Noise Analysis Tools

These tools analyze output from LTspice .noise simulations. Run a noise simulation first with a directive like .noise V(out) V1 dec 100 1 1meg, then pass the resulting .raw file to any of the tools below.

onoise vs. inoise

LTspice noise simulations produce two variables: onoise (output-referred noise) and inoise (input-referred noise). Output-referred noise is the noise spectral density measured at the output node --- it represents what an instrument would see at the output. Input-referred noise is calculated by dividing the output noise by the circuit’s gain, giving the equivalent noise you would need to inject at the input to produce the same output. Use onoise when you care about absolute output noise levels; use inoise when comparing a circuit’s noise floor against the source signal.

---

analyze_noise

Comprehensive noise analysis from a .noise simulation. Parses the .raw file and returns spectral density across all simulated frequencies, spot noise at five standard frequencies (10 Hz, 100 Hz, 1 kHz, 10 kHz, 100 kHz), total integrated RMS noise over the full bandwidth, noise figure relative to the source resistance, and an estimate of the 1/f (flicker) corner frequency.

The noise figure calculation uses the formula NF(f) = 10 log10(|noise(f)|^2 / (4 k T R)), referenced to the IEEE standard temperature of 290 K.

Parameter	Type	Default	Description
raw_file_path	str	---	Path to .raw file from a .noise simulation.
noise_signal	str	"onoise"	Which noise variable to analyze. "onoise" for output-referred, "inoise" for input-referred.
source_resistance	float	50.0	Source impedance in ohms, used for noise figure calculation.

Returns: A dict with five top-level keys:

• spectral_density — arrays of frequency_hz, noise_density_v_per_sqrt_hz, and noise_density_db.
• spot_noise — a dict keyed by frequency label ("10Hz", "100Hz", "1kHz", "10kHz", "100kHz"), each containing spot_noise_v_per_sqrt_hz, spot_noise_db, and actual_freq_hz. Only includes frequencies within the simulation range.
• total_noise — total_rms_v, integration_range_hz (two-element array), and equivalent_noise_bandwidth_hz.
• noise_figure — arrays of noise_figure_db and frequency_hz, plus min_nf_db and nf_at_1khz.
• flicker_corner_hz — estimated 1/f corner frequency in Hz, or null if not detectable.

Example

{
  "spectral_density": {
    "frequency_hz": [1.0, 1.26, 1.58, "..."],
    "noise_density_v_per_sqrt_hz": [4.2e-7, 3.9e-7, 3.5e-7, "..."],
    "noise_density_db": [-127.5, -128.2, -129.1, "..."]
  },
  "spot_noise": {
    "10Hz": {
      "spot_noise_v_per_sqrt_hz": 3.1e-7,
      "spot_noise_db": -130.2,
      "actual_freq_hz": 10.0
    },
    "1kHz": {
      "spot_noise_v_per_sqrt_hz": 1.2e-7,
      "spot_noise_db": -138.4,
      "actual_freq_hz": 1000.0
    }
  },
  "total_noise": {
    "total_rms_v": 4.7e-5,
    "integration_range_hz": [1.0, 1000000.0],
    "equivalent_noise_bandwidth_hz": 156000.0
  },
  "noise_figure": {
    "noise_figure_db": [12.3, 11.8, "..."],
    "frequency_hz": [1.0, 1.26, "..."],
    "min_nf_db": 3.2,
    "nf_at_1khz": 5.1
  },
  "flicker_corner_hz": 850.0
}

---

get_spot_noise

Get noise spectral density at a specific frequency. Interpolates linearly between adjacent simulation data points. If the target frequency falls outside the simulated range, the nearest boundary value is returned.

Parameter	Type	Default	Description
raw_file_path	str	---	Path to .raw file from a .noise simulation.
target_freq	float	---	Frequency in Hz at which to measure noise density.
noise_signal	str	"onoise"	"onoise" for output-referred or "inoise" for input-referred.

Returns: A dict with spot_noise_v_per_sqrt_hz (noise density in V/sqrt(Hz)), spot_noise_db (20 log10 of the density), and actual_freq_hz (the frequency actually used --- equals target_freq when interpolation succeeds, or the clamped boundary frequency otherwise).

Example

{
  "spot_noise_v_per_sqrt_hz": 1.2e-7,
  "spot_noise_db": -138.4,
  "actual_freq_hz": 1000.0
}

---

get_total_noise

Integrate noise spectral density over a frequency band to get total RMS noise. Computes total_rms = sqrt(integral(|noise|^2 * df)) using trapezoidal integration. This is the value you would read on an RMS voltmeter with a bandwidth equal to the integration range.

Parameter	Type	Default	Description
raw_file_path	str	---	Path to .raw file from a .noise simulation.
noise_signal	str	"onoise"	"onoise" or "inoise".
f_low	float | None	None	Lower integration bound in Hz. Defaults to the minimum frequency in the simulation data.
f_high	float | None	None	Upper integration bound in Hz. Defaults to the maximum frequency in the simulation data.

Returns: A dict with total_rms_v (integrated RMS noise in volts), integration_range_hz (two-element array [f_low, f_high]), and equivalent_noise_bandwidth_hz (the bandwidth of a brick-wall filter with the same peak density that would pass the same total noise power).

Example

{
  "total_rms_v": 4.7e-5,
  "integration_range_hz": [100.0, 100000.0],
  "equivalent_noise_bandwidth_hz": 42000.0
}