Silicon proof · Domain A
One binary. Twin and board. Matched contract.
On 2026-07-21 we ran the same, unmodified Domain A firmware ELF on a CircuitTwin twin of an STM32F103 and on a physical STM32F103 Sensor Demo Board (Domain A) over ST-LINK. Both legs exercised the DAQ path to a Winbond W25Q128. At the data-communication-contract level the sealed verifier reported verified=true, parity=true (schema v2), with the device bring-up and limits stated below — nothing estimated. Request early access via email if you want to run the same class of checks on your boards.
What ran
- MCU: STM32F103C8T6 (Device ID 0x410), FreeRTOS firmware domain-a-rtos-1.1.0
- Probe: ST-LINK/V2; flash programmed and read back byte-identical to the ELF image
- DAQ storage: W25Q128JV — JEDEC ID EF 40 18 on the physical board
- Device bring-up on silicon: 15 PASS, 2 FAIL (LSM6DS3, MPU6050), 2 NOT_FITTED (BMP280 by design)
What matched
- Physical
/daq.csvon NOR and the twin’s virtual/daq.csvboth committed and verified under the sealed schema-v2 parity contract - The twin loop caught a real BME280 model bug (impossible pressure) before it would have misled a hardware bring-up
- Full sealed packet lives under
verification-artifacts/(Domain A DAQ parity set); this page is the public summary, not a substitute for the artifacts
Honest limits
- This is one Domain A DAQ proof — not a claim across the full top-40 board list
- USB CDC did not enumerate on this board revision; boot evidence used the SWD snapshot path
- Two fitted devices failed on silicon and are reported as FAIL, not hidden