The Phonograph — Stax Edition I

This is an industrial-design specification, not an essay. The format is thesis → form → electronics → firmware → reproducibility → manufacturing → pricing → capsule → roadmap → limitations → footnotes → status. Every empirical claim is footnoted to a primary source or marked as an engineering estimate. Numbered sections. Equal-width columns. Figures called out by name (the actual render files ship as the Edition moves out of concept stage; this document is the load-bearing primitive that the renders, prototypes, and first-article inspections will be checked against).

The Phonograph is Stax Edition I in the drop-house charter¹. It is the first physical artifact in the /objects surface of the stax.dev portfolio² and the first instance of the four-component capsule format (object + essay + software + audio) that defines a Stax Edition.

1. Thesis — why this object exists

The Sonos S2 app is hated. It is hated by people who paid four-figure sums for Sonos hardware in the expectation that the controller would last as long as the speakers, and who watched the controller's user experience degrade across three forced rewrites while their amplifiers kept working³. The hate is a symptom of a broader pattern: the touchscreen-on-everything monoculture has lost the plot. Every interaction in the modern home — lights, thermostat, audio, lock, oven — has been collapsed onto the same rectangle of glass, behind the same animation system, mediated by the same notification queue. The result is that the act of turning down the music in the room you are currently sitting in has become a five-tap excavation through a phone that may not have been unlocked since the last time you turned down the music.

The Phonograph is a refusal of that pattern. It is a single-purpose, tactile, BLE-paired remote with one knob and three buttons. It does one job well: control whatever audio system is playing in the room you are currently in. It does not have a screen in Edition I. It does not have a microphone. It does not have a brand-controlled cloud. It composes with the Stax NATS substrate and the multi-room music client (project #1 in the Sovereign Stack portfolio), which means the Phonograph's button-and-knob events are open-ended NATS messages that the owner can rebind to any subject they want.

The object is a piece of the appliance layer in the language of the Mercantile Thesis⁴: a sovereign, vertically integrated, locally-owned control surface that owns the path from the owner's hand to the speaker in the room, without per-use rent flowing back to a central operator. Owning a Phonograph is not subscribing to a Phonograph service. The firmware is AGPL-3.0⁵; the schematics are public; the supply chain is SBOM-audited⁶; once you own it, you own it forever, and if Stax disappears, the next owner can rebuild it from the public repository.

That is the substrate claim. The rest of this document specifies how it is built.

2. Industrial design — the form

Figure 1 — Phonograph body, top elevation. Render to ship as prototype boards arrive.

Body. A right cylinder of approximately Ø60 mm × 28 mm height — hockey-puck silhouette, sized to live on a side table next to a couch or on a desk between a notebook and a coffee mug without claiming visual primacy. Footprint smaller than a standard coaster. The material is 6061-T6 aluminum⁷, machined from billet (not cast) for tolerance discipline at the encoder shaft and the PCB-seat shoulder.

Color options for Edition I. Two anodized finishes, both matte: charcoal and bone. Fifty units per color, one hundred total. Both finishes are Type II anodize⁸, dyed in color-matched lots to avoid the production-run drift that plagues short-run anodizers. No glossy finishes; the Phonograph is a tool, not jewelry. The charter rules out any aesthetic that reads as decorative rather than functional¹.

Top face. One large tactile knob (Ø24 mm), centered. Three programmable buttons in an arc below the knob, evenly spaced on a ~38 mm chord. Default bindings:

1. Button 1 — next track in the active zone
2. Button 2 — pause / resume
3. Button 3 — scene toggle (cycles user-defined scenes)

The bindings are owner-rebindable in firmware (§4). The default assignment is conservative: the two most-pressed buttons on any audio remote in history are next-track and pause; "scene" is the Stax contribution and defaults to a no-op until the owner defines a scene.

Bottom face. Rubber foot ring (silicone, color-matched to the body finish, ~0.8 mm thick) running the perimeter. Inside the foot ring, laser-etched:

Stax Edition I — №042 of 100
charcoal · 2026

The serial number is monotonic across the run; the color and the year are etched on the same line. An optional owner-engraving line beneath ("for L.M.K., 2026" — or any string the owner specifies at purchase) runs the same laser pass, in the same typeface, at no additional cost. Engraved owner-marks are part of the Anordain-and-Standards-Manual discipline the charter borrows¹: the object is signed by both maker and owner. The bottom face is also where the OOB pairing QR code is printed (see §4).

