Struct SpiceRunner 

pub struct SpiceRunner { /* private fields */ }

Runner for ngspice simulations.

Implementations

impl SpiceRunner

pub fn new(config: SpiceConfig) -> Self

Create a new SPICE runner with the given configuration.

pub fn check_ngspice() -> Result<String, SpiceError>

Check if ngspice is available.

pub fn simulate( &self, circuit_path: impl AsRef<Path>, input: &[f64], output_node: &str, ) -> Result<Vec<f64>, SpiceError>

Run a SPICE simulation with the given input signal.

Arguments

• circuit_path - Path to the .spice circuit file
• input - Input signal samples at internal_rate
• output_node - Node name to measure (e.g., “v_out”)

Returns

Output signal resampled to sample_rate

pub fn config(&self) -> &SpiceConfig

Get the configuration.

pub fn decimate_to_base(&self, internal: &[f64]) -> Vec<f64>

Decimate an internal-rate signal (sample_rate × oversample) down to the base sample_rate by taking every oversample-th sample.

[simulate] returns at the INTERNAL rate (to match the oversampled main runner). The standalone *_spice.rs tests instead compare against a WDF processor compiled at the BASE sample_rate, so their golden must be at the base rate too — otherwise a 1 kHz golden tone carries oversample× the samples-per-period of the WDF tone and the 1:1 comparison lines up two signals at different rates (the bug this fixes).

pub fn simulate_settled_window( &self, circuit_path: impl AsRef<Path>, input_internal: &[f64], output_node: &str, settle_s: f64, ) -> Result<Vec<f64>, SpiceError>

Run a SETTLED steady-state simulation and return the MEASUREMENT window at the base sample_rate.

input_internal is the full stimulus at the internal rate covering settle_s + measure_s seconds. The circuit is simulated over that whole span so coupling caps / bias networks reach steady state; the first settle_s is then discarded and the remaining tail is decimated to the base sample_rate. The settle boundary is floored to a multiple of oversample so the decimated grid stays phase-aligned with a WDF processor warmed up by the same settle_s.

The returned golden is at the base sample_rate: len ≈ measure_s × sample_rate, and a tone of frequency f has sample_rate / f samples per period.

impl SpiceRunner

pub fn operating_point( &self, circuit_path: impl AsRef<Path>, ) -> Result<SpiceOpSnapshot, SpiceError>

Run ngspice .op on a standalone deck and capture the DC operating point.

Composes the deck the same way Self::simulate does (stripping standalone control/analysis directives), injects a quiescent VIN v_in 0 DC 0 source, and runs an .op followed by show (per-device params) and print all (node voltages). The show table is parsed into SpiceDeviceOp records (vce = vbe − vbc, ic from the row).

This is read-only and ngspice-derived — the resulting snapshot is the golden the WDF DC bias is graded against.