Limonchito working document, published as-is. Prices and lead times are estimates from September 2026 and change often.
Lemon rev B: workshop production process (500 lemons) and prototype procedure
This covers a batch of 525 lemons (500 + 5 % scrap) and 263 cradles in a small NL workshop.
- Mechanical reference: production/MECHANICAL_REVB.md, drawings/lemon_drawings.pdf (sheets 1–10; item numbers from sheet 8).
- Board test: production/TEST_FIXTURES.md.
- Electronics: production/ELECTRONICS_REVB.md.
Steps marked STOP are irreversible or gate the next station. Every station writes one line per unit (serial, operator, PASS/FAIL, reject code) into the production sheet.
Labour minutes are direct touch time per lemon. Machine and cure time is listed separately. Numbers are planning estimates; replace them with times measured on the first 20 units.
Workshop layout
┌──────────────── ventilated casting corner (extraction >= 300 m3/h, A2P3 respirators) ───────────────┐
│ C1 scale + mixing C2 vacuum chamber 19 L C3 pressure pot 40 L (3.5 bar) + compressor │
│ C4 warm cabinet 35-40 C (moulds, post-cure rack) C5 mould release booth C6 silicone mould shelf │
└───────────────────────────────────────────────▲─────────────────────────────────────────────────────┘
│ closed cores (S4) │ cast lemons (S6)
┌──────────┐ ┌─────────────────────┐ ┌──────┴────────┐ ┌─────────────▼──────┐ ┌───────────────┐
│ S0 goods │ → │ S1/S2 ESD bench: │ → │ S4 core bench │ │ S6/S7 trim + │ → │ S8/S9 final │ → S11 pack
│ in, IQC │ │ hand solder, 2 test │ │ UV lamp, jigs │ │ inspection table │ │ assy + test, │ (+ S10 cradle
│ shelves │ │ fixtures │ │ │ │ (daylight lamp) │ │ Qi pads, LTE │ bench)
└──────────┘ └─────────────────────┘ └───────▲───────┘ └────────────────────┘ └───────────────┘
│
S3 print room: 2 SLA printers, wash + cure, support removal, insert press
- Floor space: about 25 m². The casting corner needs its own extraction and a closed door.
- Batteries: cells are stored in a metal cabinet next to S8 and are never taken to the casting corner. Cells go into lemons only at S8.
Throughput plan (5 weeks, 2 operators)
| Week | Output |
|---|---|
| 1 | Setup, 20 first-article lemons through all stations, measure cycle times, make the 8 body moulds + 5 hatch moulds (6 h cure per half) |
| 2–5 | ~25–30 finished lemons/day |
- Operator A: S1, S2, S4 (≈ 14 min/lemon, ~30/day).
- Operator B: S3 and casting S5–S7 in the mornings.
- Afternoons: both on S8–S11.
| Station | Labour min/lemon | Machine / cure |
|---|---|---|
| S0 goods-in + IQC | 0.3 | – |
| S1 PCBA hand work | 7.0 | – |
| S2 board fixture test | 3.0 | 4.5 min/board, 2 fixtures |
| S3 printed parts | 2.0 | ~90 SLA plates or outsourced cores |
| S4 core close | 4.0 | UV 2 × 60 s |
| S5 mould load + pour | 1.5 | 2 h pressure cure |
| S6 demould + trim | 2.0 | 24 h rack |
| S7 hatch skins | 1.0 | 2 h |
| S8 final assembly | 1.5 | – |
| S9 final test + pairing | 3.5 | ~5 min, 4 in parallel |
| S10 cradle (1 per pair) | 1.5 | – |
| S11 pack | 2.0 | – |
| Total | ≈ 29 min |
- Total effort: ≈ 255 h direct + ~45 h batch overhead + ~13 h rework ≈ 315 h.
S0 Goods-in and incoming inspection
- Tools: calipers, sheet 9 (component photos), 10× loupe, scale, cell tester, cabinet for lithium cells.
- Check (sample 5 % or 5 pcs, whichever is more):
- PCBAs from JLCPCB: visual inspection. Check against the JLC X-ray report for U1.
- Carriers: panel V-score intact.
- Snaptron domes: count.
- LRA and Qi coil: part number, lead length.
- SIM cards: ICCID list matches the order.
