# Glitchwave 567 — PCBWay ordering guide (rev 0.2, 2026-07-29)

> **✅ HOLD CLEARED (2026-07-29).** The enclosure fit conflict (pot knobs

> buried under PC-pin stomp mounting) is fixed by swapping RV1–RV6 to

> ALPS RK09K1130A70 (30mm shaft, same footprint, LCSC C351173) — BOM-only

> change, no board rework. See `hardware/ENCLOSURE_FIT.md`. Cleared to

> order.

> **♻️ FAB PACKAGE REGENERATED 2026-08-02.** The gerbers/drill/pos in this

> folder were last built 2026-07-25 and had gone **stale** — four rounds of

> board changes since then (input-jack move, output-jack clearance fix,

> control-board stack header moved to B.Cu, MP1584 exposed pad connected)

> were **not** in them. Anything ordered from the old package would have

> built the July board. Everything except the drill template is now

> regenerated from the current `.kicad_pcb` files by

> `hardware/review/tools/make_fab.ps1` — **re-run that script after any

> board edit, and check the file dates here against the `.kicad_pcb` dates

> before you order.**

> **⛔ DO NOT USE THE ENCLOSURE DRILL TEMPLATE.** `drill_template_1to1.pdf`

> is dated 2026-07-25 and is **wrong**: the input jack J1 moved +3.5 mm in

> x on 2026-07-31 to clear the buck inductor. The file has been renamed to

> `drill_template_1to1.STALE_2026-07-25_DO_NOT_USE.pdf`. A fresh template

> must be produced before any enclosure is drilled. This does not block the

> PCB order — it is not a fab deliverable.

> **✅ PRE-ORDER BLOCKERS CLEARED 2026-08-02.** Four things were resolved:

> 1. **Stomp path: PC-pin variant.** No board cutouts are needed — ENCLOSURE_FIT's

> two 27 × 14.5 mm cutouts were only required on the wire-lead path.

> 2. **CONTROL board now has its four M3 standoff holes** at (17, 9.5), (121, 9.5),

> (17, 108.5), (121, 108.5) — Ø3.2 mm NPTH, ≥3.69 mm to the nearest copper and

> ≥5.5 mm to the board edge. ENCLOSURE_FIT required these ("control-board retention

> must come from standoffs") and they had never been added. MAIN board deliberately

> has none: six jack bushings with nuts through the walls retain it.

> 3. **Board-to-board connector chosen** — see the assembly notes below.

> 4. **BOM is complete.** Every line on both sheets now carries a real, orderable

> part number; the 64 lines that said `generic` were backfilled.

Two boards, one enclosure (Hammond 1590XX). Order both as **4-layer, full

turnkey assembly**. Gerbers/drill/pos live next to this file; BOMs are the

sheets of `hardware/BOM.xlsx` (MAIN = 297 lines, CONTROL = 22 lines).

## Board specs (both boards identical stackup)

| Parameter | Value |

|---|---|

| Layers | 4 — F.Cu route / In1 solid GND plane / In2 route / B.Cu route |

| Size | 138 × 114 mm, corner notches 11 × 11 mm (in Edge.Cuts) |

| Thickness | 1.6 mm, 1 oz copper all layers |

| Min track / clearance used | 0.25 mm / 0.13 mm (PCBWay standard 5/5 mil capable) |

| Vias | 0.5 mm pad / 0.3 mm drill, min hole-to-hole 0.25 mm |

| Finish | HASL lead-free is fine (ENIG optional) |

| Mask / silk | Any color Jason likes; silk clipping over pads is expected |

DRC status (KiCad 10, **re-run 2026-08-02 against the current boards**):

**0 unconnected, 0 shorts, 0 clearance violations, 0 schematic-parity errors**

on both boards. MAIN = 72 warnings + 4 courtyard errors; CONTROL = 6 warnings,

0 errors. The 4 courtyard errors are the same shallow decoupling-cap kisses

(≤0.4 mm, bodies verified clear) — no action needed.

## Turnkey sourcing notes

1. **LCSC parts** — every BOM line carries an LCSC C-number (stock verified

  2026-07-25; pot line updated 2026-07-29). Pots: ALPS RK09K1130A70

  (C351173) ×6, B10k linear, snap-in, 30mm shaft.

