connect field-side GND to the controller’s isolated reference/common when the device manuals and measured potentials permit it. This is not the host USB ground.
  • The onboard resistor counts as one 120 Ω endpoint termination. The operator’s unpowered measurement of roughly 130 Ω across A+/B- is consistent with a nominal onboard termination plus component/meter tolerance.
  • Do not assume the termination provides fail-safe bias. Termination and bias are different electrical functions and must be verified independently.
  • Do not open, drill, or modify the sealed adapter as part of this project.

Approved initial topologies

Isolated two-adapter bench 

Linux USB                         Linux USB
Waveshare C #1                   Waveshare C #2
  A+  =============================  A+
  B-  =============================  B-
  GND =============================  GND
 [120 Ω]                         [120 Ω]

With power removed from both field sides, the complete bus should measure near 60–65 Ω across A+/B- because the two onboard endpoint resistors are in parallel. This topology is useful for raw serial and isolated MS/TP tests.

Existing live MS/TP trunk

The C adapter must be treated as a terminated endpoint, not a transparent mid-span probe. If the existing two physical ends are already terminated, adding this adapter in the middle creates a third termination and changes the bus load. For a live BAS/controller trunk, either:

  1. temporarily make the C adapter one physical end and ensure only the opposite end retains termination; or
  2. use an adapter whose termination can be verifiably disabled for a short, topology-compliant tap.

Do not transmit merely because raw bytes are visible. The passive gate must decode valid complete frames and tokens first.

Ubuntu inventory commands 

ls -l /dev/serial/by-id/
serial=/dev/serial/by-id/usb-FTDI_REPLACE_ME
udevadm info --query=property --name="$(readlink -f "$serial")" \
  | grep -E '^(ID_VENDOR_ID|ID_MODEL_ID|ID_SERIAL|ID_USB_DRIVER)='
readlink -f "$serial"

Record the actual USB IDs, serial, kernel driver, by-id link, kernel version, baud, topology, powered-off resistance and reference wiring in every hardware artifact. Do not hard-code a VID/PID based only on marketing material; capture what Linux reports for each physical unit.

If the driver exposes an FTDI latency setting, record it:

serial=/dev/serial/by-id/usb-FTDI_REPLACE_ME
device="$(basename "$(readlink -f "$serial")")"
cat "/sys/bus/usb-serial/devices/${device}/latency_timer" 2>/dev/null || true

Changing sysfs or setserial settings requires an explicit operator action and must be reported as part of the test environment. A low latency value is not a substitute for correct frame-stream handling in the stack.

Preflight checklist

  • USB identity and by-id path recorded
  • No other process owns the tty (fuser -v <by-id-path>)
  • Adapter field side unpowered while measuring resistance
  • Exactly two endpoint terminations in the complete topology
  • Bias/reference arrangement verified
  • A+/B- polarity confirmed from device manuals
  • Baud, MAC, Max_Master and Max_Info_Frames agreed
  • Passive valid-frame/token gate passes before TX
  • Existing supervisory workstation remains healthy