Urban Radio Astronomy with the L-Band H1 Helix

Radio astronomy is usually associated with remote, radio-quiet sites and large dishes. Volker Tympel (University of Jena) is doing almost the opposite: detecting the 21 cm neutral-hydrogen line of our own galaxy from a balcony in the middle of the city — just below a ridge that carries a broadcast mast radiating several times 10 kW. He recently moved from a home-built grid-dish “balcony radio telescope” to our compact 10-turn L-Band H1 helix (NE-LBN-010), and shared his full setup and data with us. With his kind permission, here is what he is achieving.

All measurements, plots and photographs in this post are by Volker Tympel and are reproduced here with his permission.

nolle.engineering 10-turn L-Band H1 helix on a boom, with the Jena broadcast mast on the ridge behind — The 10-turn L-Band H1 helix (RHCP) on Volker’s balcony, with the Jena broadcast mast on the ridge behind.

The challenge: radio astronomy in a city

Neutral hydrogen radiates at 1420.4 MHz, and the signal is extraordinarily faint. Volker measures the galactic contribution at below −170 dBm/√Hz at 1420 MHz. Meanwhile the man-made noise on his balcony reaches above −30 dBm at 176 MHz right at the antenna, before any amplifier — roughly a 140 dB gap between the local interference and the signal he is after. The university’s own 2.4 m dish, he notes, can no longer cope with the interference at all.

Wideband 0 to 3000 MHz RF spectrum on the balcony: broadcast and radar interference picked up by the 10-turn H1 helix and a 1296 MHz Yagi — The RF environment on the balcony: a 0–3000 MHz sweep with the 10-turn H1 helix (black): broadcast and radar signals peak above −30 dBm across 100–1000 MHz, against a noise floor near −60 dBm. Red: a 1296 MHz Yagi for comparison.

From grid dish to helix

His first “balcony radio telescope” used two KrakenRF feeds illuminating a 90 × 60 cm grid dish, feeding an SDRplay RSPdx and his own C++ processing software, tuned to survive the urban RFI. Looking for something more compact — and ultimately mountable on the university’s 2.4 m dish — he ordered our 10-turn L-Band H1 helix (RHCP), tuned to 1420 MHz.

Two KrakenRF feeds on a 90x60 cm grid dish, Volker's earlier balcony radio telescope — The earlier setup: two KrakenRF feeds on a 90 × 60 cm grid dish, feeding an SDRplay RSPdx.

Why a helix, not a Yagi, at 1420 MHz

It comes down to the match at 1420 MHz. Every nolle.engineering H1 helix is individually tuned and its match verified on the bench to VSWR ≤ 1.1 at 1420 MHz, measured at the feed connector with no cable in between. Volker compared it with a 14-element Yagi: fine on its own 23 cm band, but badly mismatched at the hydrogen line, where it “sees nothing.” The helix gives him a repeatable H1 detection.

Antenna	At 1420 MHz (the hydrogen line)	On-sky result
nolle.engineering H1 helix	tuned to VSWR ≤ 1.1 (bench, at the feed connector)	detects the hydrogen line
14-element Yagi (23 cm band)	grossly mismatched (Volker measured VSWR ≈ 36)	sees nothing

Measured on Volker’s bench: at 1420 MHz the helix is matched, the 23 cm Yagi is not.

The signal chain

Recovering a signal below −170 dBm/√Hz under kilowatts of nearby broadcast energy takes a carefully staged front end. Volker’s chain runs:

10-turn H1 helix → N-type adapters → SMA
LNA #1: a cavity resonator tuned to 1420.4 MHz feeding a TQP3M9035 (+21 dB). This first amplifier is the critical one: it needs a noise figure below 0.5 dB and very high linearity (OIP3 > +39 dBm) so the broadcast carriers cannot drive it into intermodulation. A SAW filter ahead of the LNA killed sensitivity, so the cavity does the pre-selection instead.
3× cavity band-pass filter (−0.5 dB) with DC bypass — suppresses the 80–700 MHz broadcast band by more than 70 dB and a local 1453–1471 MHz source by 30 dB.
5 m satellite coax → LNA #2: a satellite line amplifier SVS2-01 (950–2400 MHz, +20 dB), compensating the cable loss without lifting the broadcast band.
Window feed-through and a bias-tee (12 V / 120 mA)
SAW filter at 1420 MHz (−3 dB) directly at the receiver input, to reject cable pick-up
SDRplay RSPdx, with Volker’s own C++ acquisition and processing software

The front end took real experimentation: an SPF5189Z ran too hot and too noisy, and a John-Fielding ATF34143 design self-oscillated once connected to the antenna. The TQP3M9035 with an inductively-coupled input cavity is what finally worked.

