Dimensional audit — LUME v2 components
_Every dimension I've cited in the session, re-verified. Confidence level flagged on each. ⭐ = verified against authoritative source (manufacturer spec sheet, official STEP file, physical measurement). 🟡 = quoted from memory or estimate, needs verification. ❌ = known wrong or contradicted._
Full Public Reader
Dimensional audit — LUME v2 components
_Every dimension I've cited in the session, re-verified. Confidence level
flagged on each. ⭐ = verified against authoritative source (manufacturer
spec sheet, official STEP file, physical measurement). 🟡 = quoted from
memory or estimate, needs verification. ❌ = known wrong or contradicted._
Cameras
Orbbec Femto Mega (we own one, Mac4)
| Dim | Value | Source | Confidence |
|---|---|---|---|
| Width (X) | 127 mm | `lume-config.scad` FEMTO_W=127, plus official Orbbec STEP imported in `lume-components.scad` (USE_REAL_FEMTO=true) | ⭐ |
| Depth (Y) | 65 mm | FEMTO_D=65, STEP-verified bbox X[-0.6,114.6] = 115mm narrow axis but the cited 65mm is the front-face short dim | 🟡 — need to recheck which axis is depth |
| Height (Z) | 38 mm | FEMTO_H=38, STEP bbox Z[-39.6,0.6] = 40mm | ⭐ rounded |
| Rear hump (Jetson Nano) | +25 mm | Memory file `lume-v1-femto-bolt-vs-mega-decision.md`. Nano carrier sticks back beyond main 65mm body | 🟡 — measure on physical Mac4 unit |
| Mount pattern | 100×30 (X×Z) | `lume-config.scad` FEMTO_MOUNT_PITCH_X=100, FEMTO_MOUNT_PITCH_Z=30 | ⭐ |
| Cable | RJ45 (PoE) + barrel | Official Orbbec spec | ⭐ |
| Mic array | UMA-7 onboard | Orbbec spec | ⭐ but obsoleted by LUMF audio sidecar |
Y-axis ambiguity: The STEP bbox shows X[-0.6,114.6] meaning the Mega is
~115mm along its NATURAL X-axis as exported. The 127mm/65mm/38mm we use in
SCAD may have been swapped — Orbbec specs the Mega as 105×80×35mm in some
data sheets. Action item: physically measure Mac4's Mega with calipers
and reconcile. Until then, treat 127×65×38 as a 5-15
is conservative for cradle clearance.
Orbbec Femto Bolt (just ordered, Apr 27 arrival)
| Dim | Value | Source | Confidence |
|---|---|---|---|
| Width (X) | 140 mm | Memory `lume-v1-femto-bolt-vs-mega-decision.md` | 🟡 — verify on arrival |
| Depth (Y) | 39 mm | Same memory | 🟡 |
| Height (Z) | 30 mm | Same memory | 🟡 |
| Mount pattern | TBD | Not in any committed file | ❌ — unknown |
| Cable | USB-C only (bus power + data) | Orbbec product page | ⭐ |
| Mic array | None | Orbbec product page | ⭐ |
| Mass | TBD (probably <250g) | Not specced | 🟡 |
Action item: when Bolt arrives, capture official STEP from Orbbec's
hardware repo (github.com/orbbec/OrbbecHardware), drop into
`hardware/cad/references/orbbec-femto-bolt.stp`, measure with calipers
to cross-check.
Sensor sensor parity (the part that matters)
| Spec | Mega | Bolt |
|---|---|---|
| ToF chip | Sony IMX556 | Sony IMX556 |
| Depth resolution | 1024² NFOV / 320×288 WFOV @ 30fps | 1024² NFOV / 320×288 WFOV @ 30fps |
| RGB | 4K @ 30fps | 4K @ 30fps |
| Depth range | 0.25 - 5.46 m | 0.25 - 5.46 m |
| FOV (NFOV) | 65° × 65° | 65° × 65° |
| FOV (WFOV) | 120° × 120° | 120° × 120° |
| IMU | Yes (BMI270) | Yes (BMI270) |
⭐ Identical sensor silicon. The image quality and depth fidelity are the same.
The differences are exclusively form factor + connectivity + onboard compute.
Compute
GMKtec NucBox K11 (ordered, not yet onsite)
| Dim | Value | Source | Confidence |
|---|---|---|---|
| Width (X) | 142.6 mm | Memory `lume-v1-bom-pivot-beelink.md` cited from GMKtec Amazon listing | 🟡 |
| Depth (Y) | 113 mm | Same | 🟡 |
| Height (Z) | 49.5 mm | Same | 🟡 |
| Mount | VESA 75 (4× M3 underside) | GMKtec product page | ⭐ |
| Power | 19V DC barrel, 65W max | GMKtec spec | ⭐ |
| Ports | 4× USB-A + 2× USB4-C + HDMI2.1 + DP2.1 + 2× 2.5G | Same | ⭐ |
| Mass | ~520 g | Estimate from class | 🟡 |
Action item: when K11 arrives, capture exact dimensions with calipers
and reconcile with `K11_W/K11_D/K11_H` constants in lume-config.scad.
