#keyboards #lily58 #soldering #reflow #smd #solder-paste
# Overview
Method notes for populating the SMD side of the **Lily58 Pro** — 58× 1N4148W SOD-123 diodes and 58× Kailh MX hotswap sockets. Parts sourcing lives in [[Lily58 Pro Build Shopping List]]; this note is only about *how* they go on.
Short answer: both a reflow oven and a hot air station work for both part types. The oven is the better tool for bulk population; hot air is for rework. The one non-obvious risk is on the sockets, and it's a solder-flow problem rather than a heat problem.
# Design
## What can be reflowed
| Part | Reflow oven | Hot air | Notes |
|---|---|---|---|
| 1N4148W **SOD-123** diode | ✅ Ideal | ✅ | Two pads, no heat-sensitive plastic. Nothing to worry about. |
| Kailh MX hotswap socket | ✅ | ✅ with care | Housing is high-temp nylon — survives a normal profile, but with less margin than a typical SMD part. This is how factory hotswap boards are made. |
## Oven vs hot air
Oven, clearly, for the initial population. 58 diodes + 58 sockets across both halves is ~116 individual operations by hot air, and SOD-123s are light enough that airflow scatters them if the nozzle isn't well controlled.
Hot air is the right tool for reworking a single bad joint later, not for building the board.
## Order of operations
1. Diodes and sockets sit on the **same face** (the side opposite switch insertion) — one reflow pass covers both. ==Verify against the build guide for the PCB revision actually ordered before pasting; Lily58 PCBs are reversible and variants differ.==
2. Reflow.
3. Inspect, touch up.
4. Hand-solder all through-hole parts **after** the board is cool: TRRS jack, reset tact switch, OLED headers, Mill-Max controller sockets.
# Configuration
## Solder paste
**Chip Quik TS391AX — $15.95, 5cc/15g syringe**
| Spec | Value |
|---|---|
| Alloy | Sn63/Pb37 |
| Melting point | 183 °C (361 °F) |
| Particle size | T4 (20–38 µm) |
| Metal content | 88% by weight |
| Flux | Synthetic no-clean, ROL0 |
| Shelf life | **>12 months refrigerated *or* unrefrigerated** |
| Stencil life | >12 hr @ 20–50% RH |
Two reasons this one over the more commonly recommended SMD291AX: no refrigeration required, and because it's stored at operating temperature there's **no 4-hour warm-up** before use. Open the drawer and dispense.
T4 is the right powder for syringe work — finer than T3 so it flows through a small tapered tip and holds dot definition, without the faster oxidation of T5/T6.
Leaded rather than lead-free is deliberate: 183 °C peak instead of ~245 °C gives real headroom against the socket housings.
⚠️ Buy dispensing tips separately — the tip in the syringe is too coarse for per-pad dots. Want **tapered plastic, 0.4–0.6 mm ID**. Chip Quik sells a [30-pack of assorted Luer-lock tips](http://www.chipquik.com/store/product_info.php?products_id=906002).
Skip the 10cc/35g syringes. 15 g is far more than 116 pads needs and the excess ages out first.
Price and stock verified **2026-08-18**.
## No-stencil application
A stencil isn't needed here — stencils earn their keep on fine-pitch ICs, not two-terminal parts.
1. Paste at room temperature.
2. One small dot per pad, roughly the footprint of the pad, barely proud of the surface.
3. **Equal volume on both pads matters more than the absolute amount** — uneven paste is what causes tombstoning on small two-terminal parts.
4. Place with curved ESD tweezers. Paste is tacky enough to hold parts en route to the oven; don't jostle.
Run **ten diodes as a test batch** before committing to all 58. Paste volume is learned in one pass.
A hotplate is worth considering over the oven for a first attempt — flat single-sided board, and you watch the paste go grey-and-grainy → shiny-and-collapsed instead of trusting a profile blind.
# Troubleshooting
## Solder wicking into the socket contact
The #1 socket failure mode, and it's not heat. Too much paste and solder flows past the pad into the leaf contact — socket then won't accept a switch pin, or grips badly. Keep paste modest, strictly on the pads, nothing near the switch-pin opening.
This is the same class of failure already logged in [[Lily58 Pro Build Shopping List#Hot-swap socket sourcing]]: *"Every failure mode on record is installation, not manufacture."*
## Diode polarity
SOD-123 has a cathode band. One diode backwards = a dead key found much later with a multimeter. Surface tension during reflow can rotate a part slightly but will **not** flip one — orientation is entirely on placement.
Check the band against the silkscreen as each part goes down, then do a full visual sweep before the board sees heat.
## Switch-pin through-holes
⚠️ Check whether the PCB has switch-pin through-holes immediately adjacent to the socket pads. Keep paste well clear — solder bridging into a pin hole is tedious to clear.
# Decisions
- **Oven over hot air for population.** ~116 operations by hand vs one pass; SOD-123s blow away under airflow.
- **Leaded paste over lead-free.** The ~30 °C lower peak is the entire safety margin for the socket housings.
- **TS391AX over SMD291AX.** Same alloy, same metal content, a dollar cheaper, and it removes both the fridge and the warm-up wait. For one board plus a year of sporadic use, storage behaviour beats any performance difference.
- **No stencil.** Two-pad parts only; syringe dosing is adequate and a stencil is an extra order and an extra failure point.
- **Fallback worth taking seriously:** reflow the diodes, hand-solder the sockets. Socket pads are large and easy, it's ~40 minutes, and it sidesteps the wicking problem entirely. Reasonable hedge on a first reflow.
# TODO
- ==Confirm which face the diodes and sockets populate on the ordered PCB revision== before applying any paste.
- ==Decide oven vs hotplate.== Hotplate gives visual feedback on a flat single-sided board and is the lower-risk first attempt.
- ==Pull the Kailh socket datasheet for a peak-temperature number== rather than relying on the general "high-temp nylon survives a normal profile" assumption.
- ==Order tapered dispensing tips with the paste== — the included tip is unusable for per-pad dots.
- ==Run a 10-diode test batch and record the working paste volume here== before the full pass.
# References
- [Chip Quik TS391AX product page](http://www.chipquik.com/store/product_info.php?products_id=470011) — specs and pricing above
- [TS391AX datasheet (PDF)](https://www.chipquik.com/datasheets/TS391AX.pdf)
- [Chip Quik No-Fridge paste line](https://www.chipquik.com/store/index.php?cPath=470)
- [Lily58 Pro build guide (EN)](https://github.com/kata0510/Lily58/blob/master/Pro/Doc/buildguide_en.md)
---
- [[Lily58 Pro Build Shopping List]] · [[Lily58 PCB Revision Comparison]] · [[Desoldering with HAKKO FR-301]]