tiny-wb/docs/hardware-validation.md

3.2 KiB
Raw Blame History

tinySA hardware validation checklist

This checklist qualifies one physical instrument and firmware build. A passing mock or CI run is not hardware evidence. Run the validator once with a tinySA Basic and once with a tinySA Ultra; keep their reports separate.

Before each run

  • Identify the rear-label model and record the firmware shown by the device.
  • Disconnect every RF source. This test needs only the analyzer and a data-capable USB cable.
  • Close serial terminals and other tinySA software. Confirm the current user can open the USB serial port without running the validator as root.
  • If the front-panel configuration matters, save a preset. The validator changes repeat, raw zero offset, input mode, sweep limits, and pause state.

From the repository root, run exactly one command:

scripts/test-hardware.sh basic

For an Ultra, use:

scripts/test-hardware.sh ultra

The script asks when to unplug and plug the instrument so it can detect the new serial port. If discovery is ambiguous, choose from the displayed ports. To skip discovery, pass --port /dev/ttyACM0 or --port COM7. It builds the validator, runs two 101-point scans, validates the returned values, writes CSV files and an identity report under ignored hardware-validation/, and asks whether the front-panel sweep resumed.

Basic acceptance

  • The automatically selected port belongs to the Basic.
  • Both legacy (240–300 MHz low input) and direct (350–400 MHz high input) scans report PASS and create CSV files with 101 data rows plus a units header.
  • The printed version/info text matches the connected Basic and its recorded firmware.
  • The live front-panel sweep resumes after the scans; answer y only after checking it.
  • The final basic-report.txt records both CSV paths and operator_confirmed=y (or yes).

Ultra acceptance and limit

  • The automatically selected port belongs to the Ultra.
  • Both legacy and direct scans report PASS at 240–300 MHz and create CSV files with 101 data rows plus a units header.
  • The printed version/info text identifies the exact connected Ultra variant and firmware.
  • The live front-panel sweep resumes after the scans; answer y only after checking it.
  • The final ultra-report.txt records both CSV paths and operator_confirmed=y (or yes).

A passing Ultra run validates only the shared low-input command path below 350 MHz, including zero 128, binary framing, CSV export, and best-effort resume. It does not validate Ultra mode, harmonic mode, LNA behavior, ranges at or above 350 MHz, level accuracy, or frequency accuracy. Those require model-aware acquisition and a controlled reference source; do not infer them from this communications test.

Failure record

  • Keep the generated report and any completed CSV; do not edit them.
  • Record the command, host OS, selected port, exact model, firmware, and the complete FAIL: line.
  • If the sweep did not resume, disconnect any RF setup and use resume, reload the saved preset, or power-cycle the instrument.
  • Do not connect a transmitter merely to reproduce a communications failure.