Skip to content

ThingDAQ 1.1.0 — 1 MHz acquisition at 450 MHz

Latest

Choose a tag to compare

@billmccartney billmccartney released this 05 Sep 05:28

ThingDAQ 1.1.0 uses wire protocol v2—not v3. Upgrade the firmware and Python package together.

Fixed acquisition configuration

  • 450 MHz Teensy 4.0 core.
  • ADC0 and ADC1: each 1 MS/s; GPIO: 1 MS/s, with or without ADC capture.
  • 8 GPIO inputs by default; optional 16-input mode using D6–D13 and D16–D23.
  • Rate profile 4 is the only supported profile. Other rates and output modes are rejected.
  • GET_TEMPERATURE reads calibrated die temperature while idle, configured or running.
  • The legacy 4 MHz GPIO capture diagnostic is disabled and no longer advertised.
  • More ADC raw-buffer processing headroom; stricter SDK capability and STOP-tail handling.

Protocol and migration

The current contract is protocol v2, backed by the generated JSON schema. INFO is 680 bytes, STATUS is 1476 bytes, and GET_TEMPERATURE is command 0x1A / response 0x9A. Experimental 632-byte INFO clients must be updated. Historical v1 documentation is explicitly marked and its frozen schema is unchanged.

Qualification and limitations

The exact attached HEX passed a 600-second combined ADC + 16-GPIO soak with all 131 checks passing. Measured ADC/GPIO rates were approximately 1 MHz; die temperature increased from 39.7°C to 54.5°C, peaking at 55.1°C.

Four additional 20-second 8→16→8→16 input transitions passed on that HEX. The final SDK run passed its first four modes and the rate interlocks but rejected one checksum in combined 16-input mode, with no firmware raw-ring loss or packet-pressure evictions at that snapshot. The corruption's origin remains unconfirmed; this is not full SDK end-to-end qualification.

See the release qualification report and attached validation/evidence files for the final physical results, temperature measurements and failure lineage. Local validation passed 562 tests and 16,324 subtests, plus generation, lint, type and linked-memory gates. Two clean builds produced the same HEX.

The Python SDK's combined ADC + 16-GPIO path is not qualified as lossless on the remote test server's half-core CPU quota. Tests reported firmware USB queue loss and, in one run, parser rejections whose cause remains unconfirmed. Strict checks remain enabled. The independent wire validator sustains this firmware configuration. A faster host's SDK throughput remains unverified.

TimestampAligner does not support this profile's unequal-duration ADC/GPIO frames. Use read_block() / blocks() and per-sample timestamps. Tests have no declared external stimulus, so they do not establish analog accuracy or external GPIO timing. No output pins are driven.

Downloads

firmware.ino.hex is the pinned Teensy 4.0 / USB Serial / 450 MHz image. Verify it against SHA256SUMS; the attached build manifest identifies source commit e3e8417598de7494db3a93d7277ef8f0ec31bc8f and build thingdaq-636aebbecbf70691. The optional Python wheel is a private local distribution, not a PyPI publication.