Pulsed Laser

Interference

With Beam with size and Interference enabled, split copies of this source interfere through ideal beamsplitters, fully reflective flat mirrors, delay lines and phase objects. The calculation integrates the actual spectrum: fringe contrast changes with arm mismatch, and the spectrometer sees the changed spectrum. Independent sources add as powers. New sources enable this; older saved sources keep it off until you enable it. The Coherence length readout gives the arm difference at which the fringes fall to half contrast for the spectrum as set.

The combined temporal field is not reconstructed. After recombination, duration, autocorrelation and oscilloscope traces are unavailable; pulse animation stops at the recombination surface. Material and coating phase outside the supported ideal optics is not inferred. A spectral or path-budget limit is reported explicitly. See the model and bounds.

The Pulsed Laser emits the same collimated ray or 25-ray sampled beam as the CW Laser, plus a pulse train: a repetition rate, a pulse duration, and an emission offset that shifts this source's pulses in time relative to any other. That timing is what drives the travelling packet overlay, the oscilloscope view on a photodetector, chopper and AOM/EOM gating, and the two-colour temporal overlap that CARS and SFG require.

Transform-limited pulses decides what you author. On, you set the pulse duration and envelope shape, and the bandwidth beneath is computed from them, so a shorter pulse automatically becomes a wider spectrum. Off, you set the bandwidth, the chirp's sign — Positively or Negatively chirped (quadratic) — and its GDD in fs²; the transform-limited duration and the emitted pulse duration beneath are computed. The duration is therefore always derived and can never fall below the limit the bandwidth sets. Switching to chirped initializes the bandwidth from the current transform-limited pulse and resets the GDD to zero, so the emitted pulse is unchanged; previous chirped settings are replaced. Switching to transform-limited restores the last transform-limited duration and can change the spectrum. Pulse energy (average power ÷ repetition rate) and peak power are readouts, never entered; for a chirped sech² pulse the peak power is an estimate, since a dispersed sech² pulse does not keep an exact sech² profile.

The laser offers no “phase unknown”. That is a simplification of authoring, not a claim that every real laser carries a known quadratic phase: a multi-longitudinal-mode nanosecond laser, for instance, is far longer than its bandwidth's transform limit without being chirped, and cannot be described this way — author it transform-limited at its duration. Light generated on the bench can still have an unknown phase (see the crystal and OPO pages). A sketch saved before these controls opens with its bandwidth and the GDD that reproduces its saved duration, with the sign it was saved with or positive — provided that GDD fits the 10⁷ fs² range; beyond it the GDD is clamped and the duration changes. A train saved at 0 nm opens transform-limited at its duration. The bandwidth field's fixed bounds hold the transform-limited bandwidth of every authorable pulse (1 fs to 1 ms) at every allowed wavelength, so the spectrum a switch to chirped starts from is kept through a save and reload, and no wavelength or shape edit moves it. A chirped bandwidth can imply a transform limit outside the 1 fs – 1 ms the transform-limited field accepts; that is why switching to transform-limited does not derive a duration from it.

Show pulse dynamics is a drawing choice only — switching it off leaves the beam rendered as a steady CW line while every bit of the pulse physics above keeps running.

Dispersion and pulse stretching

Every pulsed detector reports accumulated group-delay dispersion (GDD) in fs². Catalogue-glass bodies add their traced distance through the selected Sellmeier material; zero-thickness lenses and objectives add the clearly marked estimates described on their own pages. A chirped laser's GDD is its own quadratic phase, added once to the path's: glass of the opposite sign therefore compresses the pulse to its transform limit before stretching it again. Light whose phase nobody authored — an OPO or crystal output declared “spectral phase unknown”, say — reports its configured duration only where the path's modeled dispersion adds up to zero (“Configured duration · zero net modeled dispersion”), and is shown as unavailable anywhere else. Unknown does not mean uncompressible; it means the phase that would decide it is not known.

When a filter, dichroic, etalon or AOTF reshapes the pulse's spectrum, the duration is worked out from the spectrum that survives. An ideal filter changes the spectrum's amplitude and not its phase, so the pulse at a detector is the numerical transform of the surviving spectrum with the source's chirp plus every GDD on the path, wherever the filter stands; a slice cut from a band that glass has fanned out carries the glass's GDD at its own wavelength. One effective quadratic phase stands for the band, so where the glass's GDD varies across the surviving band enough to move its edge phase by more than half a radian — a broad band behind thick glass — the duration is unavailable rather than approximated. A hard-edged passband gives a sinc-like pulse with side lobes, and the reading is its full width at half maximum; an autocorrelator draws the shape it assumes at that width. An etalon's output keeps its power but not its fringe comb or its transfer phase, so it is not timed. A filtered supercontinuum, which has a duration but no phase, is an assumed-sweep estimate: its duration is taken as a sweep linear in frequency across its band, a slice keeps the share its width spans, and that share adds in quadrature to the slice's own dispersed duration as a source part no compressor removes — a stated convention, not a reconstruction of the continuum's phase. The model also declines, and says why on the detector's Duration model row, when parts of one beam reach a detector by paths whose dispersion differs — separate paths do not compensate each other the way glass and a compressor in sequence do, so equal and opposite GDD on two arms is not one transform-limited pulse. Small differences, within 0.1 rad of quadratic phase across the bandwidth and 2 % in duration, are treated as one path; that is a display tolerance, not a claim that distinct fields combine. The model also declines when an aperture catches only part of a pulse that a prism or grating has fanned out by wavelength: the arriving samples must carry at least 95 % of the emitted spectral weight. Elements that transmit the pulse's whole band evenly, such as a neutral density filter or a dichroic far from its edge, leave the duration alone.

Sech² pulses use the same signed-phase construction, but their dispersed FWHM is read from a deterministic numerical Fourier-propagation table rather than the Gaussian formula. Interpolation was checked against a denser calculation to 0.12% over |GDD|/τ₀² ≤ 20 and continues with the 2π·0.315 large-dispersion asymptote. Detector, autocorrelator, probe, scope, and travelling packet all consume this one duration model. GDD remains the primary number because it is additive and meaningful even when a 150 fs pulse changes too little to notice.