# Emerald-cut bracelet — design discussion

2026-10-11. **NOT FOR MANUFACTURE.** Original parametric CAD; no restricted third-party body files were downloaded, copied, adapted or redistributed. Proposed material: **Pt950 / 5% ruthenium**, subject to Elena's alloy/process approval. A spring tongue or hinge pin must not inherit that material specification without independent engineering.

## Recommendation

Discuss the **~1.75 ct** example with **8.18 × 5.71 × 3.84 mm** diamonds, long axes across the bracelet, independent four-corner protective baskets, and **0.40 mm nominal stone-edge gaps**. Do not force the approximate 215.9 mm target. Its **219.56 mm stone section** is 3.66 mm longer; the actual ruby parcel and final wrist/clasp measurements may change the preferred choice.

The pattern is fixed: **[diamond, diamond, diamond, ruby] × 9 = 27 lab diamonds + 9 lab rubies**. These are supplier-listed individual diamond examples, not a confirmed matched parcel. Ruby face-up dimensions are matching procurement targets. The wearer expressly authorized provisional ruby envelopes; actual ruby total depths remain unknown.

| Option | Diamond L × W × D, mm | Provisional ruby D, mm | Stone section at 0.40 mm gaps | Long-axis-along-chain alternative |
|---|---|---|---|---|
| Actual 1.50 ct | 7.59 × 5.38 × 3.71 | 4.20, assumed | 207.68 mm | 287.24 mm |
| Actual 1.75 ct | 8.18 × 5.71 × 3.84 | 4.40, assumed | 219.56 mm | 308.48 mm |
| Actual 2.00 ct | 8.74 × 5.81 × 4.05 | 4.60, assumed | 223.16 mm | 328.64 mm |

Stone section is **36 × width + 35 × gap**, measured between the first and last gemstone ends. Carat weight does not determine dimensions. Tables align within each option; separate diamond/ruby seats use their own depths. Ruby-depth sensitivity of ±0.40 mm is an independently chosen envelope, not a supplier tolerance or measured range. As the ruby remains deeper, its depth controls common setting height: nominal sole-to-table heights are **4.80 / 5.00 / 5.20 mm**. Different depths require CAD regeneration, not whole-setting scaling.

## Lengths and wrist reference

The closed schematic clasp has a **14.00 mm hinge-datum span**. Two terminal half-gaps add 0.40 mm to the stone-section-to-hinge-path accounting, so the reported **clasp contribution is 14.40 mm**, not box length plus tongue length. The assumed 6 mm tongue insertion overlaps the box and must not be added again. Closed hinge-path perimeters are **222.08 / 233.96 / 237.56 mm**.

The inner wearing perimeter is separately calculated by intersecting the inward-offset basket contact planes, including a bridged-gap/clasp envelope. It is approximately **208.22 / 220.10 / 223.70 mm** for the three options. This is a nominal rigid contact contour, not skin compression, gravity drape or a fitted wrist surface. Read exact values and contour coordinates in each validation report.

The wrist estimate **8 inches or slightly less / 203.2 mm** is not measured. The render uses an explicitly illustrative oval with assumed **1.28 width/thickness ratio**. Actual wrist circumference, width and front-to-back thickness at the wearing position remain required. Stone-section length minus wrist circumference is not wearing ease. No personalized fit or comfort claim is made.

**The small 1.50 ct option intersects the illustrative wrist at the clasp box by 0.27535 mm³.** Its positive perimeter difference does not guarantee clearance against this cross-section. The two larger options have zero calculated metal/wrist intersection in the chosen pose. Do not shrink the reference or increase the stone spacing silently to hide the small option's conflict; measure the actual wrist and choose/revise the layout deliberately.

## Settings, hinges and limitations

Each basket is independently constructed in millimeters, with four rounded corner posts, local stone-cut nominal seats, an open lower frame, a gallery and integral male/fork knuckles. Retention does not rely on a shared prong between moving links. A full bezel is a protection-led alternative but shows more perimeter metal; shared-prong, half-bezel and bar tradeoffs are in the [setting research](gems-and-settings.md).

The nominal hinge uses **0.60 mm pin diameter**, **0.68 mm bore diameter**, **0.08 mm axial gaps**, and **1.30 mm knuckle outer diameter**. These are study assumptions, not approved fabrication or wear tolerances. Pin heads depict a retained-pin envelope; riveting, welding, assembly access, wear and service procedures remain unresolved. Seats have nominal zero clearance to approximate gemstone proxies. Actual corner/girdle scans, corner relief, seat finishing, residual prong section, under-gallery smoothing and setter-controlled closing are required. A rigid translation interference check is not proof of retention strength.

The CAD solver places 36 unchanged link pitches and one unchanged clasp span into a closed oval polygon. Hinge endpoints coincide without stone/link stretching. The reports include neighboring solid intersections, actual joint angles, sampled reverse/forward bending, clasp-terminal checks, nominal seat/capture probes, illustrative-wrist intersection, and STEP round-trip validity/bounds/volume. Sampled collisions identify blocked poses, not approved articulation limits. No spring-force, fatigue, impact, opening-security or production setting qualification is claimed.

