AnasSarkiz/usb-c-pd-brushed-motor

docs/OUTPUT-CAPACITOR-REVIEW.md

# A10 output capacitor review — 2026-10-03

C18 is now the unchanged official C178373 / Panasonic 35SVPK330M import, **330 µF / 35 V**, replacing C2918361. The new part is connected provisionally; it is not a completed power/assembly approval. Direct supplier stock was 251 on 2026-10-03, and its native polarized-capacitor symbol, two electrical pins/pads, full Circuit JSON schema and five native checks pass. C16/C17/C29 DC-bias and current-sharing qualification remain open. The old capacitor's 150 mA rating at 120 Hz was insufficient evidence for high-frequency acceptance or rejection; the replacement supplies a manufacturer-documented PWM-frequency rating.

The imported SMD geometry has no THT drill/paste conflict. Footprint lands differ from Panasonic's recommended F12 pattern, so assembly-process approval remains a **stage-2 blocker** pending tolerance/process review. No footprint was patched.

## Supplier imports and mechanics

Official `tsci import` models are retained unchanged. Native A4, unrouted, single-component probes separate import inspection from product placement. Four independently built candidates each pass build, netlist, pin_specification, source, schematic-placement and placement; the strict complete-document schema/source/schematic/PCB-pin/pad audit passes with zero routed traces. This is electrical/schema evidence, not final capacitor approval. Raw naming, missing-reference, pin-annotation and supplier-refetch advisories remain in the reports and are not suppressed or blanket-accepted.

| Supplier / manufacturer part | Current direct-page stock | Review result |
|---|---:|---|
| C178373 / Panasonic 35SVPK330M | 251 | Active provisional C18. 330 µF ±20%, 35 V; 18 mΩ maximum at 100–300 kHz/20°C. F12 body diameter max 10.5 mm and seated height max 12.7 mm; base max 10.5 × 10.5 mm. Imported body outline 10.452 mm, courtyard 14.000 × 10.8 mm; lands 4.500 × 1.650 mm, inner gap 4.500 mm, outer span 13.500 mm. Recommended lands are 4.4 × 1.9 mm, gap 4.3 mm, span 13.1 mm. Actual terminals 0.8–1.1 mm wide and 11 ±0.2 mm overall span fit the imported lands nominally; full tolerance/process approval is outstanding. Pin 1 is positive at the chamfered end. Two top rectangular paste apertures (3.150 × 1.155 mm) remain inside the SMD pads; no holes. Network-enabled build emits no warnings/errors. |
| C136279 / Panasonic 25SVPF100M | 444 | Import and SMD paste geometry pass. 100 µF / 25 V, 3.2 A at 100 kHz, 24 mΩ at 100–300 kHz. Imported lands 4.200 × 1.400 mm, inner gap 2.334 mm and span 10.734 mm differ from the manufacturer's recommended E7 lands (4.15 × 1.9 mm, gap 2.8 mm, span 11.1 mm). Body/base nominal dimensions reconcile; land/assembly review remains pending. Single-part motor-PWM ripple margin is insufficient under the conservative screen; a pair requires current-sharing and compensation review. |
| C133439 / Nippon Chemi-Con APSG250ELL331MJB5S | 1,352 | Exact official import and two electrical pins pass. Pitch 5.08 mm is within manufacturer 5 ±0.5 mm; 1 mm drills accept 0.6 mm leads, 1.6 mm copper gives 0.3 mm annulus. Body 10 mm diameter, maximum seated height 13 mm must be reviewed separately from copper-only component bounds. **Paste-process blocker:** four circular apertures cover the THT drills on both sides. The supported `--use-exact-footprint` reimport still produces them. No component patch or stencil workaround applied. |
| C178367 / Panasonic 25SVPF330M | 0 | Official import succeeds, but current stock is zero. Not selected; full independent geometry/native audit not completed. |
| C19191929 / KNSCHA 118EC333 | 0 | Import, two pins, native checks and SMD paste geometry pass. Current manufacturer-authored supplier PDF contradicts itself: 8 × 11.5 mm in the exact part table, 6.3 × 9 mm in the drawing. No frequency multiplier supplied. Mechanical/ripple approval blocked; not selected. Older search results showing 1,430 stock are stale. |
| C20539445 / Ymin NPLD0901E331MJTM | not refreshed | Official import fails: component not found in EasyEDA library search. Candidate rejected; no custom component created. |

Stock for unused candidates is from direct supplier HTML fetched at 2026-10-03 11:11 UTC; C178373 has its own later timestamp in supplier-C178373-A10.json; it is an observation, not reserved inventory. Exact URLs, hashes and timestamps are retained in evidence/polymer-supplier-availability-A10.json. C709533 is discontinued/out of stock and C2161943 is out of stock, so neither was imported as a production substitute.

The C133439 paste behavior is generated by the core's circular plated-hole branch, not cured by avoiding the automatic footprinter substitution. Strict schema validation accepts these positive-size apertures; that does not approve pin-in-paste or bottom-side paste for this board. A9's statement that THT paste failures were resolved refers to the earlier invalid-geometry/schema errors, not all stencil-process decisions. Existing active THT components still need manufacturing-process review before fabrication.