2. **Stomp switches SW1/SW2 — mark DNP, do NOT ask PCBWay to source them.**

  Suntsu SSWFS-S01-AC09-HWH is a Digi-Key *Marketplace* listing, Mouser has

  no listing for it or its wire variant, and has no foot-switch

  category at all** — no fab-side distributor stocks this class of part.

  It is a commodity pedal-industry switch: Amplified Parts **P-H604**

  publishes matching dimensions (12 mm × 1 mm bushing, 14.3 mm long, body

  25.3 × 12.6 mm) against this footprint's Ø12.2 hole and 25.5 × 13 body,

  and Love My Switches stocks PCB-mount ($3.99) and pre-wired ($5.75)

  versions. Buy them direct and fit them at final assembly. This takes the

  only unsourceable line off the fab order's critical path.

  ⚠️ Two numbers still unverified: the below-panel body depth (ENCLOSURE_FIT

  assumes 15.0 ± 1.0 mm — measure a real one before cutting metal), and the

  footprint `descr` says thread "15/32-32" while ENCLOSURE_FIT and Amplified

  Parts both say M12 × 1.0. The Ø12.2 clearance hole suits either; the nut

  does not.

3. **Raspberry Pi Pico (U20)** — the official Pico module, SMD-mounted flat

  (castellated edge pads, official land pattern). PCBWay can source it or

  Jason can consign two.

4. **DNP parts** — LFIL/OFIL cap pads near the LM567 are intentionally

  unpopulated ("the voice" of the pedal). They are marked DNP in the BOM.

  Do not let the fab "helpfully" populate them.

## Assembly notes

- CONTROL board: pots, stomps, and WS2812 LEDs mount on F.Cu; the 2×8 keyed

 stack header (**ref is J1 on the control board**) mounts on **B.Cu** — this

 is now true in the board file as of 2026-08-02; until then it was drawn on

 F.Cu, where it could not have mated. It uses a project-specific

 **pre-mirrored** footprint, `PinHeader_2x08_P2.54mm_Vertical_Mirrored`,

 which cancels KiCad's flip so pin N lands on pin N of J10 below. Verified

 16/16 pads and 16/16 nets against J10. **Do not substitute the stock

 PinHeader_2x08 footprint on this part** — flipping the stock one permutes

 the pin numbers (1↔15, 2↔16, 3↔13, …) and reverses the whole connector.

 Same header on MAIN (J10) mounts topside at (96, 104) — only DC nets cross.

 **The pair is now chosen and gives exactly the 11.00 mm gap the enclosure needs:**

 | Board | Half | Part | LCSC | Key dimension |

 |---|---|---|---|---|

 | MAIN, J10, **top** side | female socket | XFCN **PM254V-12-16-H85** | C46595985 | body **8.50 ±0.15 mm** |

 | CONTROL, J1, **bottom** side | male header | XFCN **PZ254V-12-16P** | C492425 | insulator **2.50 mm**, mating pin 6.0 ±0.20 mm |

 2.50 + 8.50 = **11.00 mm**, plastic face to plastic face — a positive mechanical

 stop, not a nominal. Worst-case tolerance 10.70–11.30 mm. The 6 mm pin sits inside

 the socket's 3.68–6.35 mm insertion-depth window and does **not** bottom out, which

 is what makes the plastics the datum. Verified against both manufacturers' drawings;

 the two datasheets name each other as the mate.

 ⚠️ **Do not substitute a long-pin or "stacking" header** (e.g. 13.5 mm pins). It

 would exceed the socket's maximum insertion depth, bottom out internally, hold the

 plastics apart and leave the gap indeterminate.

 ⚠️ **The socket goes on MAIN deliberately.** MAIN carries the DC jack, so its half is

 live whenever power is applied — recessed socket contacts can't be shorted by a

 dropped screw or a probe, 16 exposed pins at 18 V can. It also puts the lighter half

 on the board that hangs upside down.

- MAIN board: wall-mounted jacks (2× PJ-603A, 3× PJ-3410, 1× DC-044A) are

 right-angle parts hanging off board edges — verify they sit flush before

 wave/hand solder.

- First-article checklist is the README sheet inside `BOM.xlsx` — run it

 before boxing the assembled boards.

## Suggested order

Qty 5 of each board, turnkey, lead-free HASL, any-color mask.

The enclosure drill template is **not** in this package any more — the old one

was quarantined as stale (see the warning at the top of this file). It was never

a fab deliverable, only a 1:1 print for Jason's drill press, so its absence does

not affect the PCB order. A new one must be generated before any metal is cut.