- Cells: envelope ≤ 26.0 × 20.5 × 6.2 including the PCM fold, measure 10. OCV 3.6–3.9 V. Plug keyed with pin 1 = +. UN38.3 summary and battery DoC on file.
- Screws M2×5 T6, ruthex M2x4 inserts, PORON foam, 3M 850 film discs Ø14.
- Resin and PU lot numbers with SDS on file.
- Reject: wrong part number; a cell over the envelope; a cell without documents (quarantine the lot).
S1 PCBA hand work (ESD bench)
- Tools: soldering station, 90° carrier jig (holds the PCB flat and the carrier vertical at y = ±13.0), tweezers, twisted-pair jig, foam tape.
- Materials per lemon: 2 carriers, 2 domes + dome tape, 1 LRA, 1 Qi coil, 1 nano-SIM, 0.2 g solder.
- Break 2 carriers from the panel and deburr the edge.
- Put the PCB in the jig. Tack each carrier on its centre castellation, check it is square, then solder all 3 castellations per carrier. Carrier tilt ≤ 1° (jig gauge).
- Apply the domes with Snaptron tape, centred on the carrier pads.
- LRA: solder red → LRA_P and blue → LRA_N (sheet 10). Twist the leads and keep them under 20 mm. Put a foam pad on the LRA; it will sit at (−18.5, 0).
- Qi coil: solder the leads to QI_COIL1 / AC2, twisted, under 15 mm. The coil hangs below the board on its leads until S4.
- Punch the SIM to 4FF, insert it contacts down, close the hinged lid until it clicks.
- QC: joints shiny with no bridges; carrier square; dome centred ±0.3 mm.
- Reject/rework: carrier tilt > 1°, lifted castellation, dome off-centre.
- Time: 7.0 min.
S2 Board fixture test (TEST_FIXTURES.md)
- Tools: 2 pogo fixtures, host PC, PPK2, reference cell, Qi puck, fixture antenna.
- Steps: follow TEST_FIXTURES.md §3, steps 0–15.
- Covers: power-off checks, program, identity, battery sense, accelerometer, LED, domes, haptics, Qi charge, modem attach, provisioning, key check, lock + sleep current, APPROTECT, label.
- Label: the serial label goes on the PCB bottom beside J1. The serial is also printed onto the hatch-plate label at S8.
- QC: all steps PASS; the CSV row is closed.
- Reject: as the retest rule in TEST_FIXTURES.md says. Locked units cannot be reflashed without
recover, which erases the keys (re-provision). - Time: 4.5 min machine; one operator runs 2 fixtures, so 3.0 min labour.
S3 Printed parts (batch)
- Tools: 2 SLA printers (Form 4 class), IPA wash, UV cure, flush cutters, 400/800 grit, insert press (soldering iron with an M2 insert tip on a drill-press stand).
| Part | STL | Material | Per plate |
|---|---|---|---|
| Core top + bottom half | stl_core_* |
Tough 2000 / Siraya Blu (or outsourced JLC3DP MJF PA12 / SLA 8001) | 20 halves, on edge |
| Hatch plate | stl_hatch_plate |
Clear resin | 30 |
| Plunger | stl_squeeze_plunger |
Tough 2000 | 60 |
- Wash, cure and remove supports. Sand the parting faces flat on 400 grit on glass.
- Heat-press 2 ruthex M2x4 inserts into the core top-half bosses, flush to 0.1 mm below the ledge. Let them cool 10 s. STOP: a cracked boss means reject the half.
- MJF parts only: wipe the outside with thin UV resin and cure. This seals the porous surface for the film and the PU.
- QC: the hatch plate drops onto a golden core with an even seam and the hook engages; plunger slides freely in a Ø9.6 gauge; inserts perpendicular (M2 screw gauge).
- Reject: warp > 0.2 mm on the parting face, cracked boss, cloudy light rod.
- Time: 2.0 min/lemon (batch).
S4 Core close (sub-assembly for casting)
- Tools: core nest (holds the bottom half), UV torch (405 nm), click tester (spring probe, 20 N gauge), film applicator, IPA.
- Materials per lemon: core halves, 2 plungers, 2 film discs, 1 PORON foam 0.4 (on U1), UV resin 0.2 g.
- Put the coil flat on the coil pad in the bottom half, ferrite up, leads toward −X. Hold it with a 0.1 mm PSA dot on the core (not on a cell; no cell is present).
