feat(airframe): Rev T1 — the tilt spar becomes a structural member of the wing (both halves)

.gitignore

@@ -7,6 +7,26 @@ __pycache__/
 *.pyc
 *.pyo
 
+# Python virtual environments (never commit).  The local .venv here is 351 MB of
+# third-party packages; a single `git add -A` would push the lot.  Nothing under
+# it has ever been tracked, so this is a guard rather than a cleanup.
+#
+# It also matters to CI: `.github/workflows/ci.yml` runs `flake8` with an
+# explicit --exclude list that does NOT name .venv, so lint only passes today
+# because CI checks out a fresh tree with no virtualenv in it.  Were one ever
+# committed, the lint job would immediately fail on thousands of findings in
+# vendored code.  (Locally, `flake8 .` does report those — run it over
+# `git ls-files '*.py'` instead to see only this repo's own code.)
+#
+# NOTE for anyone running the tools: this .venv SHADOWS system Python packages
+# the repo's tools rely on (manifold3d, sympy, matplotlib are installed
+# system-wide).  Invoke tools as `/usr/bin/python3 tools/<name>.py`, not through
+# the venv.
+.venv/
+venv/
+env/
+ENV/
+
 # Claude Code agent worktrees and session data
 .claude/
 

Browncoats-workspace.code-workspace

@@ -20,6 +20,9 @@
 		},
 		{
 			"path": "../Tactical-WX-RX"
+		},
+		{
+			"path": "../engineering-pe-skills"
 		}
 	],
 	"settings": {

REFERENCES.md

@@ -293,6 +293,65 @@ aft white); controlled by FC4 node (Simon's medbay, Bay D).
 
 ---
 
+### REF-FAA-004: 14 CFR Part 23 — Airworthiness Standards: Normal Category Airplanes
+
+| Field | Value |
+|---|---|
+| **Issuing authority** | U.S. Federal Aviation Administration (FAA), Department of Transportation |
+| **Current edition** | Part 23 as restructured by Amdt. 23-64 (effective 2017-08-30), performance-based; as amended through the 2024 annual CFR edition |
+| **Official URL** | <https://www.ecfr.gov/current/title-14/chapter-I/subchapter-C/part-23> |
+| **Verified source** | Section text quoted below was retrieved and verified verbatim 2026-08-29 from GovInfo's authoritative CFR XML — <https://www.govinfo.gov/content/pkg/CFR-2024-title14-vol1/xml/CFR-2024-title14-vol1-sec23-2230.xml> and <https://www.govinfo.gov/content/pkg/CFR-2024-title14-vol1/xml/CFR-2024-title14-vol1-sec23-2265.xml>. (eCFR's HTML endpoint bot-blocks automated retrieval; GovInfo serves the same authority.) |
+| **Applicability caveat** | Part 23 governs **manned** normal-category airplanes. Serenity is an sUAS operated under Part 107 [REF-FAA-002], which imposes **no** structural certification basis. Part 23 is cited here as an **adopted engineering baseline**, exactly as `docs/structural_analysis.md` §3 already frames it — **NOT as a compliance claim.** |
+
+**Sections applied in this project:**
+
+| Section | Title | Applied where |
+|---|---|---|
+| **§ 23.2230** | *Limit and ultimate loads* | The 1.5× ultimate/limit factor used throughout `docs/structural_analysis.md` §3 and `tools/wing_spar_carrythrough.py` |
+| **§ 23.2265** | *Special factors of safety* | Justifies an additional factor on FDM-printed structure — see below |
+
+**§ 23.2230 (verbatim, GovInfo 2024 CFR):** the applicant must determine —
+(a) *"The limit loads, which are equal to the structural design loads unless
+otherwise specified elsewhere in this part"*; and (b) *"The ultimate loads,
+which are equal to the limit loads multiplied by a 1.5 factor of safety unless
+otherwise specified elsewhere in this part."*
+
+**§ 23.2265 (verbatim, GovInfo 2024 CFR):** (a) *"The applicant must determine a
+special factor of safety for each critical design value for each part, article,
+or assembly for which that critical design value is uncertain, and for each
+part, article, or assembly that is —"* … (a)(2) *"Subject to appreciable
+variability because of uncertainties in manufacturing processes or inspection
+methods."* (c) *"The applicant must multiply the highest pertinent special
+factor of safety in the design for each part of the structure by each limit and
+ultimate load…"*
+
+**Why §23.2265 matters here.** Serenity's primary structure is FDM-printed
+CF-PETG. Layer adhesion, raster orientation, moisture uptake, and
+machine-to-machine variation are precisely the *"appreciable variability because
+of uncertainties in manufacturing processes"* that (a)(2) contemplates. This is
+the regulatory hook for the FOS 4.0 joint target that `docs/structural_analysis.md`
+§3 currently describes only as *"a design-team judgment value"* — the judgment
+is sound and now has a citable basis, though the specific numeric value remains
+the project's own choice rather than anything Part 23 prescribes.
+
+**CORRECTION THIS ENTRY MAKES — stale section number.**
+`docs/structural_analysis.md` cited **"14 CFR Part 23.303"** in two places. That
+section belonged to the **pre-2017** Part 23 and does not exist in the current
+rule; the 2017 restructure moved the limit/ultimate relationship to **§23.2230**.
+This is the same class of defect this file already recorded for §23.1401 (see
+"Removed / Superseded Citations"). Both occurrences corrected 2026-08-29.
+
+**Also worth recording, to prevent a future fabrication:** the load factors
+**+3.8 / −1.52**, widely quoted as "the Part 23 load factors," appear **nowhere
+in the current rule** — they belong to the pre-2017 §23.337. Current §23.2200(b)
+requires only *"design maneuvering load factors not less than those, which
+service history shows, may occur within the structural design envelope."* This
+project's 4 g limit factor is an adopted figure under that framing, not a
+regulatory quotation. Do not cite +3.8/−1.52 to current Part 23.
+
+**Used in:** `docs/structural_analysis.md` §2, §3;
+`docs/WING_ATTACH_INTERFACE.md` §2.1; `tools/wing_spar_carrythrough.py`
+
 ## Part II — United States Federal Communications Commission Regulations
 
