A documented teaching reconstruction, not a certified digital twin.The earlier exhibit did not identify a manufactured radio or provide measured component drawings. This rebuilt receiver therefore preserves its open-breadboard character while adopting a coherent two-01A circuit. It distinguishes manufacturer facts, a photographed valve specimen, calculated geometry, and explicitly assumed model parameters. Exact visual identity to an unspecified physical radio cannot be established.
What was corrected
| Earlier exhibit | Rebuilt behavior |
|---|
| Aerial connected to the common return at the same tank terminal. | A separate primary couples the aerial to the tuned secondary; the aerial is not shorted to common. |
| Four-pin battery-style tube drawn with a separate heater and cathode. | Directly heated 01A: a continuous V filament is the cathode, with plate, grid and two filament connections. |
| Receiver behavior selected web-stream gains using a Gaussian frequency window. | The AM lab uses a loaded LC response, nonlinear grid-current rectification, grid-capacitor charge, a triode plate law and actual load equations. |
| Grid capacitor and leak were drawn, but their detection behavior was not calculated. | The exact-RF scope integrates Cg and grid current; the interactive audio circuit uses carrier-averaged nonlinear detection. |
| A single detector drove a generic cone speaker without a defensible power budget. | A second 01A, a 1:3 interstage transformer, a negative grid-bias supply and a sensitive magnetic horn form the complete audio path. |
| Transformer secondary leads terminated at the mounting foot. | Four insulated transformer terminals connect two separate windings. The horn connects in the output tube plate circuit, with no invented output-transformer stage. |
| Electrons could disappear at the grid according to random probabilities. | Slowed tracers originate on the filament and travel toward the plate. Tracer density follows calculated plate current; trajectories are explanatory, not particle-in-cell calculations. |
| Visible turns and dimensions had no documented relationship to the circuit. | The coil has 97 turns at 22 mm mean radius and 55 mm winding length. The capacitor has 10 rotor plates, 11 stator plates and 0.40 mm air gaps. |
| Internet audio, displayed signals and artificial motion were easy to conflate. | Internet playback is a separate native-browser feature. Its audio is not claimed as measured RF, and no synthetic horn motion is substituted for inaccessible network audio. |
The electrical contract
The reference valve is the type 01A: 5 V, 0.25 A directly heated filament; μ = 8; approximately 725 µS transconductance and 2.5 mA plate current at 90 V / −4.5 V. The detector uses a 45 V supply, 250 pF coupling capacitor and adjustable 0.25–5 MΩ grid leak returned to the positive filament terminal, within the manufacturer’s published detector recommendations. A 1 nF mica capacitor bypasses detector plate RF. The scope’s RF-cycle-averaged plate trace is a separate 1 kΩ winding-DC-load diagnostic; the running receiver includes the transformer and bypass dynamics.
Two separate 4 Ω filament rheostats drop the 6 V accumulator to 5 V per hot filament. The B supply has 45 and 90 V taps. The C supply provides −4.5 V relative to F− common. T1 has a 1:3 ratio, 25 H primary magnetizing inductance and 1 kΩ primary DC resistance. The 500 kΩ volume potentiometer loads the secondary. V2 drives a polarized magnetic horn represented by 2 kΩ copper resistance, 1 H inductance and a reciprocal magnetic motor. The output tube’s actual quiescent plate voltage is about 85.75 V and current 2.125 mA; the 90 V / 2.5 mA datasheet point is not substituted for this loaded operating point.
f₀ = 1 / (2π√LC) · |H| = 1 / √[1 + Q²(f/f₀ − f₀/f)²] · C(θ) = 41 pF + 334 pF × overlap. Geometry gives a 333.8 pF overlap swing; 334 pF is the nominal circuit value. L = 240 µH, loaded Q = 48, and the tuning range is approximately 530.5–1604.4 kHz. The visible winding is consistent with the approximately 240 µH single-layer Wheeler estimate.
What is calculated—and what is assumed
Calculated:LC tuning and passband; complex LC filtering of the carrier and both AM sidebands; explicit RF grid-current/capacitor KCL in the scope; carrier-averaged grid-leak behavior in the running receiver; nonlinear plate/load intersections; transformer magnetizing current, winding resistance, 500 kΩ secondary load and reflected positive-grid current; filament electrical/thermal warm-up; reciprocal horn force and back-EMF; diaphragm motion and electrical/assumed mechanical power.
Assumed or locally fitted:loaded Q, primary coupling, grid-current initial-velocity tail, thermal loss constants, interstage magnetic parameters and horn motor constants. The triode law is an empirical local fit to the manufacturer’s 90 V / −4.5 V operating point and slope. It is not a fitted full-family SPICE tube model. The directly heated filament’s distributed potential is represented by that lumped empirical characteristic, using F− as the stated voltage reference.
Not solved:a self-consistent three-dimensional Maxwell/Poisson/space-charge field, antenna radiation and loading from a measured installation, core hysteresis or saturation, manufacturing tolerances, hum/noise spectra, a particular horn’s acoustic impedance or calibrated sound pressure. The amplifier and horn are intended for small-signal teaching; very strong drive goes beyond a specimen-validated range. Electric-field arrows show direction only. Electron size/speed and diaphragm displacement are magnified. Exploded positions are display offsets; they are never used as operating geometry.
