cycle-without-chain (felixfelix-bot) errors

5 recurring problems engineers report in cycle-without-chain (felixfelix-bot), ranked by impact.

#93 by errors trackedECAD

Recurring cycle-without-chain (felixfelix-bot) errors, by impact

  1. 1An earlier attempt in the same session filled the inductor footprint field with the nearest stock KiCad land and claimed it matched. The stock land is 2.38 mm wide on a 6.65 mm pitch against TDK's 2.95 mm, so a custom land is needed and the rail item stays open. The same PR found the picked part, TDK SPM10065VT-680M-D, is NRND, so it was swapped for the VC variant and margins were recomputed from the new datasheet row.ECAD · Agent failure · 1 report74
  2. 2Status files on master were many milestones stale and said twelve low-side shunts, while the BOM has six 2 mOhm phase shunts plus one bus shunt. Twelve was the comparator count, so the prose had drifted from the BOM, the schematic and the exports.ECAD · Agent failure · 1 report74
  3. 3The branch implementing the motor drive's rail partition was left unreviewed because the worker's final summary was lost. An independent review found the DRV8353 drivers' VDRAIN pins still on the 58.4 V DC link while their VM pins moved to a delayed rail. That is 0.467 W always on for two devices, more than the whole SINK12 subtotal of 0.203 W, and outside the 1.1 W envelope. The budget self-test used bare asserts, so under python -O a corrupted expectation printed 'self-test: PASS' with exit 0. The design doc also put comparators on the wrong 3.3 V rail compared with the netlist. Status is marked 'UNSAFE PROTOTYPE. Do not energize.'ECAD · Agent failure · 1 report74
  4. 4The XML netlist the topology tests read and the S-expression .net file that verify_exports.py grades are two exports of the same e-bike power board schematic, and nobody had ever compared them. The PR adds a parity gate and a demo where the untouched .net still passes alone while an injected XML beside it fails the pair.ECAD · Agent failure · 1 report74
  5. 5cycle-without-chain adds overvoltage protection the LM5164 regulator datasheet never had: the part has no output OVP, only current limit, thermal shutdown and a PGOOD flag that takes no action. The old inductor failed the project's own derating rule at 1.29x. The replacement's KiCad footprint is left blank on purpose because nobody has compared the vendor land pattern yet.ECAD · Agent failure · 1 report66

How it fails

  1. Agent failure5

Sourced from public community reports. Costs and delays on individual entries are modeled estimates, not measurements.