Skip to content

Oscillator linewidth

The frequency spread of a selected physical oscillator output line, reported with a specified spectral estimator and width convention.

Version
v1 · 2026-10-25 · History
Domain-specific #
14427
Domain group
Natural Sciences
Origin domain
Physics
Subdomains
Oscillators, Spectral Metrology → Physics

Core Idea

Oscillator linewidth tells how spread out one selected oscillator output line is in frequency.[1][2] A report should name the output line, the way its width was measured or inferred, and the width convention. A number from a spectrum analyzer can differ from a number calculated from phase or frequency noise, even for the same oscillator.[1][2]

The width does not by itself reveal one unique noise cause.[2] In a spin-torque oscillator study, the width inferred from phase noise accounted for about 70% of the width measured with a spectrum analyzer for two studied modes.[2] This makes the measurement method part of what the number means.

Scope of Application

The same reporting pattern applies to unlike physical oscillators, including an optical laser and an electrically observed spin-torque nanocontact.[1][2] Each case has a recurring output, a chosen spectral line, and a width in frequency units. Their instruments and underlying physics differ.

The pattern can include a line-shape fit or an inference from phase or frequency noise, but an inference needs a stated model. Keller and colleagues give a phase-noise-to-FWHM relation under a one-sided white-frequency-noise convention and after neglecting amplitude noise.[2] In other studied spin-torque devices, nonwhite noise caused width to increase weakly with measurement time.[3]

Clarity

FWHM means full width at half maximum; HWHM is half of that width for a symmetric line shape. In Li and colleagues' laser fit, the Lorentzian full width was 15.0 kHz, expressed as twice their half-width parameter.[1] Their separate noise-based procedure inferred an 11.3 kHz intrinsic Lorentzian linewidth.[1] These are two qualified estimates, not a single value repeated two ways.

An oscillator's carrier frequency is also different from its linewidth. Keller's 128 MHz voltage mode had a spectrum-analyzer FWHM of 0.78 MHz.[2] A plot of phase noise can help infer a width, but it is not already a line-width number without the conversion assumptions.

Manages Complexity

Oscillators have many possible noise sources and many measurement procedures. Linewidth reporting reduces that complexity to a few questions: which line, which method, which width statistic, and which conditions matter? The answers make results comparable without pretending that all reported widths are identical.

The reporting structure also prevents a misleading short description. A laser's fitted width and its intrinsic noise-derived width can differ.[1] A spin-torque oscillator's phase-derived width can fall below its full analyzer width.[2]

Abstract Reasoning

If two linewidth numbers differ, check their definitions before treating the results as contradictory. Did they refer to the same output line? Was each number a fitted component, a whole observed peak, or a noise-model estimate? Did both use FWHM, or did one quote a half-width? If observation time or resolution affects the result, record that condition too.[1][2][3]

These checks support a careful inference about spectral coherence. They do not, on their own, identify whether phase noise, amplitude noise, drift, or the instrument produced a difference. Keller's authors presume amplitude noise contributes to the gap in their studied device but say amplifier noise obscures the intrinsic amplitude variations in their waveform data.[2]

Knowledge Transfer

The selected-line and width-convention checks transfer from optical emission to electrical voltage oscillation.[1][2] The numerical noise model does not automatically transfer: the physical dynamics and the one-sided or two-sided noise conventions still need checking. A general width of any distribution is only an analogy if no physical oscillator output line is involved.

Example

Li and colleagues measured an integrated Pockels laser.[1] A fitted Lorentzian component of its delayed self-heterodyne result had a full width of 15.0 kHz.[1] A different, correlated phase-noise procedure inferred an intrinsic Lorentzian width of 11.3 kHz from a frequency-noise floor.[1] The laser is the oscillator, its optical emission is the chosen line, and the two numbers have different estimators.

Keller and colleagues measured a spin-torque oscillator's voltage.[2] For its 128 MHz mode, the phase-noise-based FWHM was 0.54 MHz, while the spectrum-analyzer FWHM was 0.78 MHz.[2] The chosen voltage mode is the line, and the method label explains why the two widths should remain distinct.

Neighborhood in Abstraction Space

Oscillator linewidth sits in a sparse region of the domain-specific corpus (72nd percentile for distinctiveness): few abstractions share its structure, so a faithful description tends to retrieve it precisely.

Family — Unclustered & Miscellaneous (2769 abstractions)

Nearest neighbors

Computed from structural-signature embeddings · 2026-10-25

Not to Be Confused With

  • Phase-noise density: a noise plot needs a justified conversion before it supplies a linewidth.[2]
  • Half-width: a Lorentzian HWHM is half its FWHM; the convention must be named.[1]
  • Carrier frequency: where a line is centered is different from how wide it is.[2]
  • Every spectral width: an atomic transition line without oscillator output has a different carrier; a generic spectral band is a different object.

References

[1] Mingxiao Li et al. “Integrated Pockels laser”. Nature Communications 13, 5344 (2022). Fig. 2d–f and caption, pp. 3–4, and “Laser linewidth measurement,” Methods, pp. 7–8, support the two qualified laser widths and methods; full author-hosted article available at the published PDF. registry ↩a ↩b ↩c ↩d ↩e ↩f ↩g ↩h ↩i ↩j ↩k ↩l

[2] Mark W. Keller, A. B. Kos, T. J. Silva, W. H. Rippard, and M. R. Pufall. “Time domain measurement of phase noise in a spin torque oscillator”. Applied Physics Letters 94, 193105 (2009). Original full author manuscript, pp. 1–3, especially Figs. 1, 3–4 and the FWHM comparison on p. 3, supports the one-sided model and two mode comparisons; the NIST published copy is also available. registry ↩a ↩b ↩c ↩d ↩e ↩f ↩g ↩h ↩i ↩j ↩k ↩l ↩m ↩n ↩o

[3] Mark W. Keller, Matthew R. Pufall, William H. Rippard, and Thomas J. Silva. “Non-white frequency noise in spin torque oscillators and its effect on spectral linewidth”. Physical Review B 82 (2010). The NIST author abstract supports the bounded measurement-time result; the full paper was not used for detailed equations or figures. registry ↩a ↩b