Audio:laboratory sound is synthesized from the numerically solved diaphragm velocity after the circuit settles, with an explicitly uncalibrated listening gain and click-suppressed buffer changes. It preserves parameter-dependent detection, distortion and fault behavior; it does not predict room sound pressure. Internet stations play directly through an HTML audio element, allowing broadcasters that do not permit cross-origin Web Audio analysis. In listening mode, each preset has a stable virtual dial position. A smooth acquisition window blends native broadcast volume with separately synthesized, filtered static. Static is continuous when off station and while the native stream is connecting, buffering or unavailable; it fades out only after the browser reports playable broadcast audio. The virtual slots and noise are interface behavior, not measured RF frequencies or a thermodynamic noise calculation. No oscilloscope measurement or artificial diaphragm movement is claimed for those streams.
Material and construction reference
The two valves use a globe envelope close to a photographed 44 mm × 99 mm type 201A specimen. Internal geometry follows the manufacturer’s P-60 plate, grid and filament construction drawings held in the Antique Wireless Association archive: a flattened nickel anode, a 26-turn grid, a fine V filament, mica brace and spring support. Combining those documented internals with the earlier globe envelope remains an explicit historical teaching reconstruction, not a metrologically verified specimen. Wood, black phenolic, enamelled copper, mica, nickel, brass, glazed ceramic, rubber and fibreboard use distinct physically based material responses. Fine wear and finishes are procedural and representative.
Verification record
The receiver prepares nearby stations continuously and mixes program and static by tuning, without reconnecting. In the source reconstruction this reception engine passed 20 deterministic reception and lifecycle cases and the laboratory audio 10 state cases, run with controlled media doubles; they are not browser decoding or acoustic measurements. The Museum page runs the same circuit model, and its numbers are compared with the source in the Museum’s own tests. First-time remote playback still depends on network data, decoding and browser permission.
The electrical tests check the model’s equations and interventions; they do not establish the accuracy of every assumed parameter against a real receiver. In bounded transport checks on 8 October 2026 all 33 station addresses returned data: 30 MP3 or AAC audio responses and 3 HLS playlists. Transport is not a promise of decoding, long-duration uptime or worldwide access. Live playback status comes from actual browser media events. The three HLS presets require native HLS; a link to the station’s own player remains available.
Primary and specimen sources
- Type 01-A manufacturer data sheet, 1 May 1935 — Operating ratings, pinout and grid-leak detector recommendations.
- Type 201A original manufacturer sheet — Historical direct-filament construction and operation.
- The Valve Museum, type 201A specimen — Photographed globe envelope, 44 mm diameter and 99 mm height excluding pins.
- Antique Wireless Association archive — Original factory construction drawings; Standardizing Notice 3-1-201-A, June 1930. Courtesy of the Antique Wireless Association.
- The station directory lists only stream addresses that the broadcasters publish themselves; each station’s own player is linked beside it.
Source review: 8 October 2026 UTC. Geometry, textures and the AM laboratory are computed in the page; network access is only needed to listen to stations or open references.
Continuous tuning
Power-on prepares the current and adjacent stations. Recently used connections remain available, with at most five native streams. Detuning changes program and static levels while each stream keeps advancing; returning to a ready station changes its volume immediately without a new connection.
Program and static are audible together near a station. Their levels vary continuously over a 12 kHz half-width for ordinary presets, narrowed where custom slots are closer. Centering a ready station clears static. A first visit to an unprepared remote station still needs network data and decoding; buffering and station failures also restore static and are identified in the controls.
This virtual reception envelope and generated static are listening controls, not measured RF selectivity or a thermal-noise solution. Internet streams play in the browser independently of filament warm-up. The AM laboratory continues to solve the separate documented circuit.
Inspect the 16 numerical acceptance checks
- Passed · Nominal valve operating point and small-signal slope
- Passed · LC scale, capacitor monotonicity and inverse dial mapping
- Passed · Loaded resonator has the expected half-power bandwidth
- Passed · Nonlinear plate solver obeys its load line over operating range
- Passed · Transformer changes impedance/voltage without creating power
- Passed · Explicit 980 kHz grid detector rectifies and satisfies capacitor KCL
- Passed · RF time-step refinement changes mean detection by less than 1 percent
- Passed · RF scope respects power, antenna and detector controls without inventing plate current
- Passed · Heater circuit converges to 5 V / 0.25 A per directly heated filament
- Passed · Detector, amplifier and horn form an energized audio path
- Passed · Missing aerial, open detector, zero modulation and zero volume suppress steady audio
- Passed · Detuning changes the detected signal, not only a station selector
- Passed · Strong-drive grid conduction remains finite and loads the transformer
- Passed · Magnetic motor respects reciprocal force/back-EMF energy balance
- Passed · Internet material is reported as AF monitor injection
- Passed · AM laboratory sound is generated by the numerical horn motion
These tests validate the implemented equations and control behavior within the assumptions stated above. They are not measurements of a physical radio. The complete numerical record is kept with the exhibit’s archived source page.
