693.146778 psReference 50% delayUNKNOWNInspect metric contract
0.067228% / FAILCoarse delay driftUNKNOWNInspect metric contract
5 independentStatus axesUNKNOWNInspect metric contract
SIMULATION / SA-CIR-SPICE-CONVERGENCE-EVIDENCE-061
SPICE convergence evidence
Convergence Is One State
A solved run is one status, not the whole evidence chain. Compare timestep, tolerance, initial state, saved signals, metric definitions, and a deliberate singular operating point without promoting convergence to validity or signoff.
EXPLORE
Start with the observable.
Use this page when the simulation is the primary artifact. The linked lesson explains the concept; the linked lab keeps the visual replay surface available.
- Compare the reference and 500 ps cases: both solve, but only one passes the locked numerical-refinement limit.
- Inspect why tiny mean-output drift can coexist with a near-threshold delay decision flip.
- Separate missing signals, incompatible metric definitions, alternate initial state, and singular operating-point failure from model validity and signoff.
PRECOMPUTED EVIDENCE / CURRENT TAB ONLY
A solved run is one status, not the whole evidence chain.
9 cases5 independent axes0 server runs
SELECTED CASE
Under-resolved / tight / 500 ps
The converged trace exceeds at least one locked refinement tolerance.
- PASS
- 2
- FAIL
- 1
- NOT RUN
- 2
RUN STATUS
Five axes, no automatic propagation
Converged is not validated. Read each column independently.
| Case | Solver | Numerical | Model | Metric | Signoff | Delay |
|---|---|---|---|---|---|---|
| PASS | PASS | NOT RUN | PASS | NOT RUN | 693.146778 ps | |
| PASS | PASS | NOT RUN | PASS | NOT RUN | 693.191372 ps | |
| PASS | FAIL | NOT RUN | PASS | NOT RUN | 693.612769 ps | |
| PASS | PASS | NOT RUN | PASS | NOT RUN | 693.171248 ps | |
| PASS | NOT RUN | NOT RUN | PASS | NOT RUN | 405.674593 ps | |
| PASS | FAIL | NOT RUN | PASS | NOT RUN | 575.236905 ps | |
| PASS | NOT RUN | NOT RUN | FAIL | NOT RUN | - | |
| PASS | PASS | NOT RUN | FAIL | NOT RUN | 697.548110 ps | |
| FAIL | NOT RUN | NOT RUN | NOT RUN | NOT RUN | - |
Did this analysis satisfy its configured numerical stopping criteria?
Does refinement keep the scoped waveform and metric inside locked tolerances?
Has the model been validated for the intended physical range?
Are required signals, thresholds, direction, and interpolation compatible?
Has an independent governed flow accepted the result for its stated use?
RUN CARD
Conditions before conclusions.
The browser compares precomputed frozen local replay data. It never executes an arbitrary netlist or sends circuit data to a server.
Primary knobs
- selected and reference run
- status-axis inspection
- waveform signal and trace
- metric-contract definition
Saved outputs
- nine-run status matrix
- refinement and decision ledger
- precomputed waveform overlay
- metric contract and required-signal audit
- deterministic JSON and CSV report
EXPORT
Download the public artifacts.
These exports are the supported public files for review. PDK files and restricted third-party material are not redistributed.
V129 convergence evidence packet
/data/semiagora-spice-convergence-evidence-v1.json
Open file ->Original reference RC netlist
/data/semiagora-spice-convergence-rc-reference-v1.cir
Open file ->Deliberate nonconvergence netlist
/data/semiagora-spice-convergence-nonconvergence-v1.cir
Open file ->Machine-authority structural SVG
/data/semiagora-spice-convergence-rc-structural-v1.svg
Open file ->
METHODS
How the number was made.
Methods are written for citation discipline: the extraction rule matters as much as the plotted value.
Measure input-to-output 50% rising delay with linear interpolation between saved samples.
Keep this method attached when reusing the figure or metric.
Compare delay drift, waveform RMSE, maximum error, and mean-output drift against the exact reference replay.
Keep this method attached when reusing the figure or metric.
Keep solver, numerical, model, metric, and signoff status independent; downstream states never auto-pass.
Keep this method attached when reusing the figure or metric.
LIMITS
What this page does not claim.
The MVP is useful because the limits are visible. These statements prevent a teaching simulation from being cited as silicon evidence.
Convergence means the numerical criteria were met for this netlist, model, analysis, and solver configuration.
Carry this boundary into any derivative note, paper draft, or slide.
Convergence does not establish numerical refinement, model validity, metric suitability, or signoff.
Carry this boundary into any derivative note, paper draft, or slide.
All thresholds are locked educational evidence checks, not production limits.
Carry this boundary into any derivative note, paper draft, or slide.
Ideal source, linear resistor, and ideal capacitor only; no transistor, interconnect extraction, package, measurement, or silicon calibration.
Carry this boundary into any derivative note, paper draft, or slide.
The conflicting-source fixture demonstrates a singular ideal constraint only; it is not a universal convergence recipe.
Carry this boundary into any derivative note, paper draft, or slide.
No proprietary model, solver runset, process condition, production circuit, measured waveform, or qualified limit is included.
Carry this boundary into any derivative note, paper draft, or slide.
Researcher-facing businesses remain deferred to a separately verified V200 contract.
Carry this boundary into any derivative note, paper draft, or slide.
EVIDENCE
Trace the claim back to artifacts.
The evidence route remains the catalog-level browser; this page is the simulation-level reading card.
Five ngspice CSV/log pairs and one deliberate nonconvergence log are preserved in the internal evidence packet.
SA-CIR-SPICE-CONVERGENCE-EVIDENCE-061
Every public case retains settings, required signals, metric ID, sample count, status axes, reason, and source-trace ID.
SA-CIR-SPICE-CONVERGENCE-EVIDENCE-061
Five measured-replacement slots state what is required before physical validity or signoff can be evaluated.
SA-CIR-SPICE-CONVERGENCE-EVIDENCE-061
Source spine
These links anchor the public teaching model. They do not convert the replay into a qualified process recipe or signoff result.
ngspice Documentationngspice / Convergence criteria, transient analysis, solver controls, and versioned manual access.ngspice User Manualngspice / Newton iteration, voltage/current tolerances, operating-point aids, and measurement syntax.SPICE2: A Computer Program to Simulate Semiconductor CircuitsUC Berkeley EECS / Historical numerical formulation, nonlinear solution, and integration-method context.Xyce Documentation and TutorialsSandia National Laboratories / Cross-simulator terminology for nonlinear solvers, time integration, and error controls.Xyce Users Guide 7.7Sandia National Laboratories / LTE control, timestep adaptation, integration-method accuracy, and damping boundaries.Errors and uncertainty in numerical simulationNIST / Separation of physical-model, discretization, and computational error classes.KiCad Schematic EditorKiCad / Terminal, wire, crossing, and junction conventions for the original RC schematic.SemiAgora Circuit Rendered-Connectivity Standard v2SemiAgora / Electrical, symbol, vector, raster, human, and integration gates for the public schematic.
Public execution boundary
No server-side live ngspice execution, uploads, accounts, payments, comments, newsletter signup, or job/event submissions are active on this page.