Knob mechanics. Continuous rotary encoder with 24 detents per revolution, push-to-click. The chosen part is the ALPS RE160 series (EC11-style) incremental encoder with switch⁹, specifically the variant with the longest service life (≥100,000 rotations) and the firmest detent feel in the family. The detent feel is the single most important tactile property of the entire device. The reference points are explicit:

• The Behringer X32 channel-strip rotary encoder. Loud, sure, mechanical.
• The Sennheiser HD650 cup-mount detent. Smaller, but the same kind of

committed click.

Anything softer than those — Logitech mouse-wheel feel, generic consumer-electronics detents — is rejected. The push-to-click action fires at ~2.5 N axial force, well above accidental-grazing threshold and well below thumb-fatigue threshold. The custom CNC aluminum knob cap (Ø24 mm, matte-anodized to match body color) press-fits onto the ALPS shaft with a steel inner sleeve for the mechanical interface, because aluminum-on-aluminum press-fits gall over time and the steel sleeve eliminates the failure mode.

Button feel. Silent tactile switches with 0.9 N actuation force and ~0.35 mm travel¹⁰. The chosen part is the Omron B3F-1000-series through-hole tactile switch (or the ALPS SKHHAJ in a surface-mount footprint, depending on which PCB assembler is selected — both have substantially identical tactile properties). The Cherry MX Brown was explicitly rejected: it is too loud for a living-room object and the long travel is wrong for a thumb-press remote, where the Cherry's keyboard-typing-optimized travel reads as mushy. The buttons have keycaps machined from the same 6061-T6 billet as the body, with a 0.4 mm proud face above the top surface — enough for a fingernail-blind pressure to find them, not so much that they catch when the device is picked up.

Surface treatment. Type II anodize at ~25 µm thickness for both colors. The bottom face is laser-etched after anodize, leaving the raw aluminum exposed in the etched characters — high contrast against the dyed surface, and the etch is permanent (no wear-off ink).

3. Electronics — the substrate

Figure 2 — Phonograph PCB, top side. KiCad render to ship from the public repo.

Microcontroller. ESP32-C6¹¹. RISC-V single-core at 160 MHz, with BLE 5.3, Wi-Fi 6 (802.11ax), and 802.15.4 (Thread / Matter) on-die radios. The C6 is the specific Espressif part for three named reasons:

1. BLE 5.3 is the current Bluetooth Core Specification for low-energy

peripherals¹², including the Out-of-Band pairing primitive the Phonograph uses (§4). The C3 ships with BLE 5.0, which works but loses the LE-Secure-Connections-Numeric-Comparison improvements that land in 5.2 and 5.3.

1. 802.15.4 capability headroom for future Thread / Matter

participation. Edition I does not use Matter; Edition I-bis or Mark II may.

1. RISC-V matches the sovereign-stack ISA thesis. The Sovereign

Stack work at large prefers RISC-V over ARM where the choice is neutral on other axes, and the ESP32-C6 is the most mature Espressif RISC-V part with the best Zig toolchain story.

The S3 was considered (more RAM, more peripherals) and rejected: it is Xtensa, not RISC-V, and the Phonograph does not need the S3's extra compute.

PCB. 4-layer FR4, 28 mm diameter (matched to the body internal bore), 1.0 mm thickness. Layer stackup: signal — ground — power — signal, with a continuous ground pour under the BLE antenna keep-out zone. USB Type-C connector on the side of the PCB exits through a slot in the aluminum body; used for firmware flash, BLE-pairing-key provisioning at the factory, and the occasional charge cycle if the owner opts for a rechargeable LIR2032 instead of the CR2032 primary. Coin-cell socket on the back of the PCB, accepting either CR2032 or the dimensionally identical LIR2032 rechargeable.

