Pulsim Netlist Format Reference (YAML)

Pulsim uses a versioned YAML netlist format. This document describes the required structure, supported component types, and waveform definitions.

Top-Level Structure

schema: pulsim-v1
version: 1
simulation:
  tstart: 0.0
  tstop: 1e-3
  dt: 1e-6
  step_mode: variable
components:
  - type: voltage_source
    name: V1
    nodes: [in, 0]
    waveform: {type: dc, value: 5.0}
  - type: resistor
    name: R1
    nodes: [in, out]
    value: 1k

Required Fields

Field	Type	Description
schema	string	Must be pulsim-v1
version	int	Schema version (currently 1)
components	list	Component list

Optional Fields

Field	Type	Description
simulation	map	Simulation options
models	map	Reusable component models

Components

Para a matriz completa de paridade do catálogo do PulsimGui com status de suporte backend, veja gui-component-parity.

Passive Components

• Resistor: type: resistor (or R), value
• Capacitor: type: capacitor (or C), value, optional ic
• Inductor: type: inductor (or L), value, optional ic

Sources

• Voltage Source: type: voltage_source (or V)
• waveform supports dc, pulse, sine, pwm
• Current Source: type: current_source (or I) with value (DC)

Switching Devices

• Ideal Diode: type: diode (or D), g_on, g_off
• Ideal Switch: type: switch (or S), ron/roff or g_on/g_off, initial_state
• Voltage-Controlled Switch: type: vcswitch, nodes [ctrl, t1, t2], v_threshold, g_on, g_off

Power Devices

• MOSFET: type: mosfet/nmos/pmos (or M)
• IGBT: type: igbt (or Q)

Transformer

• Transformer: type: transformer (or T), turns_ratio

Waveforms

waveform:
  type: dc
  value: 12

waveform:
  type: pulse
  v_initial: 0
  v_pulse: 5
  t_delay: 0
  t_rise: 1e-9
  t_fall: 1e-9
  t_width: 1e-6
  period: 2e-6

waveform:
  type: sine
  amplitude: 2.5
  frequency: 1000
  offset: 0
  phase: 0

waveform:
  type: pwm
  v_high: 10
  v_low: 0
  frequency: 20000
  duty: 0.5
  dead_time: 1e-6
  phase: 0

Models and Overrides

models:
  m1:
    type: mosfet
    params:
      vth: 2.0
      kp: 20e-6

components:
  - type: mosfet
    name: M1
    nodes: [g, d, s]
    use: m1
    params:
      kp: 40e-6  # override

Simulation Options

simulation:
  tstart: 0.0
  tstop: 1e-3
  dt: 1e-6
  step_mode: variable    # fixed | variable
  formulation: projected_wrapper  # projected_wrapper | direct
  direct_formulation_fallback: true
  dt_min: 1e-12
  dt_max: 1e-3
  adaptive_timestep: true # override avançado; prefira step_mode
  enable_events: true
  enable_losses: true
  integrator: trbdf2   # trapezoidal, bdf1, bdf2, trbdf2, rosenbrockw, sdirk2

simulation.backend e simulation.sundials são chaves legadas e não fazem parte da superfície suportada de transiente. Use step_mode + formulation.

For thermal-port rules, strict/non-strict behavior, and complete electrothermal examples, see Electrothermal Workflow.

Solver Configuration

Use simulation.solver to control linear/iterative solver selection and nonlinear aids:

simulation:
  solver:
    order: [klu, gmres]
    fallback_order: [sparselu]
    allow_fallback: true
    auto_select: true
    size_threshold: 400
    nnz_threshold: 2000
    diag_min_threshold: 1e-12
    preconditioner: ilut        # or amg (if available)
    ilut_drop_tolerance: 1e-3
    ilut_fill_factor: 10
    iterative:
      max_iterations: 200
      tolerance: 1e-8
      restart: 40
      preconditioner: ilut      # or amg (if available)
      ilut_drop_tolerance: 1e-3
      ilut_fill_factor: 10
      enable_scaling: true
      scaling_floor: 1e-12
    nonlinear:
      anderson:
        enable: true
        depth: 5
        beta: 0.5
      broyden:
        enable: false
        max_size: 8
      newton_krylov:
        enable: false
      trust_region:
        enable: true
        radius: 1.0
        shrink: 0.5
        expand: 1.2
        min: 1e-4
        max: 10.0
      reuse_jacobian_pattern: true

You can also set Newton options directly under simulation.newton if preferred.

simulation:
  newton:
    max_iterations: 50
    enable_newton_krylov: true
    krylov_residual_cache_tolerance: 1e-8
    reuse_jacobian_pattern: true

Periodic Steady-State

simulation:
  shooting:
    period: 10e-6
    max_iterations: 30
    tolerance: 1e-6
    relaxation: 0.5
    store_last_transient: true

  harmonic_balance:
    period: 10e-6
    num_samples: 128
    max_iterations: 40
    tolerance: 1e-6
    relaxation: 0.5
    initialize_from_transient: true

SI Prefixes

Values support SI prefixes: f, p, n, u, m, k, meg, g, t.

Examples: 1k = 1000, 100n = 100e-9, 4.7u = 4.7e-6.

Where to go next

This page covers the schema of a netlist. For exhaustive per-type parameter tables, defaults, and aliases see:

• Components Reference — every electrical device (resistor, mosfet, transformer, saturable_inductor, fuse, relay, thyristor, …) with parameters, defaults, units, and notes on surrogates and YAML-vs-Python quirks.
• Control Blocks Reference — every virtual block (pwm_generator, pi_controller, lookup_table, clarke_transform, park_transform, pll, svm, …) with channels and cross-block metadata wiring patterns.
• KPI Reference — declarative benchmark.kpi: entries that score the captured trace (THD, ripple, ZVS%, T_j, etc.).
• Three-Phase Grid Library — vector-control pipelines built from Clarke / Park / PLL / SVM blocks.
• Electrothermal Workflow — the thermal: block on MOSFET / IGBT / BJT components.