The Waterfall Scrolled Green and Nothing Burned

It’s 2:47 AM and I’ve been staring at a waterfall display for three hours, waiting for the sky to ping back.

Nothing.

Well—not nothing. The GRAVES radar in France is right there on 143.050 MHz, a distant whisper of a military space surveillance system bouncing off the ionosphere and somehow reaching my RTL-SDR dongle in Alberta. That part works. The antenna works—a homebrew 6-element Yagi I aimed roughly east. The trigger circuit works; I tested it with a tone generator and watched the optocoupler fire the camera’s remote shutter jack. The Python script works, mostly. It detects amplitude spikes, debounces them, logs timestamps.

Everything works except the part where meteors cooperate.

Thirty-seven hobbies in, and you’d think I’d have learned that the universe doesn’t care about my timeline. When I mounted that APRS tracker on a foam wing for APRS Skytrace Sculptures, the packets came when they wanted to—every thirty seconds, or sooner if the tracker decided speed had changed enough to matter. I was interpolating between data points I didn’t control. This is worse. This is waiting for grains of sand to burn up eighty kilometres overhead and leave ionized trails that happen to bounce French military radar toward my specific patch of frozen yard.

The theory is elegant: meteors create plasma tubes in the E-layer of the ionosphere, between 80 and 120 km altitude. These tubes act as mirrors for VHF signals. A radar beam that would normally sail past the horizon gets briefly bent back toward Earth. If you’re listening at the right frequency, you hear a ping—a momentary reflection, Doppler-shifted by the meteor’s velocity, lasting anywhere from a tenth of a second to maybe three seconds for the big ones. The jargon: “underdense” trails are quick blips; “overdense” trails from larger meteors persist long enough that you could theoretically catch the fireball itself if your trigger latency is low enough.

My latency is not low enough. I know this because I’m watching the waterfall scroll and I can see faint diagonal streaks—aircraft, probably, showing that classic Doppler slope as they move across the reflection path—but no vertical blips. Or maybe there are vertical blips and my threshold is wrong. Or maybe I’m in a sporadic-E event, where summer ionospheric weirdness makes the beacon audible continuously and swamps everything else. It’s not summer. But maybe my gain is too high. Or too low.

The debugging loop is maddening because I can’t generate test data. I can’t schedule a meteor.

Around 1:30 AM I thought I had one. A spike on the waterfall, a quick chirp in the headphones, and the camera clicked. I got excited. I checked the frame.

Black. Nothing but black, and three airplane contrails from earlier that I’d forgotten were still in the sky.

The spike was probably an aircraft passing at just the right angle to look impulsive instead of sloped. My detection script has no way to distinguish them. It’s just looking for amplitude. I could add Doppler analysis—filter for near-vertical signatures only—but that’s another layer of code, and right now I’m too tired to think about FFT bins.