Battery. CR2032 primary lithium cell, 220 mAh nominal at 3.0 V¹⁴. Calculated service life: ~14 months at typical use (defined below), with deep sleep current target <10 µA and active current target <30 mA during a BLE event (each event lasts ~50 ms — knob turn, button press, ack roundtrip)¹⁵. Both current targets are engineering estimates against the C6 datasheet and the BLE link-layer's worst-case 50 ms wake budget; field verification ships with the Edition I run and feeds back into the spec.

The math: 220 mAh × 0.85 efficiency factor (cold-temperature derating and self-discharge over a year) ≈ 187 mAh usable. At 10 events/day of 50 ms each at 30 mA, the active-state daily consumption is 10 × 0.05 / 3600 × 30 = ~0.004 mAh. The sleep-state daily consumption dominates: at <10 µA × 24 h = <0.24 mAh/day. Total ~0.244 mAh/day. 187 mAh / 0.244 ≈ 766 days ceiling under ideal conditions; field estimate adjusts to ~14 months to account for Wi-Fi reconnect cycles, occasional firmware updates, and BLE link supervisory overhead. Heavy-use households (50+ events/day, more reconnects) should expect ~8 months; this is named honestly in §10.

Display. Optional. Edition I ships without a display. Edition I-bis (planned reprint, see §9) ships with the Sharp LS013B7DH03 1.28" memory display¹⁶, a transflective monochrome panel chosen specifically for sub-microamp idle current and zero backlight. The memory display reflects ambient light like e-paper but updates at 60 Hz like an LCD — the right tradeoff for a knob-event feedback surface that needs to show "zone: kitchen, volume: 42%" within a single knob-rotation gesture without spinning the battery.

Antenna. PCB trace antenna, inverted-F geometry, tuned for 2.4 GHz BLE. Range target: ≥10 m through one interior wall under normal residential drywall + stud construction¹⁷. Stone walls or steel-framed buildings degrade this to ≤5 m, named in §10. An external chip antenna footprint is laid out on the PCB as a fallback should the trace antenna underperform on the bench; the chip antenna is not populated in Edition I unless first-article inspection requires it.

Cost target. BOM ≤ $35 per unit (parts only, at 100-unit volume), against a retail of $280 — preserving healthy capsule margin that funds the editorial + audio components per charter §5¹. Major BOM lines (engineering estimate, refined as manufacturing quotes return):

| Component | Est. cost (USD) at qty 100 | | -----------------------------— | -------------------------— | | ESP32-C6 module | $3.80 | | ALPS RE160 encoder + switch | $2.10 | | 3 × Omron B3F tactile switches | $0.60 | | FR4 PCB (4-layer, 28 mm) | $1.50 | | CR2032 holder + cell | $0.45 | | USB-C connector | $0.40 | | Passives (caps, resistors, LDO) | $1.20 | | Aluminum body (CNC + anodize) | $18.00 | | Knob cap + steel sleeve | $3.50 | | Rubber foot ring | $0.80 | | Laser etch + QR engraving | $1.20 | | Packaging insert + paper card | $1.45 | | Total | ~$35.00 |

The aluminum body is the dominant line item, as expected for any CNC object at this volume. Doubling the run to 200 units in Edition I-bis or Mark II is the obvious lever to drop the body cost to ~$12 — but Edition I commits to 100 per charter §5, and the math closes at $35.

4. Firmware — the software substrate

Zig + Janet, single-binary, AGPL-3.0⁵. Runs on the ESP32-C6 in riscv32-freestanding-none target mode.

Build. zig build cross-compiles the firmware against the ESP32-C6 HAL and the Espressif IDF C headers (vendored, hash-pinned via build.zig.zon). The build emits a single signed .bin for OTA flashing plus a debug ELF for the JTAG path during bench bring-up. The build inherits the discipline from the existing ESP32 firmware substrate at <internal-lab>/firmware/esp32-rugged-node-v0/¹⁸ — specifically the binary framing convention (magic + version + kind + sequence_id + CRC32) is reused for the NATS-over-BLE event format below.

