Drift Scan

The DSES 60-foot radio telescope silhouetted against a twilight sky at the Haswell, Colorado site, with the operations building lit beside it.
The 60-foot dish at Haswell, Colorado at dusk — pointed near the zenith, recording around the clock.

Through most of August 2026, our 60-foot dish at Haswell, Colorado sat motionless, pointed near the zenith, while the Earth’s rotation swept the sky through its beam. A small computer at the dish recorded a spectrum of the 21 cm hydrogen line around the clock, unattended — first a six-day wide-band survey, then, after a receiver upgrade, a deeper survey at twice the spectral resolution.

What came back was the structure of our own galaxy — and a telescope measured, from the sky itself, to a hundredth of a degree.

Waterfall plot of seven days of 21 cm hydrogen-line spectra showing two bright bands where the Milky Way’s plane crosses the telescope beam.
Seven days of hydrogen spectra from the deep survey, stacked by sidereal time. The two bright bands are the plane of the Milky Way drifting through the beam — once in Cygnus, once toward the galactic anticentre — repeating identically every sidereal day.

The run, by the numbers

  • 26,000 spectra across the month — not one missing
  • Hydrogen at 0.10 km/s velocity resolution, with the receiver’s own artifacts moved entirely out of the band
  • The radio galaxy Cygnus A detected on all seventeen observing days, its transit time agreeing to ±2 seconds across three receiver setups
  • Telescope beam measured at 0.69°, sensitivity about 4,000 Jy — using the sky itself as the ruler
  • Dish pointing pinned to a hundredth of a degree: one fixed 0.15° offset remains, and the next campaign is designed to remove it

Watching the spiral arms

Hydrogen atoms throughout the galaxy broadcast at exactly 1420.4058 MHz. Because the gas orbits the galactic centre, the line reaches us Doppler-shifted — and the shift tells us how fast each cloud moves and, through the geometry of galactic rotation, roughly where it is. When the beam crossed the plane in Cygnus, the spectrum resolved into separate components: local gas near zero velocity, and gas in the outer spiral arms rushing toward us at 45 and 70 km/s.

Hydrogen-line velocity profiles from two surveys overlaid, showing multiple gas components at the Cygnus and anticentre crossings.
Hydrogen profiles at the two galactic-plane crossings, from the wide survey (gray) and the deep survey (teal). In the Cygnus direction the line splits into local gas and outer-arm components at −45 and −70 km/s — spiral structure, measured from the Colorado plains.

A radio galaxy, every day at the same second

Averaging each spectrum away from the hydrogen line turns the same data into a continuum survey — a map of total radio power around the sky. One object dominates: Cygnus A, one of the strongest radio sources in the sky, powered by a supermassive black hole 600 million light-years away. It swept through our beam every sidereal day, at the same second, at the same strength.

24-hour radio continuum plot showing a strong spike from Cygnus A and broader emission from the galactic plane.
The whole sky at our declination, folded over five days of sidereal time. The spike at 20 hours is Cygnus A; the smaller bumps beside it are the Gamma Cygni supernova remnant and the Cygnus X star-forming region. The broad rise at 3–5 hours is the plane of the Milky Way itself.
Seventeen days of Cygnus A transit timing and amplitude measurements showing a constant 48-second offset from prediction.
Seventeen days of Cygnus A transits. Top: transit time against the precise astronomical prediction — a constant 48-second offset, steady through three receiver configurations and a mount-maintenance visit (dotted line). Bottom: transit strength by day. The offset is the telescope’s one remaining pointing unknown: 0.15° of boresight, now precisely measured.

The sky calibrates the telescope

Cygnus A is effectively a point source, so its transit measures the telescope itself: the width of the curve gives our beam (0.69°), its strength gives our sensitivity (about 4,000 Jy system equivalent flux density, measured directly by the September calibration move), and its timing checks the whole chain — clocks, software, mount — to the second. In August that chain closed: a maintenance team read the dish’s position encoders on-site, and the encoders and the sky now agree. The mount repositions to twenty arcseconds, and the one remaining unknown — a fixed 0.15° east-west boresight offset — is measured so precisely that the next campaign can be designed around removing it: stepping the dish across Cygnus A night by night to map the beam and pin the pointing for good.

September update: the mount, measured

In early September the dish was deliberately moved half a degree in elevation to put Cygnus A at the center of the beam — the first step of the calibration campaign. That one move measured the telescope’s true sensitivity (about 4,000 Jy), revealed that the mount sags slowly in elevation — about 13 arcseconds per day, confirmed three independent ways, including a rise in the daily Cygnus A signal that was predicted before it was observed — and traced an apparent encoder offset to a stale mount calibration: the control system’s site coordinates proved correct — confirmed against a GPS receiver at the site — and a single +0.123° offset in the encoder-to-elevation conversion, the signature of a mount calibration not run in over a year, explains the sky, the encoder, and the control display at once. Every one of the campaign’s twenty-nine daily measurements fits a single model.

Twenty-nine days of Cygnus A transit measurements across three telescope configurations, all matching one model of a fixed beam with slow elevation sag.
Every Cygnus A transit of the campaign against one model: a fixed beam whose elevation sags slowly, at a rate measured from the mount’s own readouts. The decline, the jump at the September 5 move, and the rise after it — predicted before it was observed — all follow the same dashed line.

A receiver that got better mid-campaign

Halfway through the month we upgraded the receiver configuration so that its self-generated artifacts land outside the science band entirely. The payoff is right at the heart of the measurement: the hydrogen line’s rest velocity, which previously had to be patched around an artifact, is now a clean measurement at 0.10 km/s per channel.

Two overlaid hydrogen spectra near zero velocity showing a blanked artifact region before the receiver upgrade and clean data after.
The hydrogen rest-velocity region before (gray) and after (teal) the receiver upgrade. What was a repaired hole in the band is now well-sampled data.

What’s next

The scan is running right now — the dish keeps recording a spectrum every 64 seconds while you read this. Next: the pointing-calibration campaign. On the next site visit the dish moves half a degree, placing Cygnus A dead on-axis; each following visit steps the beam across the sky’s calibrator sources to map the beam and remove the last fraction of a degree of pointing error. After that: longer hydrogen surveys, and — the same dish, the same receiver — pulsar observations.

How drift scanning works

A drift scan is radio astronomy at its simplest and most robust: park the telescope, let the Earth’s rotation do the scanning, and record continuously. Every object at the beam’s declination passes through once per sidereal day, so a fixed dish surveys a full ring of sky every 23 h 56 m — with perfectly repeatable geometry and no tracking errors. It is the oldest survey technique in radio astronomy, and it remains one of the best ways to characterize a telescope and its site.

Reports

  • August 2026 Hydrogen Drift-Scan Campaign (Report 2, Revision C, September 2026) — the definitive analysis behind this page: both surveys, seventeen days of Cygnus A, the telescope’s true pointing, the calibration-campaign plan, and the September addendum with the calibration-move results. Supersedes the Six-Day report.
  • Lowering the Barrier to Radio Astronomy — DSES Science Meeting, September 28, 2026 (slides, PDF): the Workbench 1.6.0 story built with its users — the Haswell drift scans, Ray’s and Rich’s home stations — and the EVE-26 modem.
  • Status Report (6 December 2025)
  • DSES-60 1420 MHz Early Study (21 November 2025)

Observations and analysis by DSES members using the DSES Radio Astronomy Workbench (formerly the DSES Spectrum Analyzer), our in-house open-source software — source on GitHub, releases at gpstime.com — running on a Raspberry Pi at the Haswell site. Page updated September 2026.