FAST Radio Telescope (3D)

A Blender-built 500 m bowl with a 25 m reference dish: set area, noise and a dispersed burst, then dedisperse it and read the S/N.

Switch to 2D

Presets

Playback

Dish

Burst

Integration

Camera

Layers

Quality

Model check

The science

What the model computes

A radio telescope is a bucket for photons: it gathers power from a source in proportion to its collecting area and competes against its own receiver noise. A fast radio burst (FRB) adds a second problem, because the cold plasma between the source and us delays low frequencies more than high ones. The page chains the three real steps: collecting area to sensitivity, dispersion delay, and dedispersion with a matched filter.

Equations

  • Area and gain. Aeff = η π (Dill/2)², gain G = Aeff / 2k in K/Jy, SEFD = 2 k Tsys / Aeff (1 Jy = 10-26 W m-2 Hz-1). FAST has a 500 m spherical bowl but only about 300 m of it is illuminated at a time, which is why the slider "illuminated fraction" exists.
  • Radiometer equation. Noise on a flux-density estimate averaged over bandwidth B and time τ with two polarisations: σS = SEFD / √(npol B τ).
  • Dispersion. Δt(f) = kDM DM (f-2 − fhi-2) with kDM = e²/(2π me c) · (1 pc) = 4.1488 ms GHz² cm³/pc. Across 1.0-1.5 GHz, DM = 400 gives 0.92 s. Inside one channel the pulse is smeared by 8.3 μs · DM · ΔνMHz · νGHz-3.
  • Matched filter. After dedispersion at the true DM, the burst has effective width weff = √(w² + tchan² + tsamp²), peak S w / weff (fluence conserved) and a boxcar of that width gives S/N = Sobs √(npol B weff) / SEFD.

Checks (run in this page)

  1. The FAST preset has 70 686 m² illuminated and a gain near the published 16 K/Jy (the aperture efficiency 0.63 is tuned to that figure, Jiang et al. 2020).
  2. A Monte-Carlo of 300 noisy 512-channel boxcar sums reproduces the closed-form S/N.
  3. kDM is rebuilt from e, me, c and the parsec, and a brute-force trial-DM scan of a simulated burst peaks at the injected DM.
  4. Averaging 64 noise samples lowers the rms by √64 = 8, the root of the radiometer equation.

Honest simplifications

White Gaussian noise only (no RFI, no baseline ripple, no gain drift); constant system temperature and aperture efficiency (η = 0.63 is a single tuned number, not a measured curve); a flat-topped burst with flat spectrum (real FRBs are often band-limited); no scattering tail, no incoherent-versus-coherent dedispersion details, no pointing loss, no beam shape. The 3D bowl is a 1 unit = 10 m drawing of a FAST-like geometry, not survey data, and the dispersed packets are drawn with the delay stretched to a few seconds so you can see the order. Independent educational model, not affiliated with any observatory.

What to try

  • Load the 25 m preset: the same burst falls below S/N 5, the 144 times larger illuminated area of FAST sees it clearly.
  • Raise DM to 1000 with a narrow burst: channel smearing dominates the width and the S/N collapses even though the flux is unchanged.
  • Look at the DM scan: it peaks at the true DM and falls off on both sides, which is how real searches find the DM.
Loading model...
Loading Blender model...
Dynamic spectrum: the dispersed burst in noise
Dedispersed profile with boxcar matched filter
Trial-DM scan: S/N peaks at the true DM
Steady-source S/N versus integration time (sqrt t)

FAST Radio Telescope: Collecting Area and Fast Radio Burst Detection

Compare a 500 m spherical bowl radio telescope, of which about 300 m is illuminated at a time, with a 25 m reference dish and see why collecting area decides whether a faint fast radio burst is found. The simulation sweeps a dispersed burst across a 1.0-1.5 GHz band, dedisperses it with a trial-DM scan, applies a boxcar matched filter and compares the measured signal-to-noise ratio with the radiometer equation, with four in-page numerical checks. The bowl is a Blender-built FAST-like scene drawn at 1 unit = 10 m.

Made with 🟠 Blender · Three.js