## Ripple and compensation screen

scripts/buck-output-review.py implements the TI TPS54360 §7.3.14–16 continuous-conduction equivalent circuit with explicit candidate profiles. Nine independent limiting-case comparisons against TI Eq.11 pass. The active C178373 profile uses its 18 mΩ specification and screens all 5/9/12 V rails with a deliberately broad 16,200-case sensitivity sweep. **Threshold failures remain; this is not an approved compensation design.** Exact results are in evidence/buck-output-C178373-A10-review.json. The earlier unused C133439 14 mΩ sweep (1,570 violations) remains historical in buck-output-review-A10.json.

The varied GM, ESR, delay and MLCC parameters are assumptions, not guaranteed part corners. Constant 0.1/0.6 Ω ESR models exceed either candidate's specification at 100–300 kHz and cannot describe actual full-band response. Crossovers above 48 kHz are outside this simplified model's design range; their phase margins do not establish hardware instability. Nominal plots are preliminary. A frequency-dependent capacitor/regulator model, DCM/light-load review and transient validation are required before changing or approving the network.

C178373's manufacturer frequency factor is 0.7 throughout 10–100 kHz. At approximately 20 kHz this gives **3.08 A** from the 4.4 A allowable 100 kHz ripple rating, conditional on the manufacturer's **Tx ≤105°C** limit. The conservative motor/buck screen below is about **2.467 A RMS**, leaving conditional PWM ripple margin. At 105°C < Tx ≤125°C the 100 kHz rating falls to **1.39 A** (0.973 A after the factor), so this is explicitly not a 2 A motor rating through 125°C. Lower-frequency startup/mechanical envelopes, mixed-frequency current sharing and the unspecified ≥500 kHz multiplier at the 600 kHz buck frequency still require qualification.
For C133439, manufacturer 10 kHz ripple multiplier 0.60 screens 5 A as 3 A; using this lower tabulated multiplier for approximately 20 kHz is a conservative engineering assumption pending manufacturer confirmation. A signed motor-current waveform can approach the 2.423 A current-limit screen at 50% duty; treating it only as a unidirectional I√D(1−D) load is not conservative. Buck triangular-ripple RMS is ΔI/√12. Combining the independent PWM and switching components gives approximately **2.467 A RMS**. This is below 3 A conditionally, but does not resolve lower-frequency mechanical envelopes/startup/reversal, waveform/current sharing, temperature, or the unspecified multiplier above 500 kHz at the current 600 kHz buck frequency.

No phase-margin, current-sharing, thermal, charger or motor measurement is invented. The board remains an untested prototype; placement and routing are disabled.

## Reviewed evidence

evidence/polymer-import-audit-A10.json, polymer-native-checks-A10.json and polymer-geometry-A10.json identify the exact current imports and generated pads/paste. C133439 pre-exact results are preserved separately. C178373/C136279/C133439/C19191929 PCB renders and manufacturer mechanical/table pages were inspected, including the C178373 temperature/ripple footnotes and F12 recommended-land drawing. The current KNSCHA supplier PDF reproduces the drawing contradiction; both supplier and older manufacturer-authored copies are retained. Plot, full sensitivity failures and Python dependency manifest are in evidence/buck-output-*A10 and power-review-python-dependencies-A10.txt.

References: [Panasonic SVPK catalog](https://industrial.panasonic.com/cdbs/www-data/pdf/AAB8000/AAB8000C226.pdf), [Panasonic F12 land patterns](https://industrial.panasonic.com/ww/products/pt/os-con/models/35SVPK330M), [Panasonic SVPF catalog](https://industrial.panasonic.com/cdbs/www-data/pdf/AAB8000/AAB8000C177.pdf), [Chemi-Con exact C133439 part](https://www.chemi-con.co.jp/en/products/detail-condenser.php?part_number=APSG250ELL331MJB5S), [PSG catalog](https://www.chemi-con.co.jp/products/relatedfiles/capacitor/catalog/PSGLL-e.PDF), [KNSCHA exact supplier datasheet](https://datasheet.lcsc.com/datasheet/pdf/6ed2fa56348efb259b15876ad7072a94.pdf?productCode=C19191929), [TI TPS54360](https://www.ti.com/lit/ds/symlink/tps54360.pdf).

## A14 manufacturer-model update

The exact Panasonic simulation ZIP/curve data are now available and retained unchanged. A14's frequency-dependent CCM screen passes12,960 cases at the original thresholds, with77.77° minimum margin and31.848 kHz highest crossover. Model/S-parameter agreement is0.22%; published curve differences and missing temperature data are documented. Earlier constant-ESR failures remain historical, not hidden. See BUCK-MODEL.md. This is a conditional design screen and does not complete land-process, thermal, DC-bias,600 kHz current sharing or physical approval. Hardware/BOM remain A11 and routing/placement are disabled.