Log — Loudspeaker Design Wiki

Append-only history. Entries: ## [YYYY-MM-DD] <op> | <title> (ingest/query/lint/split).

[2026-06-28] split | loudspeaker-design-wiki created from _inbox cluster (3 sources)

Spun out of the hub _inbox loudspeaker-design cluster the moment a 3rd coherent source landed — three arrived within ~2 minutes via Telegram, so the cluster completed on arrival rather than sitting parked. Founding corpus: 3 source pages + 3 Thing pages + synthesis.

ts-parameters-explained (TechArticle, T3, audiojudgement.com) — the T/S parameter set explained with DIY cases.
loudspeaker-primer-hackaday (TechArticle, T2, hackaday.com) — driver fundamentals, crossovers, why enclosures exist, MDF cabinets, the DSP shortcut.
micka-enclosure-calculator (WebApplication, T2, micka.de) — enclosure calculator consuming T/S parameters; the design loop closed.
Thing pages: thiele-small-parameters, loudspeaker-driver, loudspeaker-enclosure.
Synthesis thesis: loudspeaker design is predictive, not empirical — datasheet numbers → simulated box → response, before any wood is cut. Open questions: how T/S params are measured; crossovers/room acoustics; the DSP bridge to ../speech-audio-wiki.
Why a new spoke, not speech-audio-wiki: that spoke owns audio AI models (TTS/STT/audio-gen); this is physical electroacoustics (drivers/enclosures/transduction) — different domain, only the word “audio” shared. Scope boundary recorded in CLAUDE.md.
Publisher/maker entity nodes deferred (thin at founding). Registry block added to ../wikis.md; hub _inbox/thiele-small-parameters.md park record deleted (superseded by ingest).

[2026-06-29] ingest | T/S parameters — measurement methods (Wikipedia) — quality cycle

First quality-cycle pass on the day-old spoke; closed its top open question. New source ts-parameters-wikipedia (Article, T2, en.wikipedia.org) — how T/S parameters are actually measured: free-air impedance sweep (Re cone-blocked, Fs at the impedance peak, Q factors from the curve), Vas via a second added-mass/sealed-box reading, three impedance rigs (constant-current / constant-voltage / small-series-resistor), and the small-signal-snapshot + unit tolerance caveat (Thiele=ABC, Small=Univ. Sydney, JAES 1972). Added a “How they’re measured” section to thiele-small-parameters and folded the measured-inputs point into the synthesis thesis; the synthesis “Measurement” open question is now Answered (residual sub-gap: hands-on measurement gear, DATS/REW). Not padding — answers a named open question with a T2 primary. Floor improved (T2 2→3, no T4).

[2026-06-30] ingest | Hard-drive speakers (Hackaday) — the transducer alone

Routed from Telegram. New source hdd-speakers (TechArticle, T2, hackaday.com) — Niv Singer drives hard-drive head actuators as loudspeakers: the voice-coil actuator that normally positions the R/W head becomes a speaker driver when fed audio. 4× WD Caviar 500GB, 2 per stereo channel with a crossover (one drive low / one high), platters PWM-synced to the music; plays Future Crew’s “Second Reality” (Assembly 1993). The spoke’s first unconventional-transducer source and first crossover instance. Integrated: added a voice-coil-is-the-whole-motor note to loudspeaker-driver; synthesis gained a “transducer alone” point (first step into the flagged general-electroacoustics broadening — transduction without the engineering, the opposite pole from the T/S design loop) and the crossover open-Q now has a concrete (if unconventional) instance. Maker (Niv Singer)

publisher (Hackaday) entity nodes deferred — tangential one-off hacker / publisher already prose-cited (consistent with the spoke’s zero-entity posture). Ran avoid-ai-writing. +1 page (→8).