Janet REPL. A subset of [Janet][janet] is compiled into the firmware ROM as the user-extensibility surface. Owners who want to rebind buttons or compose custom scene logic write Janet, not C. Example:

(on-button 1
  (publish "lab.sonos.zone.next-track" {:zone "kitchen"}))

(on-button 3
  (let [scene (cycle :morning :work :evening)]
    (publish "lab.scene.activate" {:scene scene})))

The Janet snippet is uploaded via USB-C and stored in a dedicated partition; firmware OTA updates do not overwrite it. This is the firmware equivalent of an Emacs .init.el — your customizations survive every update.

[janet]: https://janet-lang.org/

NATS-over-BLE. The Phonograph publishes button and knob events as NATS messages, tunneled over a BLE GATT characteristic to a paired Mac/Linux host running the bridge. The subject pattern is:

lab.phonograph.<serial>.event

with the event payload encoded as JSON:

{
  "ts": 1747200000.123,
  "kind": "knob.rotate",
  "delta": +3,
  "modifier": null
}

Knob rotation emits knob.rotate with delta in detent counts. Button presses emit button.press with the button index (1, 2, 3). Button + knob combos (button-held during rotation) emit with the modifier field set to the held button's index. The host acknowledges each message via:

lab.phonograph.<serial>.ack

within ≤200 ms, otherwise the Phonograph re-emits up to three times before backing off. The serial-number qualifier in the subject means multiple Phonographs in the same household are addressable independently — one per major room.

Pairing. BLE Secure Connections with Out-of-Band (OOB) pairing key¹³. The OOB key is encoded into a QR code printed on the bottom face of the device at manufacturing. The QR contains:

stax-phono://v1/
  ble_addr=AC:67:B2:01:00:2A
  oob_key=<32 bytes hex>
  serial=00042
  fw_min=v0.1.0

On first pair, the host scans the QR with any standard camera (the Stax companion app does this in-process, but a phone camera also works to read the URI). The OOB key seeds an LE Secure Connections Numeric Comparison handshake; the QR is the trust root, not a one-time code. Re-pairing requires line-of-sight to the bottom of the device, which is the right tradeoff: stealing a Phonograph and walking out the door does not get the thief access to the original owner's NATS bus, because the thief never sees the QR.

OTA updates. Signed firmware over BLE. The signing public key is compiled into ROM at the factory; the private key is held by Stax. The AGPL license ensures anyone who forks the firmware can replace the key with their own, so the OTA path is a Stax-signed convention, not a lock-in mechanism. Update flow:

1. Stax publishes a signed .bin to the public repo and the release

feed.

1. The companion app on the paired host downloads the .bin.
2. The host opens a BLE-MTU-aware transfer to the Phonograph (~2 minutes

at BLE 5.3 throughput).

1. The Phonograph verifies the signature against the in-ROM public key,

then atomically swaps to the new image via the dual-partition A/B scheme.

If the signature check fails, the update is discarded and the device continues on the old image. There is no fallback to unsigned firmware.

Composability. Each Phonograph is a NATS event source. Beyond the default bindings, owners can:

• Rebind buttons → any subject via Janet (above).
• Subscribe other devices (Sonos bridge, Hue bridge, garage door, Tesla

charger) to Phonograph subjects.

• Chain Phonographs by running multiple devices on the same household

NATS bus — each addressable by serial number.

The default Sonos binding maps to the existing UPnP driver at <internal-lab>/sonos-cli/sonos.py¹⁹, which already handles the local zone discovery and the transport-control endpoints. The Zig rewrite under <internal-lab>/sonos-cli/src/ will eventually replace the Python prototype; either driver speaks the same NATS subjects.

5. Reproducibility — the open-hardware posture

Per the Editions charter §12⁵:

• Schematics. Published as KiCad source in the public Phonograph

repo at github.com/SMC17/phonograph. KiCad is the chosen tool because it is itself open-source; proprietary EDA formats are rejected as substrate.

• BOM. Published as a JLCPCB-importable CSV, with manufacturer part

numbers, distributor part numbers (Digi-Key and Mouser), and the current quoted price column. The CSV is regenerated against the current pricing on each tagged release.

• Mechanical CAD. STEP file published in the same repo for the body

+ knob + foot ring. Fusion 360 source is held by Stax in the workshop; STEP is the portable format, and any owner with FreeCAD, Onshape, or SolidWorks can reproduce the body from the STEP. The Fusion 360 internal format is acknowledged as a portability gap and is on the migration list — when the FreeCAD parametric workflow reaches feature parity, the source-of-truth migrates.

