VisorMesh SB-LoRa v0.1
A Springboard module for the Handspring Visor that carries a Seeed XIAO nRF52840 + Wio-SX1262 stack running MeshCore companion firmware. The Visor talks to it over a bit-banged SPI link into an SC16IS740 UART bridge. The XIAO runs from its own LiPo and keeps meshing while the Visor sleeps or the module is unplugged.
Status: prototype design, not yet built. DRC is clean (0 unconnected pads) and the netlist was cross-checked against the Springboard Development Guide pin table, the SC16IS740 datasheet pinout, and Dmitry Grinberg's springSD v1.1 connector footprint. Order 5 boards and build one before trusting it.
This README covers rev 1 (hw-v0.1). Rev 2 (flash, bus UART, XIAO and Wio side by side, 4 layers) is laid out in rev2/; see REV2.md for its status and what has to happen before it can be ordered.
What's in here
| Path | What |
|---|---|
fab/visormesh_gerbers_jlcpcb.zip |
Gerbers + Excellon drill, upload as-is to JLCPCB or PCBWay |
fab/visormesh_bom_jlcpcb.csv |
BOM, JLC format (LCSC part numbers) |
fab/visormesh_cpl_jlcpcb.csv |
Pick-and-place, JLC format, top side only |
fab/hand_fit_parts.csv |
Parts you fit yourself (J1, J2, J3, JP1) |
fab/visormesh_schematic.pdf |
Net-label schematic + theory of operation |
fab/visormesh_assembly_top.pdf |
Assembly drawing with reference designators |
kicad/ |
KiCad 7 board + project + custom DRC rule (opens in KiCad 7 or 8) |
source/ |
The scripts that generated everything. netlist.py is the single source of truth |
../../firmware/ |
MeshCore patch: Xiao_nrf52_companion_radio_uart env (companion protocol on D6/D7) |
renders/ |
Top and bottom copper renders |
Ordering (JLCPCB)
- PCB: 2 layers, 1.6 mm, 1 oz, any color. ENIG recommended (flat pads for the 0.65 mm TSSOP, SC-70 parts and the straddle connector); lead-free HASL works too.
- Board is 50.43 x 75.15 mm. The bottom 50.47 mm is the in-slot section, using springSD's exact outline. The top 24.6 mm sticks out of the Visor and carries the XIAO stack.
- Assembly: top side, Economic PCBA is fine. Upload BOM + CPL. Check part rotations in JLC's placement preview before paying, especially U1-U7, Y1 and D1; KiCad and JLC disagree on rotation for some SOT/SC-70 footprints. Pin 1 of each IC is marked on the silkscreen and assembly drawing. The CPL already includes the corrections found in the first order's preview (U1, U2, U4 at 270; U3, U5-U7 at 180), so the dots should land on the pin-1 triangles without adjustment. Still glance at the preview: JLC can change its part models.
- Extended parts: SC16IS740 (C407851), 74LVC1G80 (C353857), crystal (C2901666) and probably the SN74LVC1T45 (C9382) and 74LVC1G04 (C10237). Expect extended-part loading fees.
Parts you fit by hand
- J1 Springboard connector: Hirose IC1HA-68RD-1.27SH(51), the PC-Card-style straddle receptacle springSD uses. It clamps the board edge; pins 1-34 land on the top pads, 35-68 on the bottom.
-
Obsolete (no longer stocked at DigiKey or Mouser as of Nov 2026). Sources: new-old stock on eBay or brokers (search Octopart for the exact part number with the
(51)suffix; about $24 each), or harvest one from a used Springboard module. -
The Springboard connector core is a standard PCMCIA 68-pin connector with different housing keying (Dev Guide §5, §7.1). Handspring used four board-offset variants (−0.3, 0, +0.3, +0.6 mm); a harvested connector's offset sets how high the board sits in the slot, so compare it against a (51) or dry-fit before soldering.
-
An earlier note here suggested XKB X1321FV-2x34 (LCSC C2881777) as a substitute. That was never verified: the closest LCSC listing (X1321FVS-2x34, C2881881) is a vertical SMD socket, not an edge straddle, and won't fit this footprint.
-
Pin 1 is marked on silk at the right-hand end of the top row.
-
Without the Hirose part. The order for one was cancelled (Nov 2026). Rev 1 only uses these Springboard pins, so any receptacle that fits the slot needs about 12 good joints:
Signal Pins GND 1, 34, 35, 68 CD1*, CD2* (to GND, or the OS sees no module) 36, 67 VCC (VSB) 17, 51 CS0*, CS1* 7, 42 A8, A17 12, 46 D0 30 IRQ* 16 RESET* 58 VDOCK (18, 52) is only needed if JP1 is bridged. Temporary options, best first:
- Harvest a straddle connector from a junk Springboard module (backup modules, Eyemodule, MiniJam and so on). It fits the footprint as designed; check its board offset against the slot first.
- Donor connector on wires. Cut a junk module's PCB just behind its connector, cut its traces, and wire its pads for the pins above to the matching J1 pads (or vias on the same nets). The donor strip sets the height in the slot; rev 1 sits outside, glued or taped to it.