[2026-07-09] ingest | CamillaDSP (RASPIAUDIO) — open DSP active-crossover appliance

Routed from hub (Telegram). New source camilladsp (SoftwareApplication, T2, github.com/RASPIAUDIO) — an open-source Rust audio DSP engine packaged by RASPIAUDIO as a flashable Raspberry Pi 5 appliance: active crossovers (Linkwitz-Riley 24 dB/oct), parametric EQ, per-driver delay/time alignment, routing; their 8xOUT board (Pi 5 I2S ×4 → 4 stereo DACs = 8 outputs), CamillaGUI (port 5005), USB-audio gadget mode; pitched to active speaker builders as an open miniDSP alternative. New concept node crossover (DefinedTerm, sourced to the primer) — the frequency-splitting network, passive LC vs active/DSP, slopes & Linkwitz-Riley; the integrate-the-drivers layer after the T/S→box loop. Integrated: answers two flagged open-Qs — the DSP bridge now has a concrete open tool (reframed as a multi-amp + correct-in-software design philosophy) and crossovers get their first proper (active/DSP) instance + their own node (vs the ad-hoc split in hdd-speakers). Scope-boundary check: this DSP is signal-chain correction of a real transducer → stays in this spoke (hardware/acoustics), not speech-audio (models); bridge kept as cross-link. Remaining gap noted: passive LC crossovers still unsourced. Entities (RASPIAUDIO org, CamillaDSP author) deferred per the spoke’s zero-entity posture. T2 (first-party repo; mild vendor incentive on the board framing). Ran avoid-ai-writing. +2 pages (→10).

[2026-07-10] ingest | Chesky Audio LC2 reference monitor (Future Audiophile review) (hub-routed, Telegram)

Ingested the Future Audiophile review of the Chesky Audio LC2 (~$1,995/pr reference bookshelf; Lucca Chesky). The wiki’s first commercial finished-loudspeaker page: new chesky-audio-lc2 (Product, source, T3). Grounds the concepts and adds one genuinely new alignment node — passive-radiator (DefinedTerm): a motor-less tuned “drone cone,” the 4th alignment beside infinite-baffle/sealed/vented (LC2 = twin 8″ radiators → 40 Hz in a bookshelf). Updated loudspeaker-enclosure (added passive-radiator alignment + the 3D-printed acoustically-inert cabinet as a material-not-mass answer to cabinet resonance). Synthesis: three new design- tension bullets (cabinet-material, port-vs-drone-cone, sensitivity-not-wattage) + first Contradictions entry (“measured vs. heard” — subjective review vs. the measure-first thesis). Driver details (aperiodic dome tweeter, tri-laminate cone, cast basket) captured in-page. T3 — single subjective review, no gated/anechoic measurement; specs are the maker’s. Maker node (Chesky Audio) deferred per spoke policy (page on recurrence). share.google link resolved via firecrawl. avoid-ai-writing applied (clean). +2 pages, 1 concept updated.

[2026-08-08] ingest | IEC 60268-5 replacement + AES T/S manufacturing-variation study — research pass

Two sources via the hub Quality Cycle’s research pass, against this spoke’s #1 corpus-wide edge (first T1 — it was the last zero-T1 spoke). Both qualify at the T1 bar and both close it: T1 0 → 2.

iec-60268-5-replacement — geoff-hill, chair of AES SC-04-03 and a member of IEC PT 60268-22, on why IEC 60268-5 was withdrawn and what replaced it (60268-21 / 60268-22 / 63034). Tiered T1 by what it rests on — the author’s own committee work and the standards’ contents — not by the masthead it appeared under.
ts-parameter-manufacturing-variation — Laurin & Reichard (Penn State), AES Convention 125, 2008: T/S parameters measured across twenty nominally identical drivers, with a follow-on phase correlating spread with failure mode and lifetime. E-library listing only — the page says so and claims no figures from inside the paper.