 > "Can't stop the signal." — Mr. Universe. We can, however, stay inside Part 15/95 limits while we transmit it.
@@ -1185,6 +1244,47 @@ FEA cross-check (ANSYS Workbench, linear-elastic model, Table 11): simulated fle
 
 **Used in:** `airframe/README.md` §5.1
 
+### REF-MATH-001: Melissen, J.B.M. — "Packing and Covering with Circles"
+
+| Field | Value |
+|---|---|
+| **Author** | Johannes Bernardus Marinus Melissen, Utrecht University |
+| **Publication** | PhD thesis, Universiteit Utrecht, 1997 |
+| **Official URL** | <https://dspace.library.uu.nl/handle/1874/25091> — Utrecht University Repository, open access |
+| **Retrieved** | 2026-08-29 |
+| **Scope applied** | The proven-optimal ratios for packing *n* equal circles inside the smallest enclosing circle, for *n* ≤ 7. Only these small-*n* cases are used; they are proven optimal in this source and are reproduced throughout the standard circle-packing literature. |
+
+**Values used (enclosing diameter ÷ single-circle diameter):**
+
+| *n* | Ratio K(*n*) | Closed form |
+|---|---|---|
+| 1 | 1.00000 | 1 |
+| 2 | 2.00000 | 2 |
+| 3 | 2.15470 | 1 + 2/√3 |
+| 4 | **2.41421** | **1 + √2** |
+| 5 | 2.70130 | 1 + √(2(1 + 1/√5)) |
+| 6 | 3.00000 | 3 |
+| 7 | 3.00000 | 3 (hexagonal + centre) |
+
+**Why this is catalogued rather than treated as common knowledge:** the *n* = 4
+case decided a load-bearing geometry change. Two independent sources — the
+owner's `docs/plans/2026-08-27-nacelle-wiring-plan.md` and the external Gemini
+conversation it derived from — both specified an 11.0 mm spar bore for four
+10 AWG conductors. Four Ø5.5 mm circles circumscribe **13.28 mm** at zero
+clearance, so 11.0 mm does not fit at all, and the error propagated because the
+constant was carried in prose rather than computed. It is now computed, from
+this table, in `tools/spar_bundle_fit.py`, which refuses to interpolate for any
+*n* outside the tabulated set.
+
+**Caveat on the input, not the ratio:** the ratio is exact; the 5.5 mm wire OD
+it is applied to is **not** a verified figure. `current-specification/bom_revS.csv`
+records no outside diameter for `WIRE-10AWG`. See `tools/spar_bundle_fit.py`,
+which prints that caveat on every run.
+
+**Used in:** `tools/spar_bundle_fit.py`;
+`airframe/openscad/wings/wings_s1223_revo.scad` (`SPAR_BORE_OD` derivation);
+`docs/WING_ATTACH_INTERFACE.md` §2
+
 ## Part XI — FDA / CDRH Laser Product Regulations
 
