Add moonbounce post with 9 interactive scenes, rename cascaded to stacked

content/blog/moonbounce.md

@@ -0,0 +1,105 @@
++++
+title = "Moonbounce: the absurd thing that works"
+date = 2026-05-01
+draft = false
++++
+
+<link rel="stylesheet" href="/css/scene.css">
+
+<div class="scene-host" data-scene-host>
+
+Take an antenna. Point it at the Moon. Transmit. A microscopic fraction of what you sent makes it across 384,000 km of empty space, smacks into the pocked gray surface, scatters off in every direction because the Moon is a deeply unimpressive mirror, and a microscopic fraction of that comes back. A few kilowatts leave; a few attowatts return. On the other side of the planet, someone pointed at the same Moon hears it.
+
+This is moonbounce. As a method of having a radio contact it is absurd in a way that very little else in the hobby manages. You can sit a non-ham at a reasonable HF station and explain how a signal gets from Texas to Argentina, and if they squint they'll accept it. Try to explain moonbounce with a straight face and you look slightly unwell.
+
+It does, annoyingly, just work. You need respectable antennas, respectable power, and a computer willing to spend about a minute per transmission sifting what came back from what didn't.
+
+The whole story is really the link budget, so what follows is the budget broken into the scenes that make it what it is.
+
+## The geometry
+
+Two stations on opposite sides of a rotating Earth. One Moon, tidally locked so the same face points our way, drifting through a 27-day orbit that is tilted about 5 degrees off the ecliptic for good measure. For a contact to happen, both stations have to have the Moon above the horizon at the same instant. Some days that's a few hours. Some days that's zero.
+
+<div id="scene-geometry"></div>
+
+The Moon dot is cheerfully oversized and the orbit is nowhere near to scale, because to scale it would be a single pixel most of the way across the page and you'd learn nothing. The rule is the useful part: mutual visibility exists only when the Moon is above the horizon at both ends at once. For a North American station working Europe that's maybe a couple of hours a day. For stations further apart, there are nights where the window simply isn't there, and you do laundry instead.
+
+Serious EME operators keep [ephemeris data](https://en.wikipedia.org/wiki/Ephemeris) within arm's reach and check common-window charts before firing the station up, the way pilots check weather.
+
+## How far the Moon is on any given night
+
+The Moon's orbit isn't circular. Perigee is around 356,000 km, apogee around 406,000 km, and the eccentricity shoves the distance around by roughly fourteen percent over a month. That's a round-trip path-length difference of about a hundred thousand kilometers between best and worst, and since moonbounce path loss goes as distance to the fourth power (once out, once back), it costs you about 2.3 dB extra at apogee compared to perigee.
+
+<div id="scene-distance"></div>
+
+Serious EME operators watch for perigee nights the way HF operators watch the solar flux. Same antenna, same amp, same point on the sky, but the signal is 2 dB stronger simply because the Moon couldn't be bothered to be as far away that week.
+
+## The obscene amount of path loss
+
+Free-space loss at 144 MHz over 384,000 km one way is about 187 dB. Round trip doubles that to 374. The Moon isn't a perfect reflector either: its radar albedo is somewhere around 6 or 7 percent at VHF and UHF, which scoops out another 12 dB. The number you actually put in a link budget lands near 252 dB at 144 MHz, 259 at 432, and 271 at 1296.
+
+<div id="scene-pathloss"></div>
+
+For comparison, a quiet VHF terrestrial contact might be 130 dB from antenna to antenna. Moonbounce is roughly a hundred and twenty decibels worse than a normal hobby. Every one of those decibels has to come from somewhere: antenna gain, transmit power, receiver noise figure, integration time in the demodulator. There are only so many places to find 120 dB, and after you've looked in all of them you start looking in sillier ones.
+
+## Why the Moon is not a mirror
+
+If the Moon were a polished metal sphere it would reflect specularly and moonbounce would be a different sport. The real Moon is rock, dust, craters, and mountains on every scale from millimeters to kilometers. At the wavelengths hams use (2 m, 70 cm, 23 cm, and a few shorter) the surface is optically rough, so the energy scatters off in every direction it feels like.
+
+What comes back to Earth is the coherent sum of reflections from everywhere on the lit disk, each with its own path length and its own Doppler shift relative to the rest, because the Moon is not quite rotating in lockstep with Earth. The apparent rocking is called [libration](https://en.wikipedia.org/wiki/Libration), and although the optical excursion is less than 8 degrees over a month, it's enough that different surface patches are moving at slightly different radial velocities. The returned signal is the sum of all those patches and the sum drifts in and out of constructive and destructive interference as they slide past each other.
+
+<div id="scene-libration"></div>
+
+That produces a characteristic fading pattern: deep, not quite random, with timescales from a fraction of a second at 10 GHz to tens of seconds at 50 MHz. If you've ever listened to a strong moonbounce signal in the raw, you've heard it: a CW note that ought to be steady instead shimmers and hollows out and comes back, over and over, while the operator just keeps sending. It's also why the transmissions are 30 or 60 seconds long in the first place. You need to average over several fading cycles for anything to decode reliably.
