The $150 Hydrogen-Line Receiver, End to End

People write in asking what the minimum is. Not the minimum that would technically work on a good night with a lot of luck, but the minimum that will reliably show the hydrogen line to someone who has never done this before. This is that list. Dee’s receive-chain paper describes what runs at Sixmile Run. This is the version that fits in a hatchback and costs about what a decent pair of boots does.

I keep the cost spreadsheet here, so I’ll be specific about money where I can and honest where I can’t.

What you’re building

Four things in a line: a horn antenna, a low-noise amplifier screwed directly onto the horn’s connector, a few metres of coax, and an RTL-SDR dongle plugged into a laptop. The LNA gets its power up the coax from the dongle. That’s the whole receiver.

The two parts that do the real work are the RTL-SDR Blog V3 (listed at $35 to $40) and the Nooelec SAWbird+ H1 (listed at $45 to $55). Between them that’s roughly $80 to $95. The rest of the $150 is the horn, the adapters, and the coax, and how much of that you actually spend depends heavily on what’s already in your garage. If you have a drawer of SMA adapters and a roll of foil tape, you’re well under. If you’re buying every piece new, plan on spending close to the full amount.

Parts list

Electronics

• RTL-SDR Blog V3. It has a software-switchable bias tee that puts about 4.5 V up the antenna port, which is what powers the LNA. That feature is the reason we name this dongle specifically instead of “any RTL-SDR.”
• Nooelec SAWbird+ H1. A filtered LNA built for 1420 MHz. The SAW filter is the important half: the band around the hydrogen line is shared with GPS, Iridium and Inmarsat downlinks a short way up the spectrum, and a cheap dongle with nothing in front of it will happily overload on those.

Antenna

• Two sheets of foam board, the kind sold for school posters, at least 1 m on the long side. Or thin aluminium sheet if you have a brake and patience.
• One roll of aluminium foil tape (HVAC tape, the real metal kind, not the plastic “foil-look” kind).
• A panel-mount connector for the probe: SMA or N, female, with a solder cup or a long centre pin.
• A short length of stiff copper wire or brass rod, about 2 mm, for the probe.

Cables and adapters

• A short SMA male-to-male jumper, or a direct adapter, to join the probe connector to the SAWbird’s input.
• A few metres of coax from the SAWbird’s output to the dongle. Because this run comes after the LNA, it does not need to be good cable. Bench 3 below shows why.
• Whatever adapters match the connectors you ended up with. Count them before you order. Everyone forgets one.

Building the horn

The horn is a pyramid with its point cut off: a rectangular box at the narrow end (the throat, which is really a short length of waveguide) flaring out to a larger rectangular mouth. The throat box holds the probe. The flare gives you gain and narrows the beam.

Size. Use the dish and horn calculator set to 17 dBi on a WR-650 throat, which is Bench 1. It gives a mouth of about 675 by 514 mm and a flare roughly 480 mm long. That’s what we’d build first. If you need it smaller, 15 dBi (Bench 2) shrinks the mouth to about 542 by 404 mm with a 262 mm flare. Don’t go much below that; you are trading away the beam’s ability to pick out the Galactic plane from everything around it.

Throat. WR-650 is a standard waveguide size with an inside broad wall of 165.1 mm and a narrow wall of 82.55 mm. Its cutoff for the dominant mode is about 908 MHz and the next mode doesn’t propagate until about 1816 MHz, so at 1420 MHz it carries exactly one mode, which is what you want. Make the throat box with those inside dimensions, closed at the back, open at the front where the flare begins. Give it some length: a box at least 150 mm deep is easy and leaves room to adjust the probe.

Probe. The probe is the bit of wire that actually picks up the field inside the waveguide. Two numbers place it:

• Length: a quarter of the free-space wavelength. At 1420.406 MHz the wavelength is 211.1 mm, so the probe sticks 53 mm into the box from the broad wall.
• Distance from the back wall: a quarter of the guide wavelength, not the free-space wavelength. Inside WR-650 at 1420 MHz the guide wavelength stretches to about 274 mm, so the probe sits about 69 mm in front of the back wall.

