A laboratory optical test reported a heterodyne signal with a signal-to-noise ratio above 2 at a spectral flux density of 188 pW per nanometre, measured at 1.56 micrometres with 0.4 milliseconds of integration. The result came from a two-arm, single-spectral-channel prototype using an artificial stellar source. The work is an arXiv version 1 preprint dated 28 August 2026.
The central idea is to use a broadband serrodyne heterodyne arrangement that could separate measurements by wavelength. The paper evaluates an architecture intended to support spectral multiplexing and simultaneous J- and H-band coverage. That is a design goal, however, rather than a result from a multi-channel experiment: only one H-band spectral channel was demonstrated.
The laboratory arrangement
The prototype was a two-arm, single-spectral-channel, polarization-maintaining heterodyne interferometer built from telecom-fiber components in the H band. Its local oscillator was a broadband femtosecond laser with approximately 200 nanometres of optical bandwidth. It emitted approximately 20-femtosecond pulses at 80 megahertz, which were stretched to approximately 12.5 nanoseconds using 750 picoseconds per nanometre of accumulated chromatic dispersion.
To mimic a stellar source in the laboratory, the setup used a continuous-wave SLED with a 60-nanometre full width at half maximum, meaning the stated spectral width was measured at half the peak. Tunable fibre-coupled filters matched that source spectrum to the local oscillator. An 8-nanometre spectrum was selected for single-channel operation.
Detection used matched photodiodes and differential subtraction. That balancing step was intended to suppress common-mode laser noise and the steady direct-current background while retaining the heterodyne beat signal. The experiment varied the optical path difference in delay scans and, at each delay, extracted the modulus of the Pearson correlation coefficient from the 1-megahertz Fourier component using pseudo-lock-in detection. The output was a correlation trace across optical delay.
What the test actually measured
At strong SLED power, the measured signal agreed with the calculated Fourier transform, and fringe visibility reached 0.98. Fringe visibility is the contrast of the interference signal across its bright and dim states. The authors identified detector and beam-splitter imperfections as likely limits on the measured value.
For the detection-limit test, the paper estimated SNR by dividing a fitted amplitude by the residual standard deviation, using a high-power reference correlation trace. Detailed analysis found a correlation-trace SNR above 1.4 at 3 nanowatts and 400 microseconds. Using the conventional squared-fringe-visibility definition, that corresponds to approximately 2, with per-arm spectral flux density below 188 pW per nanometre at 1,560 nanometres.
Those figures need to be read with the paper's two SNR definitions in mind. The abstract reports the experimentally demonstrated result as an SNR above 2 at 188 pW per nanometre, 1.56 micrometres and 0.4 milliseconds, while the detailed trace analysis uses the above-1.4 correlation measure before translating it to the visibility convention. No interval or replicate-based uncertainty was reported for the result.
The gap between measurement and projection
Noise was a separate weakness. The continuous-laser control was reported as shot-noise limited, but serrodyne measurements with optical skew or without high-pass filtering were not. Even with proper balancing and filtering, the remaining relative intensity noise, or RIN, was twice the expected shot-noise level. The experimental data were fitted with a three-component noise model.
The authors' scaling discussion projects that 16 CWDM wavelength channels, combined with 40 milliseconds of integration, could reach sensitivity below 10 pW per nanometre in the H band, even with uncooled photoreceivers. But the experiment demonstrated only one H-band spectral channel, so the multiplexed figure remains a projection rather than a measured result.
The selected configuration was also reported to correspond to a spectral resolving power of approximately 8,600, matching the photoreceiver bandwidth. For a general reader, that is a measure of how finely the arrangement is expected to distinguish nearby wavelengths. The figure is tied to the selected configuration, not to a demonstrated multi-channel measurement.
Taken together, the work is a laboratory proof of principle for a broadband, single-channel H-band measurement. It shows a useful correlation signal and near-unity visibility at strong source power, but it leaves the central engineering questions open: whether multiplexed channels can deliver the projected sensitivity and whether technical noise can be brought down to the shot-noise limit.
Paper data and sources
Original title: Broadband heterodyne interferometry with a chirped femtosecond laser in the H-band
Authors: Félix Gudin, Nicolas Forget
Journal/Repository: arXiv
Status: Preprint, not yet peer-reviewed
First online: 2026-08-28
DOI: Not available
Original paper · Full text