Results: reproducible galaxy transits

With the helix and cavity-tuned LNA, the galactic hydrogen line comes through reproducibly at about 0.5 dB above the noise floor (for reference, the KrakenRF feed on the 90 cm dish gives roughly 1–1.5 dB). Volker pointed the fixed antenna at two elevations — 65° and 35° — roughly due south, and let the Earth’s rotation carry the galaxy through the beam, twice per pointing. Each spectrogram below plots Doppler velocity (km/s) horizontally against time vertically, with signal level in colour; every frame is a 15-minute average of about 192,000 spectra.

So where on the sky does that signal come from? Because the antenna stays put and the Earth turns, each pointing slowly sweeps an arc across the sky. Drawn in galactic coordinates and coloured by the measured signal, the hydrogen brightens right where the beam crosses the plane of the Milky Way:

Sky map of the four pointings' paths, coloured by hydrogen-line signal, peaking on the Milky Way — Each pointing’s path across the sky (galactic coordinates), coloured by the measured H I signal. The dotted circles show the helix’s wide beam; the signal peaks where that beam crosses the Milky Way.

The four runs looked in very different directions — the night runs toward the crowded inner Galaxy (Aquila, Vulpecula), the midday runs out toward the sparse outer edge (Gemini, Monoceros), which is part of why the night data is a little stronger:

Top-down Milky Way schematic of the four look-directions, inner Galaxy versus outer edge, with peak velocities — The same four pointings seen from above the Galaxy. Velocities are the peak gas speed measured in each direction.

The pointing matters. Because different elevations look down different galactic longitudes, the hydrogen shows up at different Doppler velocities — the direct fingerprint of galactic rotation. (The longitudes work out to roughly 60°/190° at 65° elevation and 30°/230° at 35°, for the night and midday passes.) In the 65° night scan you can even make out a second, higher-velocity spiral arm around −90 km/s.

Plotted as signal against the hydrogen’s Doppler velocity, a single run is easier to read: a strong peak from nearby gas in the local arm, and a fainter, faster shoulder from a more distant arm. The antenna never moves — the Milky Way drifts through its wide beam as the Earth turns:

Left: hydrogen signal versus Doppler velocity with local and distant arm features. Right: fixed-beam drift cartoon — Left: one night run as signal vs. Doppler velocity — nearby gas near 0 km/s, a more distant arm near −90 km/s. Right: the helix is fixed; the Earth’s rotation carries the Milky Way through its beam.

HI line spectrogram, 65 degrees elevation, night — 65° elevation, night

HI line spectrogram, 65 degrees elevation, midday — 65° elevation, night

Measuring the grid dish’s beam with the Sun

Volker measures antenna patterns with the Sun: with the antenna fixed, letting the Sun drift through the beam traces the pattern directly. This drift-scan of his grid-dish setup (December 2025, at 16° elevation) shows a clean main lobe with the expected sidelobe structure at about ±8.7°. This is the pattern of his dish, not the helix: a 10-turn helix has a much broader beam — tens of degrees wide — and none of these sharp sidelobes.

Antenna beam pattern measured by a solar drift-scan — A solar drift-scan (Sonnendurchgang) of Volker’s 90 × 60 cm grid dish — not the helix: the Sun sweeping through a fixed beam traces the antenna pattern directly, main lobe plus sidelobes.

Calibration

The whole front end has its own frequency response — and the final SAW filter even drifts for the first couple of hours — so every spectrum is flattened against reference lines: 50 Ω minimum/maximum terminations and blank-sky references at each elevation. Correcting locally on the running minima and globally to a 0 dB offset is what makes a 0.5 dB signal measurable.

Frequency-response reference lines around 1420 MHz — Reference lines around 1420 MHz: 50 Ω min/max terminations and blank-sky references used to flatten every spectrum.

What’s next

Volker is taking it further. He is characterising several new front-end amplifiers, and exploring simultaneous RHCP + LHCP reception: two counter-rotating 10-turn helices would, on his estimate, buy about 0.7 dB (a factor of 1.4) — at the cost of a second amplifier chain. A shorter, capacitively-loaded feed (like the KrakenRF PCB dipole) might also tame the long helix wire’s tendency to collect city noise. The longer-term goal is to move the front end onto the university’s 2.4 m dish.

Thanks

Thanks to Volker Tympel for sharing his work and letting us publish it. It’s a good example that H1 radio astronomy doesn’t need a dark-sky site or a big dish. If you would like to try the hydrogen line yourself, the antenna he used is our L-Band H1 Helix, and our earlier practical guide to H1 radio astronomy is a good place to start.