 ### REF-FDA-001: 21 CFR Part 1040 — Performance Standards for Light-Emitting Products
@@ -1706,10 +1806,20 @@ Per-application operating mode differs even though the hardware is now identical
 | Application | Qty | Mode | Range |
 |---|---|---|---|
 | Cargo winch | 1 | Continuous rotation, gateway closes position on the AK7455 spool encoder (REF-SENSOR-008) | Multi-turn, unbounded by the servo itself |
-| Nacelle tilt | 2 | Position, firmware soft-limited | −5°…140°, backstopped by the existing CF-PETG hard-stop blocks in the external gear train (`docs/NOZZLE_DRIVE_TRADE.md`) — mechanically independent of the servo's own (now-removed) rotation pin |
-
-Removing the pin on all three is a deliberate commonality choice, not a requirement of the
-position-mode applications: LibreServo replaces the servo's potentiometer with a 360°
+| Nacelle tilt | 2 | **CORRECTED Rev T1c (2026-08-30):** continuous rotation, Pilot closes position on the AK7455 nacelle encoder (REF-SENSOR-008) — the same pattern as the winch above | **Multi-turn** (1.438 rev of actuator per 145° of nacelle), unbounded by the servo itself. Soft-limited to −5°…140° *at the nacelle*, backstopped by CF-PETG hard stops in the external gear train |
+
+> **The row above changed at Rev T1c and the change is structural, not editorial.**
+> The tilt stage became a REDUCTION (tip 14T/50T, `i` = 3.571), so the drive shaft must
+> turn **1.438 revolutions** over the 145° sweep and no limited-rotation servo can drive
+> it at any horn radius. Removing the rotation pin is therefore **required** on the tilt
+> servos now, not merely convenient — and the AK7455 became **load-bearing for control**
+> rather than telemetry, because a multi-turn drive without absolute feedback does not
+> know where the nacelle is. See `docs/WING_ATTACH_INTERFACE.md` §4.3b/§4.3c and
+> `docs/TILT_DRIVE_CONTROL_SPEC.md`.
+
+Removing the pin on the winch is a deliberate commonality choice; on the tilt servos it
+is now a requirement (see the note above). It is not a requirement of the
+position-mode applications generally: LibreServo replaces the servo's potentiometer with a 360°
 absolute magnetic encoder (AEAT-8800, 16-bit), so position feedback and soft-limit
 enforcement no longer depend on the mechanical stop the pin used to provide. See
 REF-SENSOR-013/014/015 below for the individual part records, and

