app
Adom Fields
Public Made by Adomby adom
Adom Fields: where the current and the heat go on your board.
main
Colby Knox
adom-fields 0.10.8: capacitor dielectric from the MPN's vendor code (issue 31); Murata C0G in the PDN model
f01bf13
adom-fields 0.10.8
- A capacitor's dielectric from its MPN (issue 31). The capacitor-temperature check knew a part's rating only when the text "X7R", "C0G" or a °C rating appeared in its value, footprint or MPN. Samsung, Murata, KEMET and Walsin encode the dielectric as a letter or two inside the part number, so a Samsung CL05B104KO5NNNC (X7R, 125 °C) was taken as X5R (85 °C), and an LED strip with 44 such caps at about 78 °C got a false finding.
- The MPN field's vendor codes are now read:
- Samsung CL (A X5R, B X7R, C C0G, F Y5V);
- Murata GRM, GCM, GCJ, GRT and GJM (5C C0G, R6 X5R, R7 X7R, C7 X7S, C8 X6S, D7 X7T, F5 Y5V);
- KEMET C… (R X7R, P X5R, G C0G, U Z5U, V Y5V);
- Walsin (N C0G, B X7R, X X5R, F Y5V).
- TDK and Yageo spell the dielectric out (X7R, NPO), and that was already read. X8R and X8L now count as 150 °C.
- A code not in the table is never guessed: the part stays at the assumed 85 °C, as before.
- Each entry in
checks.capacitorsnow hasdielectric,ratingFrom(the value, or "MPN CL05B104KO5NNNC (Samsung B code)") andmpn. - The finding for capacitors whose rating is still assumed now says why: the dielectric couldn't be read from the value or the MPN field, or how many have no MPN field. Before, it said the dielectric "isn't in the part number". The fix it suggests starts with filling in the MPN field.
- The MPN field's vendor codes are now read:
- Murata C0G in the PDN model. The C0G check read the wrong two characters of a Murata part number. A GRM1885C… above 1 nF was modelled as a class-II part that loses capacitance under bias. It now uses the same decoder.
Checked:
- A test LED board with six 100 nF 0402 caps at about 80 °C:
- 0.10.7 flagged all six against an assumed 85 °C.
- 0.10.8 reads the three Samsung CL05B and the one Murata GRM155R7 as X7R (125 °C), and flags only the cap with no MPN field and the one whose MPN has no known code. The message counts the one with no MPN field.
- Unit tests: 25 part numbers across Samsung, Murata, TDK, Yageo, KEMET and Walsin, including unknown codes that must stay unread.
- Real boards, every module, 0.10.7 against 0.10.8:
- The decoder read the real MPNs right: StickHub's Murata GRM188R61A226 as X5R, the 12 V buck's Walsin 0603B104 as X7R and Samsung CL21A226 as X5R, and the tinytapeout demo's Samsung CL10A as X5R.
- No finding changed, except the reason wording on the BQ25792 board, whose parts have no MPN field.
- PDN output is identical apart from solve time.
Not checked: noah's LED strip board itself (no file).
# adom-fields 0.10.8
- **A capacitor's dielectric from its MPN (issue 31).** The capacitor-temperature check knew a part's rating only when the text "X7R", "C0G" or a °C rating appeared in its value, footprint or MPN. Samsung, Murata, KEMET and Walsin encode the dielectric as a letter or two inside the part number, so a Samsung CL05B104KO5NNNC (X7R, 125 °C) was taken as X5R (85 °C), and an LED strip with 44 such caps at about 78 °C got a false finding.
- The MPN field's vendor codes are now read:
- Samsung CL (A X5R, B X7R, C C0G, F Y5V);
- Murata GRM, GCM, GCJ, GRT and GJM (5C C0G, R6 X5R, R7 X7R, C7 X7S, C8 X6S, D7 X7T, F5 Y5V);
- KEMET C… (R X7R, P X5R, G C0G, U Z5U, V Y5V);
- Walsin (N C0G, B X7R, X X5R, F Y5V).
- TDK and Yageo spell the dielectric out (X7R, NPO), and that was already read. X8R and X8L now count as 150 °C.
- A code not in the table is never guessed: the part stays at the assumed 85 °C, as before.
- Each entry in `checks.capacitors` now has `dielectric`, `ratingFrom` (the value, or "MPN CL05B104KO5NNNC (Samsung B code)") and `mpn`.
- The finding for capacitors whose rating is still assumed now says why: the dielectric couldn't be read from the value or the MPN field, or how many have no MPN field. Before, it said the dielectric "isn't in the part number". The fix it suggests starts with filling in the MPN field.
- **Murata C0G in the PDN model.** The C0G check read the wrong two characters of a Murata part number. A GRM1885C… above 1 nF was modelled as a class-II part that loses capacitance under bias. It now uses the same decoder.
**Checked:**
- A test LED board with six 100 nF 0402 caps at about 80 °C:
- 0.10.7 flagged all six against an assumed 85 °C.
- 0.10.8 reads the three Samsung CL05B and the one Murata GRM155R7 as X7R (125 °C), and flags only the cap with no MPN field and the one whose MPN has no known code. The message counts the one with no MPN field.
- Unit tests: 25 part numbers across Samsung, Murata, TDK, Yageo, KEMET and Walsin, including unknown codes that must stay unread.
- Real boards, every module, 0.10.7 against 0.10.8:
- The decoder read the real MPNs right: StickHub's Murata GRM188R61A226 as X5R, the 12 V buck's Walsin 0603B104 as X7R and Samsung CL21A226 as X5R, and the tinytapeout demo's Samsung CL10A as X5R.
- No finding changed, except the reason wording on the BQ25792 board, whose parts have no MPN field.
- PDN output is identical apart from solve time.
**Not checked:** noah's LED strip board itself (no file).