• Firmware. AGPL-3.0-or-later in the same repo. The license choice

is deliberate: anyone may fork, modify, and ship; they must publish their fork upstream. This is the Red Hat / Linux model per charter §12⁵.

• SBOM. Each Edition ships with a [sentinel-sbom][sentinel-sbom]

-emitted SPDX manifest of every component and its source — every Zig dependency, every Janet module, every C library, every firmware partition, with hash-pinned versions and provenance traces. The SPDX file is on the QR-code landing page (stax.dev/phonograph/<serial>), not just in the repo — supply-chain transparency by construction, per the sovereign-edge thesis.

[sentinel-sbom]: https://github.com/SMC17/sentinel-sbom

6. Manufacturing — the partner-shortlist research lane

Per charter §7¹. Names below are placeholders pending outreach; the spec commits to three quotes per component before any production commitment lands. The active partner research lane is tracked publicly at /workshop/2026-Q2-phonograph-manufacturing.md (a future Workshop entry that begins when manufacturing outreach begins).

| Component | Partner shortlist (research, not committed) | | -------------------— | ------------------------------------------------------------------------------------— | | CNC body | Premier Solutions Pro (CT, US), Hubs / Protolabs (low-volume), Xometry (aggregator) | | PCB + assembly | PCBWay, JLCPCB (low cost); MacroFab (US, premium) | | Knob cap + sleeve | Same CNC partner as body, finished in same anodize lot | | Anodizing | Contracted via CNC partner; specify Type II, RAL color match per lot | | QR + serial engrave | Laser-etched at body-finish step (same CNC partner) | | Foot ring | Standard silicone O-ring stock, color-matched dye | | Packaging | Custom rigid box (Smock or similar US shop); foil-stamped Edition number; kraft wrap | | Product card insert | Hatch Show Print (Nashville, the canonical letterpress shop, per charter §7) |

Decisions land in /workshop/2026-Q2-phonograph-manufacturing.md as the partner research progresses publicly. No partner commitments in this document. Vendors who ship the Edition get a credit in the colophon per charter §7¹.

7. Pricing, run size, distribution

Edition I run: 100 units total, 50 per color. No bonus units. No "founder edition" tier within the run. No NFT-style variants. Every buyer gets the same capsule for the same price, per charter §5 and §9.

Retail price: $280. Premium-but-honest, against a ~$35 BOM. The gross margin funds the editorial + audio components of the capsule (the essay in /journal/, the 60-minute audio program in /canon/audio/, the companion software's continued AGPL maintenance) — the four-component capsule is more than the object¹.

Distribution: direct from stax.dev only. No Amazon. No Shopify-app-bloat. No reseller markup. Payment via Stripe.

Waitlist + purchase-window mechanic per charter §8:

1. Edition I opens as a 7-day waitlist at stax.dev/edition-i.

Free to join. Capped to run size + 20% buffer (120 names).

1. On day 7, the waitlist converts to a 7-day purchase window.

Everyone in the window who clicks Buy gets a unit.

1. Tuesday morning 9:00 AM Eastern, never midnight. Civilized hours.
2. If the run undersells, units stay listed at $280 until they sell or

the next Edition ships, whichever comes first.

1. If it oversells, the next Edition's first slots go to the

waitlist overflow.

Shipping. USPS Priority Mail for the US; DHL eCommerce for international. Every package includes an explicit customs note: the declared value matches the receipt and there are no "gift" undervalued declarations. International buyers pay their own duties; the customs form is honest.

Anti-flipping clause. Per charter §12: first-year transfers must be at price-paid-or-below to maintain warranty. This is soft enforcement — no DRM, no per-device activation server — but the brand stance is explicit and the warranty mechanism gives it teeth.

8. The capsule — what ships with the Phonograph

Per the Editions charter §6¹, the Phonograph is one of four components in Capsule I. The capsule is the unit — no Phonograph ships without all four components in working state.