What they change. The predictive loop rests on measured inputs and the spoke had never sourced the regime those inputs come from. It now has one, plus the finding that the regime expired — and that it expired because 60268-5 assumed an “entirely passive” device, which is exactly the active/DSP speaker this spoke reached from the hobbyist side (camilladsp). The DSP bridge turns out to be why the standard broke, not a side-branch. Folded into synthesis; thiele-small-parameters carries the datasheet-comparability caveat.

The second source upgrades the standing measure-your-own-driver advice from builder folklore to a measured subject, without supplying the magnitude.

Entities: 3 created — geoff-hill, audio-engineering-society, iec. Laurin and Reichard were not paged: abstract-only knowledge, no recurrence, same call as the hub’s precedent on thin one-off paper authors.

Edges refreshed: first-T1 closed; successor edge added — the standards’ own text, since the spoke now knows which documents its numbers come from and holds none of them.

[2026-08-10] ingest | Audio DSP guide (American Hard Bag) — the environment that can’t be designed

Routed from the hub (Telegram). americanhardbag.com/blogs/installation-guides/audio-dsp — a Harley-audio retailer’s explainer, in the installation-guides section of its own store. T4: no author, no measurement, and it recommends the product categories it sells.

New pages: audio-dsp-hardbag-guide (source), audio-dsp (concept), in-vehicle-audio (environment). camilladsp already covered DSP as a tool, so the concept page holds the function set rather than duplicating it.

Why a T4 earned three pages. The spoke’s thesis assumes the acoustic load is knowable in advance — a box you specify, a room that holds still. A motorcycle has no enclosure and a noise floor that tracks road speed. loudspeaker-primer-hackaday calls DSP a shortcut around getting the passive design right; here it is the only variable left, because the environment was never designable. That bounds the thesis instead of contradicting it: the predictive loop covers the transducer and its load and stops at the door of the space.

One asymmetry recorded: chassis placement makes path lengths uneven (so time alignment is necessary, not optional), but the listener doesn’t move, which makes single-position correction worth more here than in a room.

Vendor spec checked and failed. The quoted 85 ms / 4096-sample delay is ~29 m of path difference at 343 m/s; a vehicle needs under 3 ms. Buffer headroom quoted as capability. The arithmetic is on the page.

Growth edges added: #9 vehicle audio measured (this entered on one T4 page plus geometry), #10 what DSP correction costs (fixed vs floating point, cumulative filter error, the excursion ceiling on any EQ boost — the corpus can’t say).

[2026-08-10] research | Hi-fi bass reproduction, and what a used speaker sells for

Directed research pass (no source handed in). The curator asked for research on hi-fi bass gear and its sellability, routed here from the hub. Two sources fetched, five pages written.

Sources. svs-sealed-vs-ported (SVS, T3 — a subwoofer maker arguing both sides of its own catalogue) and hifi-value-guide-potteries (Dave Spence / Potteries Auctions, 2026-04-23, T3 — an auction house that profits from consignments). Both are trade parties with a stake; recorded as such.

Two fetches failed with HTTP 403 — Audioholics’ “Ten Nonsense Myths About Subwoofers” and Headphonesty’s February 2026 resale-depreciation dataset. Not transient socket errors, so no retry loop; both are logged as open items on loudspeaker-resale-value and in synthesis.md growth edges #11/#12. The neutral, measured group-delay comparison this pass wanted is still missing.

New pages. group-delay and room-gain (DefinedTerm) fill two concepts the spoke had been referring to without paging — group delay was already listed among the four curves a calculator predicts, and never defined. loudspeaker-resale-value is an explicit adjacency page: economics, not electroacoustics, kept because it is where a design decision meets a price.

What changed in the thesis. The sealed-vs-vented tension is restated as a currency — extension and output are bought in milliseconds, and the tuning frequency sets the exchange rate. And the predictive loop gets a second boundary next to the room: the used market prices brand and condition, not measurement, so predicting quality and predicting resale are different problems.

loudspeaker-enclosure updated to link both new concepts. index.md regrouped accordingly.