TODO.md

@@ -35,6 +35,104 @@
 - [ ] Verify AK7455 off-axis geometry + pinout vs datasheet (REF-SENSOR-*)
 - [ ] ASTM D3039/D695 coupon test: CF-PLATE-2MM bending allowable (thwarts
       built at FOS 8.5/8.7 vs the conservative 300 MPa stand-in only)
+- [ ] **ASTM D3039/D695 certificate for the 20 x 16.3 CF SPAR TUBE.** New at
+      Rev T1 and now load-bearing: the spar is the wing's primary bending
+      member at FOS 9.0 against the same unverified 300 MPa stand-in
+      (`docs/TILT_SPAR_ANALYSIS.md` §3.6.3). The 4130 rows above no longer
+      govern the flight load path.
+- [ ] **Measure the procured 10 AWG silicone wire OD.** `bom_revS.csv` records
+      none; the 5.5 mm figure the Ø20.4 spar bore is derived from is an
+      ASSUMPTION. A larger real OD re-opens the airfoil trade
+      (`tools/spar_bundle_fit.py` prints this caveat on every run).
+- [x] **Tilt servo angular range (180 vs 270 deg) — CLOSED 2026-08-29, VOID.**
+      Owner direction: the drive turns the shaft MORE THAN ONE REVOLUTION, so
+      the stage is a reduction and the actuator's own travel no longer sets the
+      ratio. Built: module 0.8, 14T pinion / 50T ring, i 3.571, shaft 1.389 rev
+      per 140 deg, C 25.6 -> station 53.6.
+- [ ] **ACTUATOR RE-SELECT (replaces the item above).** A multi-turn output means
+      the tilt drive is no longer a limited-rotation servo but a
+      continuous-rotation gearmotor or stepper, closed on the AK7455's absolute
+      nacelle angle. The DS3225 is ~17x oversized on torque AND now the wrong
+      kind of device. NOTE this makes the encoder load-bearing for CONTROL, not
+      telemetry: a multi-turn drive without absolute feedback does not know
+      where the nacelle is.
+- [ ] **LG-11 coupon — DEMOTED at the wing root, still a gate elsewhere.**
+      Owner ruled the cargo bay clear, so the root joint was re-designed around
+      it: socket for shear + bonded 80x60 flange for the moment, FOS 29.2 at the
+      STANDING 5 MPa figure (29.2 / 87.6 / 274.6 at 5 / 15 / 47 MPa). The coupon
+      no longer decides whether that joint is buildable, only how small the
+      flange could shrink. It remains a gate for the tenon path and the thwarts.
+- [ ] **Aero revalidation of the re-lofted wing (CFD or bench).** The section
+      is no longer S1223 -- root t/c 12.14 -> 17.72 %, tip 18.93 -> 26.70 %.
+      Every aero figure in this repo citing this wing is unverified until this
+      lands, including the 7.6 N cruise-lift figure and everything derived
+      from it. The built sections are designated S1223/t17.7 (root) and
+      S1223/t26.7 (tip) -- S1223's CAMBER LINE with the thickness envelope
+      scaled x1.46 / x2.20; camber is unscaled and exact (8.67% at 49.0%
+      chord), so zero-lift angle and lift-curve slope partially survive at the
+      ROOT on thin-airfoil grounds and not at all at the tip. CL_max, L/D and
+      the cruise-lift figure do NOT survive. Needs XFOIL or a
+      TRANSITION-SENSITIVE RANS run at Re 1.3-1.8e5 -- a fully-turbulent model
+      will misrepresent the separation bubble. `tools/wing_cfd_openfoam.py` is
+      still blocked on mesh generation. See docs/flight_envelope.md's banner.
+
+#### 0.8.1 — Wing attach interface (Rev T1), open requirements
+
+→ detail: `docs/WING_ATTACH_INTERFACE.md` §5; fuselage detail:
+`airframe/fuselage-mid/WBS.md` §1.1.1.5
+
+Wing side is BUILT. These are the two joints it publishes.
+
+- [x] WA-R1/R1b, R2, R4, R6 — fuselage side BUILT 2026-08-30 (Rev T1c). Socket
+      Y +21.00 / Z +66.85 / D20.4 / 18.5 mm; F688ZZ and both tie rods deleted;