Object. The Phonograph itself, packed in a custom rigid box. The box contains:

• The Phonograph (charcoal or bone, per order).
• A paired Stax Edition card (5.5 × 8.5", letterpress, foil-stamped

Edition number, etched serial number printed in matching ink). Hatch Show Print stock.

• A kraft-paper wrap around the device.
• A CR2032 cell, pre-installed under a pull-tab insulator (slide

insulator out to activate).

• A USB-C to USB-A cable for first-pair provisioning (15 cm, woven

sleeve, not bagged).

• The printed companion-essay insert (see below).
• A printed QR-code card linking to the software + audio components.

Essay. "Why The Sonos S2 App Is Bad — And What That Tells Us About The Touchscreen-On-Everything Monoculture" — companion piece in /journal/, ~3,500 words, primary-source footnoted in the Mercantile Thesis discipline⁴. Written separately, published to stax.dev/journal/sonos-s2-app-and-the-touchscreen-monoculture.html. A printed insert in the box reproduces the essay as a stapled chapbook on Mohawk paper, set in Charter to match the digital surface².

Software. The multi-room music client — a separate Stax software project, AGPL-3.0, single-binary. Companion macOS app + iOS app. The client speaks Sonos UPnP, AirPlay 2, Spotify Connect, and the Stax NATS bus. Downloads via the box's QR code → stax.dev/phonograph → platform links. The macOS app installs to /Applications and registers a launch agent for the NATS bridge; the iOS app is sideloaded via TestFlight in Edition I and moves to the App Store in Edition I-bis if distribution warrants. The client is sovereign in the same sense the firmware is sovereign: AGPL, no telemetry, no required login, runs forever even if Stax disappears.

Audio. "The Phonograph as Object Lesson" — 60-minute Lineage Mode audio program, history of single-purpose audio remotes from telegraph keys to studio fader handles to the Sonos volume knob. Bundled access via the box's QR → stax.dev/canon/audio/phonograph. The audio is streamed (high-bitrate AAC) and downloadable (FLAC); both formats are included, no upcharge for the lossless version. The audio is licensed CC-BY-NC 4.0 per charter §12⁵ — owners may share with attribution, commercial reuse requires a license.

The four components cite each other. The essay names the Phonograph as its load-bearing example; the Phonograph's QR card names the essay as its source-thesis; the software's --about flag prints the citation chain back to the essay; the audio's outro references the firmware repo. The capsule is a 4D citation of the same idea, instantiated in four substrates.

9. Edition history (forward-pointer)

This is a roadmap, not a date commitment. Per charter §3¹, the cadence is "when the engineering is right, not by deadline."

Edition I — The Phonograph (this document). 2026 Q2 target, 100 units, $280, charcoal + bone. Status: concept (this spec, no prototype boards yet).

Edition I-bis — The Phonograph with display (winter 2026 reprint). A reprint of Edition I with the Sharp LS013B7DH03 memory display added. Same body, same firmware substrate, same price band (likely $300 to absorb the display BOM cost). Resolves the "open-loop knob events" limitation named in §10. Run size: 100 units, same color split. Status: named-but-not-committed; ships only if Edition I sells through and the display integration passes bench testing.

Phonograph Mark II — Stax Edition IV candidate (per charter §6). Improved remote with feedback from Edition I owners. Candidate features (decided after Edition I ships):

• Wired-DAC output mode for headphone amplifiers (becomes a

knob-plus-DAC, not just a remote).

• Larger memory display (1.8") if the I-bis display proves useful.
• Optional rechargeable LIR2032 + USB-C charging by default (CR2032

remains compatible).

• Color expansion: a third anodize color, decided by community vote

among Edition I owners.

Status: candidate, not committed. Charter §6 explicitly defers Edition IV until Editions I–III have shipped. The Phonograph Mark II competes against three other Edition IV candidates (the Codex Deck, the Workshop Volume I, the Sovereign Stack Toolkit) for the slot.

No drops at midnight. No dates promised. The roadmap exists so that Edition I buyers know what compositional moves are possible, not so that future Editions create urgency for the current one.

10. Honest limitations

Per the lab-notebook discipline²: every limitation named explicitly, no hand-waving.

