Circuit Lab: electronics, pedals, bending and oscillator research

Inside the project

Prepared with the project’s development agent from source and retained evidence. This is a project-owner assessment, not an independent customer review. Dated checks were not rerun for this website.

Circuit Lab is most useful as an inspectable engineering laboratory: it connects component laws to circuit equations, then exposes the resulting behaviour through measurements, plots and native experiments. Its strongest foundation is the separation between an equation solver, reusable Swift packages, a Workbench interface and instrument-specific research. Readers can follow a nonlinear circuit or capacitor experiment from its assumptions to retained numerical evidence. The 0.36 validation record, dated 16 October, reports 522 passing root tests and six required live ngspice regressions. These are recorded results, not tests rerun for this review. The current coverage map contains 403 catalogue entries, including 293 planned entries; that inventory is not a count of calibrated device models.

MXR Envelope Filter research demonstrates the value of this approach. The topology-derived averaged model has passed the documented full-range comparison cases and supports an offline Gear Locker demonstration. A separate factored optimisation passes 35 identical-signal circuit cases, 15 native checks and four Swift tests, but still lacks accepted quiet-machine timing. Its numerical success cannot establish live playback readiness. The AUv3 is a separate research prototype: its live mode is explicitly illustrative, while the averaged circuit requires host-declared offline rendering and fixed controls. The 17 October Logic investigation reproduced an editor-driven allocation mismatch and repaired that path in version 0.2.2. The recorded regression reached 1,536 successful blocks with zero rejections; the older unexplained fault count remains unattributed.

Circuit Bending now has a stronger native demonstration than yesterday's buffered-preview description suggests. The main Circuit Lab 0.22.2/build 222 report records shared MIDI input, direct bounded rendering, output-route selection and Panic controls. Its retained session processed 640,000 frames across 1,250 callbacks with zero observed underruns or late callbacks. This is a useful, bounded demonstration of the supported fixture, not general circuit compilation or a promise about arbitrary hardware. Physical keyboard feel and combined Performance-window audio acceptance remain separate questions.

Stellar provides the oscillator side of the estate: modular synthesis, modulation graphs, coupling and nonlinear dynamics, with oversampling comparisons rather than a blanket sound-quality claim. Its 13 October report records 119 passing workspace tests and independent FM, PM and PWM comparisons. A separate 32-node, 64-route, 4x speaker soak completed 60 seconds without observed budget misses; dense graphs still overloaded. Existing synthetic WAV exports offer an immediate media candidate, subject to the website curator's clearance.

LF-X1 is earlier: a DC–1 kHz instrument design pack with numerical tools and Phase 0 evidence. The retained checks support its filter algebra and planning, not a built transmitter, receiver or calibrated instrument. Priorities across the family are clear: finish MXR performance and host/device acceptance, keep Bending's supported fixture boundaries visible, characterise Stellar workloads, and take LF-X1 through its simulation and implementation gates before hardware claims.

Features & evidence

Workflow walkthrough

A guide to the native workflow. These steps describe the app; only the recordings and images embedded on this page are available as website demonstrations.

1. Open the Workbench screenshot and identify the shared scope/measurement workflow; retain its historical version caption.
2. Inspect the nonlinear-charge figure to connect a component law with its numerical experiment.
3. Read the MXR comparison figure beside the distinct offline averaged and illustrative AU preview labels; show the diagnostic editor capture without claiming optimized live admission.
4. Offer the existing synthetic Stellar sine and metallic-network WAVs as optional audio players after curator clearance; compare the source types, without claiming listening acceptance.
5. Finish with the recorded Bending test summary and LF-X1 Phase0 report; label a Bending interaction video as a future capture rather than an available demo.