# D15 — E4-H160: An aperture–dispersion ceiling on range resolution in wavelength-steered lidar

**Batch:** 7 · **Selected rank:** 15/30 · **Tractability:** 8/10 · **Significance:** 7/10

## 1. Original hypothesis and refined claim

The original: in a wavelength-steered optical phased array the beam angle is a function of optical frequency, so the chirp that buys range resolution also sweeps the beam across the scene. A target is illuminated only while the beam covers it, so the effective bandwidth saturates at `B_eff = c/(λD|dθ/dλ|)` and the range resolution floors at `δR = λD|dθ/dλ|/2`, independent of chirp bandwidth and sweep time.

The scaling survives and the derivation can be exact. Three things are wrong or missing:

- **(R1)** The ceiling is `δR = max(c/2B, λD|dθ/dλ|/2)`, with crossover `B* = c/(λD|dθ/dλ|)`. The number of usable range bins is `N = B_eff T = cT/(λD|dθ/dλ|)`, independent of B: extra chirp bandwidth adds neither resolution nor bins.
- **(R2)** The floor **grows linearly with aperture**: a larger aperture makes range resolution *worse*. The original's §9 says the opposite — only a larger aperture improves range resolution — contradicting its own formula, where `B_eff ∝ 1/D`.
- **(R3)** For a monostatic reciprocal array the window is the overlap of transmit and receive dwells, `T_eff = T_dwell − 2R/c`, so resolution degrades linearly with range and `R_max = cT_dwell/2 = (cT/2)(B*/B)`: over-chirping costs range.

## 2. Refined scientific description

In a wavelength-steered array the antenna and the waveform are one instrument. The aperture applies a transverse phase gradient that sets the beam direction; a chirp changes the optical frequency, hence the gradient, hence the direction. The chirp therefore walks the beam off the target, and the receiver sees only that fraction of the sweep during which the target stays inside the diffraction width; range resolution is set by the spectral support of the *received* signal, not the laser's tuning range.

The coupling is pure geometry: aperture sets the angular acceptance, dispersion sets how fast frequency converts to angle, and their ratio is the usable bandwidth. Laser tuning range, modulator speed and sweep time cancel, so the ceiling belongs to the antenna, not the source — and post-processing cannot recover photons the off-target part of the sweep never returned.

## 3. Verified supporting literature and prior-art assessment

[EST] FMCW range resolution `c/2B` for a linear chirp with coherent detection. [EST] The grating/array-factor relation between inter-element phase and beam angle, and the diffraction-limited beamwidth `≈λ/D` for an aperture of diameter D. [EST] The time–bandwidth product as the determinant of range resolution, and the Fourier reciprocity between spectral support and impulse width.

[REC] On-chip wavelength-steered optical phased arrays and grating-array lidar demonstrations, consistent in field-of-view claims but not re-verified here. [REC] FMCW lidar at 1550 nm with resolutions quoted from bandwidth alone.

[UNV] A measured range-resolution-versus-bandwidth curve for a wavelength-steered array, which is the one dataset the hypothesis needs and the one I could not confirm exists. [UNV] That such arrays are characterised by field of view rather than instantaneous bandwidth — plausible from the literature's emphasis, not verified.

Prior art: wavelength steering, FMCW ranging and the diffraction limit are textbook. Narrow here is the product — the *ceiling* and its linear growth with aperture. I found no source stating either, and the original's own §9 mis-states the sign.

## 4. Proposed physical or mathematical model

Aperture phase: `φ(x,t) = (ω(t)/c) x sinθ(ω(t))`, so the transverse phase gradient is `κ(t) = ∂φ/∂x = (ω/c) sinθ(ω)`. The beam points where `κ` matches the target's transverse wavevector. Differentiating at fixed x,

`dκ/dt = (2π/λ) cosθ (dθ/dλ)(dλ/dt) = −(2π/λ)(dθ/dλ)(λ²/c)(B/T) cosθ`,

using `dλ/dt = −(λ²/c)(B/T)`. The aperture can resolve a change in gradient only up to its own spectral width, `Δκ = 2π/D`. Setting `|dκ/dt| T_dwell = Δκ`:

`T_dwell = cT / (λ D |dθ/dλ| B)`,

and the effective bandwidth is the part of the sweep falling inside that window, `B_eff = B T_dwell/T = c/(λD|dθ/dλ|)`. Hence `δR = c/(2B_eff) = λD|dθ/dλ|/2` for `B > B*`, and `c/2B` below.

