20261025-0012-air-hub-single-side-io Air Hub / Air Hub Pro: all connectors, key and status LEDs on one side

status: completed
priority: P2
owner: worker/session-20261015-1500
createdAt: 2026-10-25 00:12

Description

On 2026-10-25 the owner asked for the button, the connectors, the switch, the Ethernet port and the status LEDs to sit on one side of the set-top-box enclosure where possible; if not possible, the Ethernet port and the M8 socket go on a side of their own ("尽可能把按钮，接口，开关，网线状态放在一边，如果不可以的把网线，m4 独立放一边").

Today (revision D, 104 x 90 mm): RJ45 and RP-SMA in the left wall, USB-C in the front wall, M8 in the right wall (lead to J1 at the top right), PAIR key and SYS / RADIO LEDs on the top face. There is no power switch.

Proposed scope (awaiting approval):

Plan A, preferred: the left wall becomes the only I/O side. Top to bottom: M8 panel socket, RP-SMA, USB-C, PAIR side key, SYS / RADIO right-angle LEDs with light pipes in the wall, RJ45 (unchanged). Board changes: the M8 input and CAN chain (J1, fuse, TVS, LM74502 + FET, ACM7060, filter, TCAN3413, ACT45B, NUP3105L) moves from the top right to the top left next to the K7805; the USB-C, USBLC6, HUSB238, SMF15A and B360A move to the left edge between the SMA and the RJ45; the USB pair to the core board (pins 55 / 56, bottom edge) is routed on L3 past the PHY; the PAIR key becomes a side-actuated switch and the LEDs right-angle parts at the left edge. The board may grow about 10 mm in y if the left half does not fit. Antennas unchanged (MG24 right edge, C6 front edge, 433 MHz via the SMA).
Plan B, fallback if A does not route or makes EMC worse: RJ45 and M8 alone in the left wall (input chain moved to the left as in A, RP-SMA moved to the top wall above the RFM310); USB-C, PAIR side key and the two LEDs grouped in the front wall next to the USB-C.
Enclosure model, renders, interference check, product READMEs and docs updated as for task 20261024-2357.

Decisions (2026-10-25, owner):

The two status LEDs become one right-angle RGB LED (Kingbright APFA3010LSEEZGKQBKC, common anode, active-low cathodes through 1k / 1k / 330R) on core-board pins 72 / 73 / 74 (LCD control pins, no display); pins 83 and 99 are freed. The PAIR key becomes a side-acting Panasonic EVQ-P7A01P. No power switch.
Plan A was tried on paper and dropped: the core board has the USB pins on its bottom edge and RMII on its left edge, so a USB-C in the left wall needs a 480 Mbit/s pair (or the 12 V VIN path) across the RMII fan-out, which costs EMC margin. Plan B is implemented: left wall M8 + RJ45 only (input and CAN chain moved to the top left, the 5 V module turned round, the 3.3 V buck next to the core board), front wall USB-C + PAIR + RGB light pipe, RP-SMA in the rear wall above the RFM310 (433 MHz block moved to the top right).

Acceptance criteria:

ERC 0 and net verification clean; DRC 0 unconnected, 0 errors, 0 parity issues.
EMC pre-check errors not above 23, the USB pair change explained.
All I/O in one wall (plan A) or RJ45 + M8 alone in one wall (plan B); renders and 0 interference.
Docs gate passes.

ActiveForm

Moving the Air Hub I/O to one side

Dependencies

blocked by: (none)
blocks: (none)

Notes

The Air Hub Mini is out of scope. No commit until the owner asks.

Result (2026-10-25):

Schematics regenerated for both variants: ERC 0, net verification 0 mismatches. Core sheet: one RGB LED with three resistors on pins 72 / 73 / 74, side-acting PAIR key; pin map updated.
Board re-placed and routed (104 x 90 mm unchanged): DRC 0 unconnected, 0 errors, 0 parity; warnings: 4 library mismatches and 1 silkscreen-to-edge (the RGB LED sits on the board edge by design). EMC pre-check 58 findings, 22 errors (was 23); SPICE 22 / 22 pass.
Enclosure: left wall M8 + RJ45, front wall USB-C + flush PAIR plunger + RGB light pipe, rear wall RP-SMA, closed top; new front and rear renders; board / enclosure interference 0.
Cost: +$0.26 per board (Pro $48.50, Air Hub $37.42 at 100 boards); reference board R1.18, specification R1.21, product plan totals, EMC evaluation and product READMEs updated.