- Lower the board onto the 4 seat ribs. The carriers drop into the lower halves of the windows.
- Put a plunger in each window from inside, flange against the wall, tip toward the dome.
- Run a bead of printing resin on the bottom-half parting face. Lower the top half onto it; the hold-down tab clamps the +X tip of the PCB. Cure with UV 60 s per side, then wipe the seam.
- Press a PET disc over each window from outside. Rub it down with the applicator and make sure the edge is sealed.
- Click test (STOP): press each plunger through the film with the spring probe. You must feel a click on each side at 4–6 N probe force, and the plunger must return. There is no power yet, so this is mechanical only.
- Fit the protective cap: a printed dummy hatch, which also keeps dust out.
- QC: seam gap < 0.1 mm, no resin inside the windows, both clicks present.
- Reject: no click (re-open is impossible after bonding; the PCBA can be recovered by cutting the core, so budget 1 %).
- Time: 4.0 min.
S5 Mould load and pour (casting corner, batch of 4 moulds)
- Tools: 4 body moulds, 16 hatch plugs, 8 clamp plates + M5 bolts, scale (0.1 g), mixing cups, vacuum chamber, 40 L pressure pot, timer, warm cabinet.
- Materials per mould: ~42 g Polytek PT Flex 70 (A+B 1:1), pigment, Ease Release 200, green brush-coat PU (0.1 g per leaf).
- Before starting: PU pouring requires the EU diisocyanate training certificate.
- Take the moulds from the warm cabinet (35 °C). Two thin release coats, flash 5 min.
- Leaf: brush 0.1 g of green-pigmented PU into each leaf recess of the bottom half. Leave it 3–4 min to gel.
- Remove the dummy hatch from 4 closed cores and put a hatch plug on each: 2 pins into the insert holes, plug seated on the hatch ledge.
- Lay each core+plug into the bottom half; the plug flange rests on the top face. Close the top half on the keys. Clamp plates on, bolts snug. Stand the mould on its +X edge.
- Mix 84 g (2 moulds) for 45 s, vacuum 60 s. Pour each runner mouth in 20–30 s, until PU shows in the vent. Top up.
- Into the pressure pot within 4 min of mixing, 3.5 bar, 35 °C, 2 h. Repeat for moulds 3–4 with a fresh mix.
- QC: every vent shows PU; pot pressure logged.
- Reject: short shot (vent dry), which is found at S6.
- Time: 1.5 min per cavity; cycle 2 h 20 min per mould; 3 cycles/day.
S6 Demould, trim, inspect
- Tools: fresh scalpel blades, flush cutters, daylight lamp, 10× loupe, skin thickness gauge (spring caliper), go/no-go ring for the hatch outline.
- Depressurise and unclamp. Pull the plugs, open the mould, lift out the lemons.
- Cut the runner at the gate and the vent at the stem stub, flush to 0.2 mm.
- Trim the flash at the parting line and around the hatch opening.
- Inspect for voids > 0.3 mm in view, blocked dimples, seam flash > 0.2 mm, film wrinkle showing through, leaf fully green and bonded, stem stub complete.
- Rack for 24 h at room temperature before handling at S8. PT Flex 70 reaches full properties in 7 days, so allow 7 days before pack-out or drop testing.
- Reject: void on a squeeze pad or the Qi flat; skin < 1.2 mm at a squeeze pad (gauge); leaf lifting. A reject core can be recovered by cutting the skin off (PCBA saved).
- Time: 2.0 min.
S7 Hatch skins (casting corner, open moulds)
- Tools: 5 × 8-cavity hatch moulds, locating pins, scale.
- Release coat. Pour 1.8 g PU into each cavity, textured face down.
- Press a clear hatch plate onto the 2 locating pins, skin side down, until the flange seats.
- Cure 2 h at 35 °C. Demould and trim the overflow at the seam edge. Punch the 2 screw counterbores clean.
- Stick the label (CE ≥ 5 mm, bin symbol, model "Lemon B", serial blank, address) on the underside. Stick the PORON 1.5 pad on the press ribs.
- QC: no voids over the LED window; skin 0.8 mm over the rod (gauge).
- Reject: void over the window, plate not seated.
- Time: 1.0 min.
S8 Final assembly (cells enter here)
- Tools: T6 driver set to 0.10 Nm, ESD mat, cell cabinet.
