Board-to-Board Interconnect Candidates
Connector families for carrying the 15V/3A rail (plus signal) between the PD board and the LED/control board.
Board P (PD front end) and the LED/control board are separate PCBs per the epic's board-split decision. This page evaluates connector families for that link — power (15V, contract cap 3A) plus whatever status/signal lines the architecture pass ends up needing (e.g. an ATT/PDOK line per the Board-P audit). All parts lookup-verified against the local JLCPCB DB (fetched 2026-07-25; per-part lookup dates noted).
Candidate 1 — JST XH, 6-pin (zudo-pd baseline)
| Spec | Value |
|---|---|
| Part | B6B-XH-A(LF)(SN) (C144397) |
| Pitch | 2.5mm |
| Rating | 250V / 3A per contact |
| Pins | 6 |
| Stock (tier) | 47,935, partdetail |
This is the connector zudo-pd used, with power contacts doubled (2 of the 6 pins ganged per rail, both for the 15V feed and the return) specifically because a single XH contact's 3A rating sits with zero margin against this project's 3A contract cap. That leaves 2 of the 6 pins for signal/status.
Pros: proven in the sibling project, cheap, small footprint, positive-locking housing, widely stocked. Cons: a single contact has no headroom at all over the 3A cap — doubling is mandatory, not optional, and it consumes 4 of the 6 available pins just for power+return before any signal line is added.
Candidate 2 — JST VH, 2-pin (headroom alternative)
| Spec | Value |
|---|---|
| Part | B2P-VH(LF)(SN) (C160315) |
| Pitch | 3.96mm |
| Rating | 250V / 10A per contact |
| Pins | 2 |
| Stock (tier) | 207,876, partdetail |
VH is the "step up" connector family precisely for cases like this: a single contact carries 3.3× the 3A contract cap with margin, so no doubling is needed — one pin per rail is enough. A 3-pin (VH-3A / B3P-VH, C160316) or 4-pin (VH-4A / B4P-VH, C160317) variant in the same family would add signal pins without touching the power-margin story.
Pros: real single-contact margin over the 3A cap, same connector family scales to more pins if signal lines are added, robust for hand assembly/rework. Cons: larger 3.96mm pitch and taller housing than XH — bigger board-edge footprint; splitting power (VH) and signal (a separate small connector) into two connectors is more parts/more assembly steps than XH's single 6-pin combo, unless a wider VH variant is used for everything.
Candidate 3 (for comparison) — Molex-style KK 2.54mm, 4-pin
| Spec | Value |
|---|---|
| Part | 2510-4A (C2847671) |
| Pitch | 2.54mm |
| Rating | 250V / 3A |
| Pins | 4 |
| Stock (tier) | 104,421, partdetail |
Included to show that XH isn't an outlier — most connectors in this 2–2.5mm pitch class top out at 3A per contact, i.e. exactly at this project's cap with zero margin. This family would need the same doubled-contact treatment as XH if selected; it isn't a genuine alternative to the margin problem, just a different footprint with the identical constraint.
Comparison
| Connector | Pitch | Per-contact rating | Margin over 3A cap | Doubling needed? | Pin count used for power+return |
|---|---|---|---|---|---|
| JST XH (C144397) | 2.5mm | 3A | None | Yes (zudo-pd already does this) | 4 of 6 |
| JST VH (C160315) | 3.96mm | 10A | 3.3× | No | 2 |
| Molex KK 2510 (C2847671) | 2.54mm | 3A | None | Yes | 4 of 4 (no pins left for signal without a larger variant) |
Questions the architecture pass must answer
Is XH's proven-but-zero-margin doubling scheme (already validated by zudo-pd's design, though that project's PCBAs were never bench-confirmed per issue #5) acceptable here, or does the LED board's higher continuous current profile (vs. the PD board's own downstream) justify the headroom VH offers?
How many signal/status lines actually need to cross this connector (ATT/PDOK status, anything else)? That fixes whether XH's remaining 2 pins are enough, or whether VH needs a companion small-signal connector.
Board-edge space and mechanical stacking (is this a stacked or side-by-side board arrangement?) — XH's smaller footprint vs. VH's larger one may be decided by enclosure/board-outline constraints not yet fixed at this research stage.