All six exported STEP assemblies round-tripped as **113 valid solids each**, preserving overall bounds within 0.00001 mm and total volume within 0.001 mm³. Displayed closed poses have maximum joint angles of approximately **13.59 / 13.63 / 13.64°** and no neighboring-link or terminal intersections. The sampled **25° and 30°** poses collide; some **−5° reverse bends** also collide. These are discrete geometric samples, not a certified travel range. The smallest checked pair clearance is approximately **0.04 mm at the pin/bore**, a deliberate nominal play assumption—not the 0.40 mm gem gap.

## Clasp: explicit original schematic

Amber geometry is an **unvalidated envelope schematic**, not supplier CAD. It shows a **12 × 8 × 4 mm assumed box**, a folded tongue, **two separate external safety loops**, pivots/studs and bracelet-end knuckles. Full extents, including external safeties, are reported separately. The **0.25 mm tongue-side clearance**, **0.25 mm carrier end gap** and **6 mm insertion overlap** are assumptions. The exploded view separates parts for inspection; it is **not an animation or proof of a viable opening sequence**. Spring action, internal locking detent, release actuator, safety-catch detents, opening sweeps and release/retention forces are not established.

[Stuller's double-safety box-clasp family](https://www.stuller.com/products/21346/100947?groupId=12517) is an inquiry lead, not an approved platinum assembly. Catalog wording alone does not confirm two external latch geometries or matched internal parts. [Clasp and license research](licenses-and-clasp.md) records the larger supplier candidate, alternative leads and precise missing specifications.

The supplied CGTrader files were **not inspected** because permissions for the AI-assisted workflow and public reusable derivative STEP/webmesh redistribution were not established. Their listing previews are not treated as dimensional evidence. Original construction avoids passing off restricted or unverified geometry as an inspected supplier mechanism.

## Deliverables and reproduction

Each option offers laid-out and articulated STEP solids, top/side SVG drawings, a validation report and an optional fit-gauge STL. The viewer displays the CAD tessellation and rigid placement matrices; it does not resize individual gems to improve appearance. Compare-all uses one orthographic camera/world scale. Setting/link, closed clasp and exploded-clasp views provide close-ups. Labeled PNG export records the actual current rendering, camera direction and major limitations. Gem optics, color and platinum finish are approximate.

The source ZIP contains exactly eight reviewed, repo-relative files: the original builder, drawing helper, parameters, pinned Python requirements and four factual bracelet research records. No private email, ring CAD or credentials are included. From the extracted source root:

```sh
python3.12 -m venv .venv
.venv/bin/pip install -r cad/requirements.txt
.venv/bin/python cad/bracelet/build_bracelet.py
```

Edit `cad/bracelet/parameters.json` and the sourced/provisional dimension records before rebuilding. The generator fails on its validity, connectivity, closure, intersection and round-trip requirements. Changing dimensions is not manufacturing approval. The repo's viewer commands are `npm run dev:jewelry`, `npm run build:jewelry` and `npm run preview:jewelry`; they are isolated from the ring build. `deploy/package_jewelry_release.py` packages an explicit public-file allowlist.

### Optional fit gauge

The STL is **two rigid continuous envelope halves**, not an articulated jewelry mesh. It removes separate stones, seats, all small hinge features and the working-clasp concept. Width is constant gemstone length + 0.80 mm; thickness follows the tallest stone envelope. **Wider clasp/latch protrusions are omitted.** This is a simplification for assessing gross inner contour and bulk, not exact clasp fit, drape or comfort.

Import in **millimeters at 100% scale**, with slicer auto-fit disabled. Two halves are separated by 16 mm in the export; move them independently on the build plate as needed, then bring the planar cut faces together to reconstruct the nominal inner contour. Orient the broad side on the print bed and inspect support/elephant-foot effects. Measure the print against CAD before using it as a gauge. Do not force a rigid gauge over the hand or wear it as finished jewelry. SVG drawings use consistent comparison scales but are not calibrated 1:1 print templates.

## Questions for Elena

1. Measure snug wrist circumference, width and thickness; agree ease and drape before choosing the final inner perimeter.
2. Can you quote all three targets as **27 matched diamonds plus 9 matched red lab rubies**, with per-stone L/W/D, corner/girdle geometry, color/cut matching, actual quantities and delivery? No evidence establishes that 2 ct is cheaper.
3. Approve protective independent baskets versus bezels, finished prong/seat sections, under-gallery edge finishing, Pt950/Ru process and allowable gap/hinge travel.
4. Provide a matched box/tongue/two-external-safety assembly's SKU, dimensioned open/closed drawings, connection span, insertion/release clearances, material/temper and validated retention/release/fatigue data. Confirm permission before sharing supplier CAD publicly.
5. Retain 219.56 mm stone section or revise the targets after the actual parcel, wrist and clasp are known? Do not hide conflicts by stretching stones or forcing impractical gaps.

No stones were purchased or reserved, and no jeweler or supplier was contacted.