Bar enclosure (current v1)
| Dim | Value | Source | Confidence |
|---|---|---|---|
| Width (X) | 500 mm | LUME_WIDTH | ⭐ committed |
| Height (Z) | 120 mm | LUME_HEIGHT | ⭐ committed |
| Depth (Y) | 85 mm | LUME_DEPTH | ⭐ committed |
| Wall thickness | 3 mm | WALL | ⭐ |
| Internal cavity | 494 × 114 × 79 mm | derived | ⭐ |
| Split plane | Y=0 (centered front/rear) | SPLIT_Y=0 | ⭐ |
| X-split for plate fit | X=80 mm | SPLIT_X=80 | ⭐ committed |
| Wordmark | "LUME" 16mm Helvetica Bold, 1.5mm deboss | WORDMARK_* | ⭐ |
| Femto cradle position | front center, Y=-LUME_DEPTH/2+3 | hardcoded in `lume-shell.scad` | ⭐ |
Pod (T-form, v1, just shipped)
| Dim | Value | Source | Confidence |
|---|---|---|---|
| Width (X) | 200 mm | POD_W | ⭐ committed |
| Height (Z) | 120 mm | POD_H = LUME_HEIGHT | ⭐ |
| Depth (Y) | 130 mm | POD_D | ⭐ |
| Wall | 3 mm | POD_WALL | ⭐ |
| Internal cavity | 194 × 114 × 124 mm | derived | ⭐ |
| Bar→pod offset | centered behind bar (POD_X_OFF=0) | ⭐ | |
| Fillet transition | 12mm chamfer | POD_FILLET_R | ⭐ |
| VESA pattern | 100×100 on pod back | POD_VESA_PITCH | ⭐ |
Cross-checks
K11 in pod (single-sensor config)
K11 142.6 × 113 × 49.5 in pod cavity 194 × 114 × 124. K11_D 113 vs cavity
height 114 = 1mm clearance (⚠ tight on Z). K11_W 142.6 vs cavity width 194
= 25.7mm slack each side. K11_H 49.5 vs cavity depth 124 = 74.5mm slack
behind for cables + Mega Jetson hump (25mm).
⚠ Risk: the 1mm Z clearance on K11_D is too tight after print warp +
seam gap. Recommend bumping POD_H from 120 to 130mm OR rotating K11 90°
on its side. Rotated 90°, K11_D 113 goes along bar X (no constraint),
K11_W 142.6 goes along Z (vs 114 cavity = ❌ doesn't fit). So Z-bump is
the answer. Action item: change POD_H to 130mm in v2 SCAD.
Two sensors total volume
| Sensor mode | Required cavity (mm³) | Where |
|---|---|---|
| Mega only | 127×65×38 + hump | front bar, current v1 design |
| Bolt only | 140×39×30 | front bar, fits with 50mm depth slack — could shrink bar to 60mm |
| Both in same bar | 267×min(65,39)×max(38,30) ≈ 267×39×38 if side-by-side | requires bar width >280mm — fits in 500mm |
| Both same bar wide-angle | each at 30° outward from center | needs 350-400mm horizontal span — fits |
Two sensors fit in the existing 500mm bar laterally with 100mm of design
slack. The Y-axis depth depends on which orientation. Both in landscape
(thick face forward) takes 65mm (Mega) — within current 79mm internal.
Both in portrait would push past current depth.
Print volume
Neptune 4 Max plate: 420×420×480mm. Largest single print: shell_front_L
at 330×42.5×120mm — well within plate. Pod halves are 200×130×60mm — also
within. Adding a second sensor cradle inside the same bar adds maybe one
~80×80×40mm internal mount — trivial. No print-volume concern for any
two-sensor config.
Open dimensional questions
1. Mega axis labels — the FEMTO_W/D/H constants may have X/Y axes
swapped relative to Orbbec's published spec. Verify with calipers on
Mac4's unit before committing v2 SCAD. ETA: 5 minutes once you have
physical access.
2. Bolt dimensions — only memory-quoted, never STEP-verified.
Verify on arrival with calipers + STEP import.
3. Bolt mount pattern — completely unknown. Without it we can't design
the Bolt cradle. STEP file from Orbbec's hardware repo will reveal.
4. K11 actual footprint — cited from Amazon listing image, not measured.
Verify on arrival.
5. Mega Jetson hump exact protrusion — cited as +25mm but could be
anywhere between 18 and 32mm depending on how it's measured. Verify.
6. K11 fan clearance requirement — vendors typically spec 50mm intake
clearance for sustained 65W loads. Our pod_compute_sled has 8mm floor
gap — recheck whether that's intake or exhaust side.
Promotion Decision
Attach run IDs, datasets, metrics, and reproduction commands.
Source Anchor
lume-commerce/hardware/design/v2-twin-sensor/DIMENSIONAL-AUDIT.md
Detected Structure
Evaluation · References · Architecture