- Single-sided SMT PC Card receptacle glued to the top at the edge. Top-row tails (1-34) go to the top pads where they line up (staggered tails need short jumpers). The 8 used bottom-row pins (35, 36, 42, 46, 51, 58, 67, 68) are wired around the edge to the bottom pads. The board then sits about 1 mm off the contact centre line, so dry-fit it: the edges may need filing to run in the slot's rails.
- Receptacle from a junk PCMCIA card. Same contacts, but its housing keys may block the Springboard slot, and a straddle part is sized for a thinner card PCB than 1.6 mm. Try it only if one is to hand.
Whichever you use, check continuity from each used pin to its net and do the bench-power checks in BRINGUP.md before it goes in the Visor.
-
- J2/J3: two 1x7 2.54 mm female headers (low-profile if you can find them). Your XIAO+Wio stack plugs in here. You can also solder the stack's pins straight into the holes.
- JP1: leave open. See VDOCK below.
Mechanical / orientation
- The XIAO sits with components facing up (same side as U1-U7) and USB-C at the outer edge. Silkscreen labels (D0..D6, 5V, GND, 3V3, D10..D7) show which header pin goes where. Check them against your stack before plugging in. If your stack only exposes pins upside-down, the labels tell you it's mirrored.
- Solder your LiPo to the XIAO's BAT+/BAT- pads first. The carrier doesn't touch the battery; the XIAO's own charger handles it over USB-C.
- In-slot components are all 1.25 mm or shorter, on the top side only, which is lower than springSD's microSD socket. The bottom side inside the slot is a flat GND pour.
- Assumption to verify: like springSD, the top edge of the 50.47 mm section lines up with the slot opening, and the extension (with the stack and antenna) protrudes like a battery-module bump. Dry-fit a bare board before soldering anything.
Electrical summary
- Bus SPI (same wiring as springSD, so its 68k routines port directly):
- A CS0* access latches A17 and A8 on its rising edge. Through Q-bar,
nCS = NOT A17andMOSI = NOT A8. - A CS1* read produces one SCLK pulse; MISO comes back on D0 through a 1k resistor.
- SC16IS740 is SPI mode 0. Register byte: bit7 = R/W (1 = read), bits 6:3 = register, bits 2:1 = channel 00.
- A CS0* access latches A17 and A8 on its rising edge. Through Q-bar,
- UART: 14.7456 MHz crystal (SCTF SX3B14.7456F2010F30, 20 pF load; replaced the out-of-stock SXT32413BA07 rated 13 pF), so 115200 baud with DLL=8, DLH=0 (prescaler 1). 64-byte TX/RX FIFOs. IRQ* goes straight to the Springboard IRQ* (the Visor has the pull-up). RESET* follows the slot RESET*.
- Power isolation: only U5-U7 (SN74LVC1T45) touch both domains, and both of their ports go Hi-Z when either supply is 0 V. Removing the module, a LOWBAT power cut, or a flat LiPo can't back-power the other side.
- Visor-present: XIAO D0 is high when the slot is powered. The firmware can stop pushing frames when it's low; the companion queues messages anyway.
- Slot budget: a few mA active, against the 100 mA limit. To stay inside the 100 uA standby limit, put the SC16IS740 to sleep (IER bit 4) when the Palm app is idle.
- VDOCK (JP1): when bridged, VDOCK (4.75-6.2 V, only present in a charging dock) feeds the XIAO 5V pin through a Schottky so the XIAO charger can top up the LiPo while docked. The standard Visor USB cradle almost certainly doesn't supply VDOCK on an AAA Visor. If you bridge JP1, never plug USB-C in while docked, because VDOCK would back-feed the USB host.
Bring-up order
BRINGUP.md has the full plan: checks and expected readings for each step, and what each line of the Visor's SB-LoRa Test means. In short:
- Bare board, before soldering anything: check continuity of VSB (J1 pins 17/51) and GND, then dry-fit in the Visor slot.
- After JLC assembly, fit J1 only. Bench-power VSB at 3.3 V through a current-limited supply. Expect a few mA, and the crystal at 14.7456 MHz on Y1.
- Insert into the Visor with nothing on J2/J3. Port springSD's bus routines and read the SC16IS740 scratch register (SPR, reg 0x07): write and read back a pattern. That proves the CS0/CS1 latch path.
- Flash the XIAO with
Xiao_nrf52_companion_radio_uart(patch in../../firmware/). Bench-test it over a 3.3 V USB-TTL adapter on D6/D7 with meshcore-cli before plugging the stack into the carrier. - Plug in the stack. Configure the SC16IS740 for 115200 8N1, send CMD_DEVICE_QUERY (
<+ len +0x16 0x03), and look for a>-framed DEVICE_INFO reply.
Known limits / things I could not verify from here
- Mechanical fit of the extension against the Visor body and the exact slot depth (see the assumption above).
- The J1 footprint is copied pad-for-pad from springSD v1.1 (verified against its IPC-356 netlist). I could not check it against the Hirose drawing.
- JLC part rotations (see Ordering).
- There is no
.kicad_sch: the schematic is generated fromsource/netlist.pyas a PDF, and the board's nets come from the same file. To regenerate on a machine with KiCad 7, Java and Freerouting 2.1:source/make.sh(autoroutes), thenmake_fab.py,verify.py,draw_schematic.py. Routing is nondeterministic between runs.