E01 independent nominal geometry and exported-file review

Two detailed digital assemblies now exist: the ND64 acoustic head (67 parts)
and the captive manual gravity cartridge (42 parts). These are unbuilt
mechanical references. Neither is a physically validated fracture device or a
complete calibrated computer measurement system.

Current independent results
- cad-v3/CAD-INDEPENDENT-REVIEW.json: 182 nominal geometry checks, zero failed
  checks. Source hashes stayed unchanged during the review.
- exports-v2-review-v1/CAD-EXPORT-INDEPENDENT-REVIEW.json: 296 exported-file
  checks, zero failed checks. Source/output hashes remained unchanged.
- Exact final export directory reviewed: generated-v2. The initial generated-v1
  export and prior reviewer records are retained as development history.

The geometry review does not call the author's collision checker or tests.
It checks kernel validity and one connected solid per part, its own pairwise
intersections, actual manufacturer-pattern baffle holes, eighteen assumed 3mm
diameter axial tool approaches, twenty-nine sampled impact travel positions,
release-pin support and withdrawal, positive tip/bushing retention and upper
and lower overtravel negative controls. Nominal positive-volume collisions
are absent in the reviewed configurations. These are discrete nominal checks,
not a proof over every tolerance and every continuous motion.

The export review independently reimports 111 individual STEP files and four
whole-assembly STEP files. It checks the two functional assemblies plus a
process-fit coupon and driver-pattern gauge. All 21 custom/template/metal-part
STL meshes have closed consistently oriented edges and no degenerate faces at
the declared 0.00001mm weld scale. The largest STL-versus-kernel volume
difference is approximately 0.0334%, reflecting tessellation approximation.

Part categories matter
Acoustic head: 5 custom parts, 2 gasket templates, 60 purchased nominal reference
instances (including the driver, fasteners and grommet).
Impact cartridge: 6 custom parts, 2 gasket templates, 4 custom machined parts,
30 purchased nominal reference instances.
Two further custom pieces are the fit coupon and driver-pattern gauge.
The STL count includes templates and machined-part reference geometry. It does
not mean every STL should be printed. Purchased reference shapes are not
production drawings and do not establish procurement compatibility.

Physics correspondence
The impact CAD z axis is opposite the numerical downward x axis. Nominal tip
release 85mm, upper limit90mm, target0, pad engagement-2mm and hard capture-5mm
correspond exactly to the declared dynamics datums. The upper and lower
constraints are geometrical retention limits. A trajectory crossing the hard
model boundary is a failed dynamics case, not a predicted hardware stop force.

The original enclosure's 6mm bosses only touched the wall, producing multiple
disconnected solids. The corrected7mm bosses overlap the wall and reduce the
Boolean air volume to0.7039780004475063L. The original E01 model used0.7060914L.
Use the separately audited corrected-volume composite for current geometry;
the older numerical files remain valid for their explicitly historical volume.

Repairs and retained evidence
Development found and corrected a foot rib rotated along the wrong dimension,
an incomplete communicating-air domain, disconnected mounting bosses, rod/pad
interference, a front cover/base overlap and one collapsed raw tip-mesh pole
facet. The mesh fix only welds at0.00001mm and removes collapsed facets; it does
not fill holes or change the CAD physics. The original raw generation remains
retained. The per-pixel depth renderer also replaces a drawing method that
showed hidden internal triangles over outer surfaces. Current assembled PNGs
were visually inspected; they are CAD illustrations, not photographs.

The initial independent cad-v1 review failed only because exporter/test sources
changed while it was checking. Its geometric results were otherwise clear.
cad-v2 rechecked stable geometry, and cad-v3 binds the final exporter-era source
snapshot. These failed/superseded records are included rather than overwritten.

Open engineering gates
The assembled moving group must be weighed. Using the explicitly assumed
material densities, the CAD envelope estimate is12.2047g, not exactly the12g
nominal dynamics case. Sleeves made of a different material change it. Actual
mass, guide/bushing tolerances, shaft alignment, static breakaway force, pin
drag, wear and jamming have not been measured. The model's smooth friction law
does not establish whether a real slider releases under its own weight.

Frame, keeper, retainer, fastener and pad strength are unqualified. A connected
solid and a closed mesh are not a load or fragment-guard rating. Buffer/contact
stiffness and damping have not been derived from a printed part or measured.
The reference cantilever fixture and target-plane registration are separate
unfinished mechanisms. Manual reset/release is implemented geometrically;
automatic release, repetition and programmed BPM are not.

For the acoustic head, actual flange thickness, terminal/cable envelope, gasket
compression, leakage, fastener torque, enclosure compliance and connector fit
remain unmeasured. The grommet dimensions are assumptions. There is no complete
verified no-solder cable/sensor/supply purchase list or current under$1000 quote.
Use the separate electrical-interfaces document to distinguish the proposed
wired chain from connections that have actually been verified.

No acoustic focusing, material fracture, perfect split, large-specimen scaling,
physical efficiency, thermal protection, novelty or patentability is proved by
the geometry or numerical consistency checks in this package.