Mount the connector in the centre of one broad wall at that distance, solder the wire to its centre pin, and trim to length. Cut a few millimetres long and trim down if you have a way to measure return loss. If you don’t, cut to 53 mm and leave it.

Flare. Cut four trapezoids from the foam board so that they meet the throat box at one end and the mouth dimensions at the other, with the flare length the calculator gave you. Line every inside surface with foil tape before you assemble, overlapping the strips by a centimetre so there’s no gap running along the length of the horn. Tape the outside joints. The foil carries the currents; the foam is just there to hold it in shape.

The foil-tape horn is not a precision instrument. It will be a dB or so off the calculator’s gain, the corners will be soft, and if it gets rained on it will be finished. None of that stops it from showing you the hydrogen line.

Wiring it up

Probe connector → SAWbird+ H1 input. Keep that joint as short as you physically can. Every adapter and centimetre of cable before the LNA adds noise in full; Bench 3 counts 0.2 dB of adapters here and that alone is about 14 K of the receiver’s 60 K.

SAWbird output → your coax → RTL-SDR Blog V3. Mount the SAWbird on the horn itself with a bit of tape or a zip tie so its weight isn’t hanging off the probe connector.

Power. Turn on the V3’s bias tee. On Linux and macOS that’s rtl_biast -b 1 from the RTL-SDR Blog driver tools; most Windows SDR software has a bias-tee checkbox in the device settings. The SAWbird has an indicator LED, and if it doesn’t light, nothing downstream will work. Check this first, every time.

One warning I’ve learned the expensive way: the bias tee puts DC on the antenna port. Never plug a bias-tee-enabled dongle into something that shorts DC to ground, such as some antennas or certain test equipment, without checking first.

Sanity checks before you point it at the sky

1. Plug and power. Dongle seen by the computer, bias tee on, LNA light on.
2. Noise floor rises with the LNA on. Tune to 1420.4 MHz with any spectrum software and note the floor with the bias tee off. Turn it on. The floor should jump by many dB, and you should see the gentle hump of the SAW filter’s passband appear. If nothing changes, the LNA isn’t getting power.
3. Hand test. Point the horn at open sky, then hold your hand over the mouth, close to the throat. The floor rises: your hand is near 300 K and the sky is much colder. It’s a crude test of a real thing, and it tells you the horn is coupling energy into the probe.
4. No towers in the beam. Point it away from buildings, cars and the house. Ground and walls are warm noise sources and they fill a wide beam easily.

If all four pass, you’re ready to observe. The observing session itself is LAB-009.

Why the cheap coax is fine

Bench 3 runs this chain through the noise calculator with 5 m of RG-8X after the LNA. Receiver temperature comes out at about 60 K, essentially the same as with low-loss cable, because the LNA’s gain divides down everything after it. Move the LNA to the laptop end of that same cable and the receiver goes to about 220 K, and the integration time for the same signal goes up nearly sevenfold. That’s the entire argument of LAB-011, and it’s why the only money in this build goes on the two electronics parts.

What this will and won’t do

It will show you the hydrogen line. On the inner Galactic plane, with a few minutes of integration and a reference spectrum to divide out the bandpass, the line should stand clearly above the noise. It won’t give you fine angular resolution: a 17 dBi horn has a beam roughly 20 to 25 degrees across, so you see broad swaths of the Milky Way, not individual clouds. LAB-010 covers what you give up and what you gain by moving to a dish.

For a printed reference that covers this class of receiver in more depth than we can here, The Radio Sky and How to Observe It is still the one we hand people.

---

The equipment and book links are affiliate links. If you were going to buy the dongle and the LNA anyway, buying through them sends a few dollars to the hosting bill, which I am the one who has to look at every month.