- Materials per lemon: 1 cell, 1 hatch, 2 screws M2×5 T6.
- Check the serial label on the lemon matches the batch sheet, and write the serial on the hatch label.
- Plug the cell into J2 (keyed; it latches). Lay the cell in the bay, label up, lead folded toward J2.
- Hook the +X tongue under the rim, lower the hatch, drive 2 screws to 0.10 Nm.
- Squeeze both sides: the click haptic confirms power.
- QC: seam even; screws flush in the counterbores; haptic present.
- Reject: no haptic (check the plug, then the cell).
- Time: 1.5 min.
S9 Final test and pairing
- Tools: 4-slot test rack with Qi pucks (the production puck model), laptop on the backend admin view, 1 golden partner lemon.
- Put the lemon on a Qi slot and run the final test in TEST_FIXTURES.md: - charge tick within 5 s, backend shows charging; - accelerometer pick-up event; - LED breath visible through the window; - both domes register only together; - pulse round-trip to the golden lemon within 25 s.
- Pair: assign partner IDs in the backend (A/B of one box). Send a pulse A→B and B→A.
- Charge state for shipping: 30–60 % (UN3481 / PostNL).
- QC: all PASS; pairing logged; serial ↔ ICCID ↔ key row complete.
- Reject: fail → S8 rework (plug, hatch) or quarantine.
- Time: ~5 min, 4 in parallel → 3.5 min labour.
S10 Leaf cradle (263, one per pair)
- Tools: T6 / PH1 driver, cradle cover jig.
- Materials: leaf shell, cover, puck (Samsung EP-P2900 or the chosen one), foam spacer, 3 × M2×6 PT screws, 4 Bumpon feet.
- Press the puck into the pocket, logo down, cable toward the stem. The crush ribs centre it.
- Lay the cable in the underside channel and out through the stem groove.
- Put the foam spacer on the puck (stack to PUCK_H_MAX − puck height + 0.5). Fit the cover and 3 screws. Stick 4 feet on.
- Test with the golden lemon: charge tick within 5 s, no rattle, the lemon sits in the nest without rocking.
- Reject: puck loose, lemon not charging (swap the puck), cable pinched.
- Time: 3.0 min per cradle = 1.5 per lemon.
S11 Pack
- Box per pair: 2 lemons (30–60 % SoC), 1 cradle, manual card (includes the battery replacement instructions and the online URL, GPSR and CE text), insert.
- Outer carton labelled for UN3481 / SP188 where needed. Only 2 lemons with cells per PostNL parcel.
- Time: 4 min per box = 2.0 per lemon.
Appendix A: prototype procedure (one unit, R1/R2)
For the first 10–20 units, before moulds exist.
A1. Board 1. Bench bring-up on the nRF9151 DK and the 1NCE SIM as in rev A: attach, eDRX floor current on the PPK2 < 150 µA. 2. Board from JLCPCB. Do S1 by hand and S2 with a bench setup: DK as programmer, bench supply 3.9 V on TP8 via a 100 mA limit.
A2. Printed parts
1. Print one set: cores, hatch plate, 2 plungers, and 1 master stl_master_lemon_skin.stl, SLA 50 µm.
2. Sand the master to 600 grit, then acrylic coat + Inhibit X.
A3. One-cavity mould
1. Put the master in the parting board. Frame 4 walls × 1 cavity cut from stl_mould_box_frame.stl, or a cardboard/LEGO box.
2. Mold Star 30: half 1 (6 h), flip, half 2 (6 h).
A4. Core and fit checks 1. Do S4 by hand. 2. Dry-fit a cell (unplugged) and the hatch before casting: the cell lifts out, J2 is reachable, the rod sits over D1 (look through the clear plate), the hook engages. 3. Click check on both plungers.
A5. Cast and finish 1. Cast per S5 with a single mould. No pressure pot in R1 is acceptable, but expect bubbles in the dimples. Record the void count to justify the pot. 2. Cast the hatch skin in an open single-cavity mould. 3. S8 + S9 with a bench Qi pad and the backend.
A6. Measure and record - Trip force per side (target 12–20 N). - Skin thickness at the pads (cut sections of 1 scrap unit). - Qi charge current on 3 pucks. - LED visibility. - Drop test from 1 m onto carpet with the hatch closed. - Cell swap time for a first-time user (goal < 2 min).