• BLE range ≤10 m through one interior drywall partition. Stone

walls, brick walls, or steel-framed buildings reduce that to ≤5 m. Multi-room households with multiple Phonographs solve this by placing one per major zone; single-Phonograph households in RF-hostile environments are advised that the device may not reach across the building.

• CR2032 service life depends on use intensity. The 14-month

estimate assumes ~10 events/day and infrequent firmware updates. Heavy-use households (50+ events/day, regular OTA cycles) should expect ~8 months. The LIR2032 rechargeable workaround exists for owners who prefer USB-C charging over annual cell replacement.

• **QR-code OOB pairing requires line-of-sight to the bottom of the

device for first pair.** A re-pair (after firmware factory-reset) requires the same. This is a security feature, not a bug — see §4 for the threat model — but it is named honestly: if the device is mounted inside a wall sconce or otherwise inaccessible, re-pairing becomes a physical-access operation.

• No microphone. Voice control is explicitly out of scope. The

Phonograph is a tactile control surface, not an Alexa replacement. Owners who want voice in the same room can run an Alexa or a HomePod alongside; the Phonograph does not compete with that surface.

• **No display in Edition I means knob events are open-loop until the

receiving device confirms via Sonos zone state.** Turning the knob up emits the rotation event; the speaker volume change is the only visual feedback the owner gets. For ≥99% of use this is fine — the speaker's response is the feedback — but in the edge case where the paired zone is muted or the network has dropped, the knob feels unresponsive. Edition I-bis closes this with the memory display (§9).

• The host bridge must be running. The Phonograph does not

directly speak Sonos UPnP — it speaks NATS over BLE to a host that runs the bridge. If the Mac/Linux host is offline, the Phonograph is inert. This is the deliberate appliance-layer architecture: the remote is dumb, the substrate is rich, the substrate is sovereign.

• No iOS background-mode pairing in Edition I. The companion iOS

app must be in the foreground for the first pair. Subsequent BLE reconnects work in background, but the QR scan needs the camera.

• The render in Figure 1 is a render. No prototype boards in

hand. No first-article inspection has happened. Every dimension, every current draw, every tactile feel reference in this document is spec-against-datasheet or engineering-estimate; the field-verified numbers ship with the Edition I run and feed back into the spec via a Workshop log entry.

11. Footnotes

and §7 "Manufacturing partner research lane." Local path: ~/codex/methods/stax-editions-drop-house-charter.md. Amendment 0 dated 2026-05-14.

surfaces" (/objects definition) and §11 "The two known gaps" (the Phonograph identified as the RISD-tier closer). Local path: ~/codex/methods/stax-dev-portfolio-design-system.md. Amendments 1 and 2 dated 2026-05-15.

The Verge, "Sonos has apologized for botching its new app" (2024-07-24); Ars Technica, "Sonos CEO admits new app's launch was 'so botched it warrants an apology'" (2024-07-25). The controversy was load-bearing enough that the CEO's compensation was publicly tied to fixing the app; engineering estimate that the pattern of "controller rewrite that degrades the experience the user paid for" is the broader symptom this object refutes.

Utility and an Appliance" — /journal/mercantile-thesis, 2026-05-06, ~8,500 words, 20 primary-source footnotes. The appliance-layer doctrine the Phonograph instantiates is set out there in full.

for firmware and apps; CC-BY 4.0 for essays; CC-BY-NC 4.0 for audio; trade-dress protection for the capsule format.

github.com/SMC17/sentinel-sbom. The Phonograph manifest at Edition I ship time includes every component, hash-pinned, with provenance traces to the upstream source.

extrusion / ASTM B209 sheet. Yield strength ~276 MPa, tensile strength ~310 MPa, density 2.70 g/cm³. Chosen over 7075 (higher strength, harder to anodize cleanly) and 5052 (better corrosion resistance but lower workability for the encoder shoulder tolerance).