For a monostatic reciprocal array the echo returns after `2R/c`, by which time the receive beam has moved on; the overlap of the two dwell windows is `T_eff = T_dwell − 2R/c`, giving `B_eff(R) = B T_eff/T` and `R_max = cT_dwell/2`.

## 5. Variables, assumptions and boundary conditions

`λ` wavelength (m); `ω = 2πc/λ`; `D` programmed aperture diameter (m); `Λ` array pitch (m); `θ` beam angle from broadside; `dθ/dλ` angular dispersion (rad m⁻¹), equal to `1/(Λcosθ)` for a grating array; `B` chirp bandwidth (Hz); `T` sweep time (s); `Ω = |dθ/dλ|λ²B/(cT)` angular sweep rate (rad s⁻¹); `κ` transverse phase gradient (rad m⁻¹); `Δκ = 2π/D`; `T_dwell`; `B_eff`; `δR` range resolution (m); `N = B_eff T` usable bins; `R` target range; `R_max`.

Assumptions: far field (`R > 2D²/λ`); a single transverse mode and a diffraction-limited beam; linear chirp; dispersion dominated by the array geometry rather than waveguide material dispersion; a point target filling less than one beamwidth; a reciprocal monostatic array for (R3). Small parameters: `B/B*` large in the ceiling regime; `2R/(cT_dwell)` small for (R3). Domain: `θ` away from the grating's grazing limit, where `cosθ → 0` and the dispersion diverges.

## 6. Quantitative predictions

1. **Ceiling.** For `λ = 1.55 µm`, `D = 1 mm`, `|dθ/dλ| = 10⁵ rad m⁻¹`: `B* = 1.9 THz`, `δR = 78 µm`. At `B = 10 GHz` the resolution is `15 mm`; at `100 GHz`, `1.5 mm`; at `1 THz`, `150 µm`; at `10 THz` still `78 µm`. Computed from stated inputs, not measured.
2. **Aperture scaling.** `δR ∝ D` and `B* ∝ 1/D`: at `D = 1 cm` the crossover falls to `194 GHz` and the floor rises to `780 µm`; a tenfold larger aperture multiplies the floor and divides `B*` by ten.
3. **Bin count.** `N = cT/(λD|dθ/dλ|)`; with `T = 1 ms` and the 1 mm aperture, `N = 1.9×10⁹` bins across a `150 km` window, unchanged when `B` is raised tenfold.
4. **Range penalty.** `R_max = (cT/2)(B*/B)`: `150 km` at `B = B*`, `15 km` at `10 B*`, `1.5 km` at `100 B*`.
5. **Dispersion scaling.** `δR ∝ 1/|dθ/dλ|`, so doubling the grating dispersion halves the floor — the one lever the original identifies correctly.

## 7. Comparison against conventional explanations

The conventional account is FMCW theory: `δR = c/2B`, buy bandwidth, and a larger aperture is unambiguously better because it raises gain, narrows the beam and improves angular resolution. On that account the two are independent and each improves monotonically.

The separating observable is the aperture scaling at fixed bandwidth. Above `B*`, doubling the programmed aperture must double `δR` and halve the bandwidth at which the plateau begins. No conventional mechanism produces that: laser phase noise gives a floor set by coherence time and independent of `D`; detector or digitizer bandwidth gives a floor independent of `D`; a finite tuning range gives a hard cap on `B` independent of `D`. Only the angular-gating mechanism ties the floor to aperture and dispersion, and only it predicts that the plateau's onset moves as `1/D`.

A second discriminator separates the two versions of the hypothesis. The ceiling is architecture-independent; the range penalty (R3) is not. A separate, statically pointed receive telescope keeps the ceiling and loses the `R`-dependence.

## 8. Specific experimental or computational method

One method: a space–time propagation model of the array, with range compression by matched filter, swept in bandwidth at fixed aperture and in aperture at fixed bandwidth.