+      mortise 30.8 -> 12.8 (tenon fits +0.40 mm/side); nav D4.2, AK7455 D7.5 and
+      shaft D4.4 conduits cut. `wing_root_deconflict.py` now CLEAR.
+- [x] WA-R15 — tilt actuator: DS3225 + LibreServo_v4 run MULTI-TURN, closed on
+      the AK7455; fuselage spur stage m0.8 38T/38T, C 30.40, 1:1. Coaxial does
+      not fit and no gear clears the tenon in Y -> 18 mm actuator standoff.
+- [ ] WA-R3/R17 — split-collar pinch clamp is still not a part (no SCAD, no STL,
+      no BOM row). **BLOCKS wing removal/refit.**
+- [ ] WA-R16 — the tilt train is NOT self-locking and has no holding provision.
+      **BLOCKS flight release.** `docs/TILT_DRIVE_CONTROL_SPEC.md` §5.2/§7.3.
+- [ ] WA-R18 — Rev T1c is +102.8 g (+2.63 % AUW, revised by the weight audit);
+      hover T/W was ~1.19 vs a 1.2 minimum. Re-derive mass/CG/T-W.
+      **BLOCKS flight release.**
+- [ ] MA-1 — BOM printed-part masses understate by +521.6 g (13.3 % of AUW)
+      across 23 rows measured against their STLs. Reconcile + add a CI check.
+      **BLOCKS any weight statement.** -> `docs/MASS_AUDIT_CARGO_WING_ROOT.md`
+- [ ] MA-5 — hollow the actuator standoffs, -32.9 g, no structural question
+      (deflection 1.4e-5 mm at the gear mesh against a 0.05 mm budget).
+- [ ] MA-6 — `PRINT-BATT-TRAY` measures 140.2 g against a 22 g BOM row.
+- [ ] MA-7 — BOM mass column mixes installed mass, stock and GCS; add an
+      `Installed` flag before any weight statement uses it.
+- [ ] W1..W8 — ranked weight-reduction targets: battery -225 g, tray floor
+      -23.1 g, actuator standoffs -32.9 g, cradle ~-25 g; wing root CLOSED.
+      W2 (BOM reconciliation) gates all of them.
+      -> `docs/plans/2026-08-30-001-weight-reduction-targets-plan.md`
+- [ ] WA-R15a — re-measure the cargo-bay roof band: the actuators now reach
+      X -158.5 / -221.5 at Z +85.99..+112.99.
+- [ ] TILT-CTL-01..06 — tilt control loop open items (plant model, slew rate,
+      differential-tilt trip, LibreServo_v4 encoder part).
+      → `docs/TILT_DRIVE_CONTROL_SPEC.md` §8.
+- [ ] WA-R7..R12 — nacelle: trunnion bearing bore D20.0 at ring plane X ~ 28
+      (axial AND radial duty), ring gear PD 33.8 at C 26.0, ring magnet
+      ID 27 / OD 41 axially separated from that gear, the 4 x 10 AWG
+      disconnect relocated into the nacelle annulus, the nav 3-core crossing
+      at the trunnion, and confirmation of the 32 mm spar-stub protrusion.
+- [ ] WA-R13 — avionics: `docs/TILT_ENCODER_WIRING_EMI_SPEC.md` §6.1's
+      ferromagnetic-spar premise is corrected in place; confirm the firmware
+      zero-calibration procedure still covers the drive-shaft/pinion field
+      that replaces it.
+- [x] **OWNER DECISION — cargo-bay intrusion: CLOSED 2026-08-29.** The bay stays
+      clear; nothing goes inboard of hull X ~ -86 against a bay edge at -100.
+      The constraint produced a better joint (moment FOS 4.02 -> 29.2), not a
+      compromise. See `airframe/fuselage-mid/WBS.md` §1.1.1.5.
+- [ ] **Spar stub / trunnion packaging — NEW, tightest constraint on the joint.**
+      The spar must terminate >= 26 mm from the nacelle duct axis, so the stub is
+      15.0 mm (max 15.7). The trunnion bearing PAIR, the 50T ring gear and the
+      ID26/OD41.2 ring magnet must all fit inside that 15 mm, and the magnet and
+      gear are nearly coradial (r 13.0-20.6 vs 20.0) so they need axial
+      separation. 2x 6804 (20x32x7) = 14.0 mm fits; MF128ZZ retired to QTY 0.
 
 ### 0.10 Update and correct documentation touching every non-archived file.