app
Adom Fields
Public Made by Adomby adom
Adom Fields: where the current and the heat go on your board.
main
John Lauer
adom-fields 0.10.5: converter power ground and shunt values with ratings (issue 24)
7169a99
adom-fields 0.10.5
Two fixes for 4-switch buck-boost boards, both from issue 24 (thanks, Glenn):
- Resistor values with a rating or a size. A shunt's Value often carries its power rating or tolerance after the resistance:
3m 1W,3mR 1%,0R003 1W,3.0 0.75W,10k 0402. The parser joined the tokens (3m1W) and gave up, so a 3 mΩ shunt between the switch node and the inductor was not seen as a series link, and the review came backunsupported. Rating tokens (ending in W, V, % or containing ppm) are now dropped, and the resistance is the first token. Specconverter.series: ["R1"]names a series part outright when its Value says something else. - The converter's ground is the power ground. It was the first ground-named net in alphabetical order, which on a board with AGND and PGND is AGND, the controller's quiet ground. The review then saw only the small AGND decoupling caps (0.03 µF at bias, so tens of volts of ripple) and no hot loops. The ground is now the net the switch legs return to (the low-side switches' sources). Ties go to the net that more input and output capacitors share. Spec
converter.groundoverrides it, andreview.converterrecordsground,groundFromand every ground net, so a wrong pick is visible. - Checked:
- On a 4-switch test board, the shunt with Value
3m 1Wand a split AGND/PGND gaveunsupportedin 0.10.4. In 0.10.5 it gives the same numbers as the single-ground board: buck mode D 0.816, ΔI 2.31 A, hot loop C1; boost mode D 0.125, ΔI 1.37 A, hot loop C2. - On the 12 V buck, every finding and number is unchanged.
- On a 4-switch test board, the shunt with Value
# adom-fields 0.10.5
Two fixes for 4-switch buck-boost boards, both from issue 24 (thanks, Glenn):
- **Resistor values with a rating or a size.** A shunt's Value often carries its power rating or tolerance after the resistance: `3m 1W`, `3mR 1%`, `0R003 1W`, `3.0 0.75W`, `10k 0402`. The parser joined the tokens (`3m1W`) and gave up, so a 3 mΩ shunt between the switch node and the inductor was not seen as a series link, and the review came back `unsupported`. Rating tokens (ending in W, V, % or containing ppm) are now dropped, and the resistance is the first token. Spec `converter.series: ["R1"]` names a series part outright when its Value says something else.
- **The converter's ground is the power ground.** It was the first ground-named net in alphabetical order, which on a board with AGND and PGND is AGND, the controller's quiet ground. The review then saw only the small AGND decoupling caps (0.03 µF at bias, so tens of volts of ripple) and no hot loops. The ground is now the net the switch legs return to (the low-side switches' sources). Ties go to the net that more input and output capacitors share. Spec `converter.ground` overrides it, and `review.converter` records `ground`, `groundFrom` and every ground net, so a wrong pick is visible.
- **Checked:**
- On a 4-switch test board, the shunt with Value `3m 1W` and a split AGND/PGND gave `unsupported` in 0.10.4. In 0.10.5 it gives the same numbers as the single-ground board: buck mode D 0.816, ΔI 2.31 A, hot loop C1; boost mode D 0.125, ΔI 1.37 A, hot loop C2.
- On the 12 V buck, every finding and number is unchanged.