Class 2 (dyed). Coating thickness 18–25 µm. Standard finish for consumer-electronics aluminum where color and corrosion resistance matter more than the harder Type III hardcoat.

https://tech.alpsalpine.com/prod/e/html/encoder/incremental/ec11/ec11.html. 24 detents per revolution, 100,000-cycle mechanical life, push-switch integrated, ~2.5 N axial actuation force. The EC11-style is the industry-standard "good rotary feel" detent encoder; the RE160 series is ALPS's hardened variant.

https://components.omron.com/us-en/datasheet/B3F. 0.98 N actuation force (low end of the B3F family), 0.25 mm travel, 1 million-cycle mechanical life, silent operation. The ALPS SKHHAJ is the SMT equivalent at substantially identical tactile properties.

(2024). https://www.espressif.com/sites/default/files/documentation/esp32-c6_datasheet_en.pdf. RISC-V single-core at 160 MHz, 320 KB SRAM, 320 KB ROM, BLE 5.3, Wi-Fi 6, IEEE 802.15.4. Confirmed RISC-V (not Xtensa).

2021-07-13. The relevant features for the Phonograph are the LE-Secure-Connections improvements and the channel-classification updates. https://www.bluetooth.com/specifications/specs/core-specification-5-3/.

Section 2.3.5 "Pairing Methods" — specifically the Out-of-Band pairing flow. The QR code is the OOB channel; the key in the QR is used as the input to the LE Secure Connections key generation.

https://industrial.panasonic.com/cdbs/www-data/pdf/AAA4000/AAA4000COL13.pdf. Nominal capacity 225 mAh at 0.2 mA continuous discharge, 20 °C; derated to ~187 mAh under the use profile assumed in §3 (engineering estimate, field-verified in Edition I run).

Consumption." Deep-sleep current (RTC retention, no peripherals) is specified as 7 µA typical at 3.3 V; active BLE TX current is ~25 mA at 0 dBm. Both numbers used in §3 are derived directly from these. Engineering estimate; field-verified in Edition I run.

https://www.mouser.com/datasheet/2/365/LS013B7DH03-1144772.pdf. 1.28" diagonal, 128×128 pixels, transflective monochrome, sub-microamp idle current. The display is in Edition I-bis, not Edition I.

budget calculation: TX power +8 dBm, RX sensitivity −96 dBm, margin for one drywall partition (≈3 dB) and free-space path loss at 2.4 GHz. Field-verified in Edition I run. Stone or steel construction reduces by 10–15 dB and is named in §10.

<internal-lab>/firmware/esp32-rugged-node-v0/esp32-rugged-node-v0.ino. Reused conventions: binary frame layout (magic 0xA55A + version + kind + sequence_id + CRC32), USB-serial framing at 115200 baud, explicit little-endian write helpers. The Phonograph firmware inherits this discipline; the BLE GATT path replaces the serial transport.

<internal-lab>/sonos-cli/sonos.py — 229-line Python proof-of-control over the SOAP/UPnP endpoints on port 1400 of every Sonos zone on the LAN. The Zig rewrite under <internal-lab>/sonos-cli/src/ is in progress; both drivers speak the same NATS subjects. Upstream Sonos UPnP documentation: https://developer.sonos.com/ (Sonos developer docs) and Tom Coates' early write-up of the unauthenticated LAN API (https://plasticbag.org/, 2014 era — predates the cloud-required rewrite).

12. Status

status: concept per frontmatter.

Concept stage means: this specification exists; no prototype boards are in hand; no manufacturing partner is committed; no first-article inspection has happened. The /workshop log tracks the move from concept → prototype → shipped with dated entries. Specific milestones from the charter §13:

1. Manufacturing partner outreach. Three quotes per component,

committed by ~2026-06-15. Tracked in /workshop/2026-Q2-phonograph-manufacturing.md.

1. First prototype boards in hand and bench-tested. Estimated

2026-07. Field-verifies current draw, BLE range, knob and button feel. The first-article inspection blocks any further commitment.

1. **Capsule assembly — essay, software, audio all shipped to

/journal, /lab, /canon/audio two weeks before the Edition I drop.** The drop is the event; the capsule is fully present by drop day per charter §13¹.

The status field updates as the milestones land. The spec amends as field-verified numbers replace engineering estimates. Every amendment is numbered and dated, per the canon discipline.

This document is the load-bearing primitive for Stax Edition I. When the document and a prototype disagree, the document wins or the document gets a numbered amendment. Edits ship as Amendment N — DATE.