LUME CAD Approval Pack - K11 T-Form Audit

LUME CAD Approval Pack - K11 T-Form Audit

Date: 2026-04-30

Superseded for the current hardware pass by `LUME_CURRENT_BUILD_SPEC.md`, which adds the ZHAOCAILIN top display, Arducam IMX586 mount, K11-era rear pod interface, and approval-v2 STL exports.

> Do not use this as the active print plan. It is retained only as an audit trail for the earlier K11 T-form review.

This is a repo-grounded audit of the current CAD, STL exports, and prearranged 3MF print plates. It corrects the earlier "31 parts" claim.

Current Design Intent

The active design is the K11 T-form architecture:

• Front bar: 500 x 120 x 85 mm rounded enclosure, printed as four shell halves.
• Rear pod: centered pod behind the bar for the GMKtec K11, K11 sled, cable bend, airflow, and VESA mount.
• Compute: K11 lives in the pod. Do not use the old Jetson AGX Orin sled path.
• Sensors: Orbbec Femto Mega/front camera window, two SVPRO brackets, UMA-8 mic rail, LED diffuser, cable backbone.
• Top display: 1920 x 440 strip display is represented, but dimensions are still placeholder and not approved.

Mockups To Review

Open these first:

• `renders/approval/approval_form_factor.png` - fresh T-form overview, top view, side/front views, pod exploded, and current top display cutout.
• `renders/approval/approval_shell_halves.png` - all four printed shell halves, exterior and interior.
• `renders/approval/approval_print_plates.png` - the seven prearranged 3MF plate thumbnails.
• `renders/approval/approval_small_parts.png` - active small parts plus flagged legacy/optional pieces.

Fresh render files generated from current CAD:

• `renders/approval/current_tform_3q.png`
• `renders/approval/current_tform_top.png`
• `renders/approval/current_shell_top.png`
• `renders/approval/current_pod_exploded.png`

Actual Printable Queue

The current prearranged 3MF queue contains 21 STL objects, not 31.

Part Dimensions From Current STLs

Approval Findings

Do not approve the current production print queue as-is.

1. Display cutout is not safe yet.

`DISPLAY_W=310`, `DISPLAY_H=72`, `DISPLAY_BEZEL=5`, plus clearance creates a `322 x 84 mm` top cutout on an `85 mm` deep bar. That leaves roughly `0.5 mm` front/back margin before the shell curve, far below the 3 mm wall target. If your physical display is truly that deep, the bar needs a redesigned raised display cradle, deeper body, or reduced bezel strategy. If the real display is smaller, update `lume-config.scad` before printing `shell_front_L` or `shell_front_R`.

2. K11 pod volume is plausible, but sled airflow support is incomplete.

The pod internal envelope is about `194 x 159 x 124 mm`; K11 is configured as `131.83 x 124.46 x 57.91 mm`, so raw volume clearance is acceptable. The problem is the K11 bottom fan: config says minimum bottom clearance is 12 mm, but `SLED_FLOOR_CLEAR=8`, and the printed sled itself has no physical riser/feet to hold that clearance. The preview floats the sled; the printed part does not enforce it.

3. `splice_plate.stl` is required but not in any prearranged 3MF.

The shell X-seams depend on CA glue plus splice plates. `exports/splice_plate.stl` exists, but no 3MF plate contains it. Print it separately or add it to Plate 02 before shell assembly.

4. The rear port bracket is Jetson-era and not queued.

`lume-port-bracket.scad` still labels Jetson-style HDMI/USB/LAN ports. In the K11 architecture, K11 ports live in the pod, not the bar. The `shell_rear_R` port opening should either become a clean umbilical/service hatch or be removed/fill-plated. Do not print that area assuming it matches the K11.

5. `internal_fan_shroud` and `internal_thermal_baffle` are legacy/optional.

They are still in Plate 02. That may be fine as sensor airflow/partitioning, but they are no longer compute cooling for a Jetson in the bar. Treat them as optional until the internal sensor layout is approved.

6. Several docs are stale.

`README.md`, `PRINT_PLAN.md`, `DIMENSIONAL_AUDIT.md`, and parts of the validation guide still reference Jetson-era dimensions or older part counts. Use this approval pack plus `print/print-queue.md` and `lume-config.scad` as the current truth until docs are cleaned up.

What Is Safe To Do Now

Safe:

• Open and inspect all mockup PNGs above.
• Open the 3MF files in OrcaSlicer for visual validation.
• Print Plate 01 coupons only, if you want ASA fit calibration.

Not safe yet:

• Printing shell front halves.
• Printing the K11 pod as final hardware before sled airflow/riser is fixed.
• Assuming Plate 02 contains every assembly part.
• Assuming rear bar ports match the K11.

Recommended Next CAD Fixes

1. Measure the physical display active area, total glass/frame width, total depth, thickness, and cable exit.
2. Fix the display cutout strategy in `lume-config.scad` / `lume-shell.scad`.
3. Add real standoffs/feet to `pod_compute_sled()` so K11 bottom intake clearance is physically enforced.
4. Add `splice_plate.stl` to the print queue.
5. Replace the Jetson-style rear port bracket with a K11-era pod/bar umbilical hatch.
6. Decide whether `fan_shroud` and `thermal_baffle` stay as sensor-only parts or get removed from the mandatory print batch.

Print Strategy After Fixes

You do not need to print 21 parts one by one unless you want maximum inspection. Normal flow is:

1. Print Plate 01 coupons.
2. Validate fit by hand.
3. Print the small parts batch, including the missing splice plate once added.
4. Print each large shell half alone.
5. Print pod parts.
6. Dry-fit everything before inserts, glue, and final assembly.