Model: elements at pitch `Λ` with far field `E(θ,t) = Σ_n a_n exp[i(ω(t)t − κ(t)x_n)]`; a point target at range `R` and angle `θ_t` returns the delayed field, gated by the same angular response; compression is the transform of the chirp times the two-way angular window. Grid: 64 to 1024 elements, `B` from 1 GHz to 30 THz, `T` fixed at 1 ms. Control arm: a phase-steered array of the same aperture, where `θ` is set electronically and does not depend on `ω`, and the resolution must follow `c/2B` with no plateau. Nuisance held fixed: chirp linearity and sweep time, since either can produce an apparent plateau by a different route.

Acceptance: the plateau holds if `δR` changes by less than 10 per cent over a decade above `B*` while the control's `δR` falls tenfold.

## 9. Necessary controls

- Phase-steered array of identical aperture: no plateau.
- Bandwidth sweep at fixed aperture over three decades: locates `B*`.
- Aperture sweep at fixed bandwidth, 4× range: confirms `δR ∝ D`.
- Sweep-time series at fixed `B`: the floor holds while `R_max` moves.
- Grating-pitch series: `δR ∝ 1/|dθ/dλ|` at fixed `D`, separating geometric from material dispersion.

## 10. Sources of measurement or numerical error

The dominant uncertainty is the beamwidth convention: `λ/D` versus `0.886λ/D` for a uniformly illuminated aperture, and FWHM versus null-to-null, changes `δR` by up to a factor of two. This is a prefactor, not a scaling, and every prediction above should be read as correct to that factor.

Second, the far-field condition `R > 2D²/λ` is `1.3 m` at 1 mm aperture but `130 m` at 1 cm, so a short-range aperture test measures the near field. Third, material dispersion in the waveguide adds to `|dθ/dλ|` and a bandwidth sweep does not separate it. Fourth, a target wider than a beamwidth is illuminated progressively, smearing the gate. Fifth, (R3) is architecture-dependent and should not be the primary test.

## 11. Falsification criteria

- **F1.** At fixed aperture and dispersion, a `δR` that keeps falling as `1/B` past `B*` by more than 20 per cent over a decade of bandwidth kills (R1).
- **F2.** A floor that does not scale with programmed aperture — a 4× aperture change giving less than a 3× change in `δR` — kills the mechanism, since no other term carries `D`.
- **F3.** A floor that moves when only the sweep time changes indicates a coherence or processing artefact, not angular gating.
- **F4.** For a reciprocal monostatic array, an `R_max` that does not scale as `1/B` kills (R3).
- **F5.** A phase-steered control array showing the same plateau kills the attribution to wavelength steering.

## 12. Expected scientific contribution

Turns an antenna parameter into a ranging specification: aperture, dispersion and sweep time fix the floor and the bin count, and bandwidth beyond `B*` buys nothing. It reverses the intuition that a larger aperture is free — it costs range resolution linearly while buying angular resolution.

## 13. Resource and feasibility assessment

Computational for the central claim: a workstation and an open-source array-factor model; no fabricated device is needed for (R1) and (R2). Confirming it against hardware needs a wavelength-steered array with a programmable aperture and a fibre delay line, estimated at weeks of cleanroom and bench time, uncosted and unverified. No dataset is required; a published resolution-versus-bandwidth curve would allow a direct test.

## 14. Reproducibility plan

1. Fix `λ = 1550 nm`, `T = 1 ms`, `Λ` and `θ = 0`; record the beamwidth convention explicitly.
2. Compute `κ(t)` from the chirp, gate the return by the two-way angular overlap, compress by matched filter, record the half-maximum width.
3. Sweep `B` from 1 GHz to 30 THz in 40 steps at `D = 1 mm`; repeat at `D = 2, 4, 10 mm`.
4. Fit `δR = [(c/2B)² + (λD|dθ/dλ|/2)²]^{1/2}` and compare the fitted floor with prediction; a discrepancy above 20 per cent fails the model.
5. Replication: an independent full-wave or beam-propagation model returning the same floor and `D` scaling; tolerance 20 per cent on the floor.

## 15. Principal remaining uncertainties

Most likely to kill the idea: the prefactor. The floor sits within a factor of two of the beamwidth convention, and a real array's beam is not diffraction-limited — phase errors, apodisation and loss broaden it, raising the floor but also weakening the `D` scaling the test relies on. Second, a separate receive telescope removes the range penalty, leaving only the ceiling. Third, if material dispersion dominates, `|dθ/dλ|` is not the design variable and the aperture test becomes a test of the waveguide.