+
+## Doppler
+
+The Moon moves. You move. The round-trip path length changes constantly, so the frequency you get back is not the frequency you sent. At 144 MHz the peak offset from Earth rotation alone (the self-Doppler, for your own round-trip) is roughly ± 400 Hz near moonrise and moonset and passes through zero when the Moon is at its highest. At 432 MHz it's three times that. At 1296, nine. By 10 GHz it's several kilohertz, and finding your own echo is a genuine tuning exercise instead of an afterthought.
+
+<div id="scene-doppler"></div>
+
+Operators either tune for it in real time or let the computer do it. JT65 and Q65 assume the Doppler will drift during the 47-second transmit window and accommodate a linear drift inside the decoder, which is one of the small but very grown-up things those modes do that nobody notices until they try to roll their own.
+
+## Polarization is complicated
+
+At VHF, the [ionosphere](https://en.wikipedia.org/wiki/Ionosphere) rotates the plane of linear polarization ([Faraday rotation](https://en.wikipedia.org/wiki/Faraday_effect)) by anywhere from a few degrees to several full turns, depending on total electron content, path geometry, and frequency. At 144 MHz it can be huge and unpredictable. At 432 MHz it shrinks. By 1.3 GHz and above, it's mostly not a problem.
+
+On top of that there's a purely geometric offset: the plane that's "vertical" at one station is not the plane that's "vertical" at the other when the Moon is the apex of the triangle. This "spatial polarization offset" is deterministic and can be computed from the pointing angles at both ends, but it isn't zero, and it's exactly the sort of thing that quietly costs you a contact.
+
+<div id="scene-polarization"></div>
+
+Drag the receive antenna's rotation angle. When the two agree, mismatch loss is zero. When they're 90 degrees apart, the signal vanishes into the cross-polarized null, which is why any sensible 2 meter EME station either uses circular polarization or puts a rotator on the linear array. Circular costs you a flat 3 dB against a randomly polarized return, and in exchange it doesn't care about rotation at all, which turns out to be the trade you want.
+
+## Modes and thresholds
+
+Everything above was about getting the signal to and from the Moon. What happens once the faint echo reaches the receiver is a question for the demodulator. SSB voice needs about +10 dB SNR in a 2.5 kHz bandwidth, which in EME terms is not a mode so much as a dare, and the handful of stations that have ever done it have antennas visible from orbit. CW needs about 0 dB in a few hundred Hz, which a well-equipped single-yagi station can sometimes manage under ideal conditions. JT65 integrates for 47 seconds and decodes down around −24 dB in 2.5 kHz. Q65, the current weak-signal mode of choice for EME, goes another 3 dB lower.
+
+<div id="scene-modes"></div>
+
+The signal in the waterfall above is always the same power. What changes is the mode threshold. Watch SSB disappear into the noise while JT65 and Q65 sit comfortably above where they need to be. This is the whole reason every serious EME operation now lives inside WSJT-X. You trade real-time interactivity for sensitivity, and the sensitivity is the only thing that makes the link close.
+
+## The link budget, end to end
+
+Roll it all up. Transmit power plus transmit antenna gain minus path loss plus receive antenna gain gives you the received signal power. Subtract the system noise and you get the SNR. Compare it to the mode threshold and either the link closes or it doesn't.
+
+<div id="scene-budget"></div>
+
+A typical "starter EME" 2 meter station looks something like 1 kW into a 4-yagi array (around 18 dBd each end), a 0.5 dB LNA, and a Q65 decoder. That puts you maybe 5 to 8 dB above threshold on a decent night, which doesn't sound like much and isn't, but it's enough to actually have contacts with other stations that size. A single yagi and a full gallon puts you right at the threshold, where you'll hear plenty but almost nobody will hear you. A big-gun setup with a stacked eight-yagi or a 10-foot dish on a higher band pulls you 15 to 20 dB above threshold and turns the mode from a lottery into something closer to a scheduled hobby.
+
+## The ground gain trick
+
+One more thing, because it's free. When the Moon is near the horizon and your antenna is mounted above actual ground, the signal arrives via two paths: direct, and bounced off the ground. At the right elevation angle the two arrive in phase at the antenna and give you up to 6 dB of "ground gain" on top of whatever the antenna alone can do. At other angles they cancel, and you get a deep null that looks like somebody unplugged the coax.
+
+<div id="scene-groundgain"></div>
+
+Most moonrise and moonset EME sessions ride this lobe on purpose. Which means the operators who can't afford an elevation rotator and have to settle for pointing at the horizon aren't as disadvantaged as they sound. The ground is doing some of the heavy lifting for them.
+
+## What this is really
+
+Moonbounce is a hobby that asks amateurs to overcome several hundred decibels of loss using legal power and reasonable antennas. It works because every piece of the link has been optimized to the limit of what amateurs can currently do, from antenna design to noise figure to the signal-processing modes, and then the whole thing has been given thirty or more seconds of integration time to squeeze a decode out of what little returns.
+
+The first moonbounce contacts happened in the early 1950s with military surplus gear and a lot of patience. Sixty years later, with a laptop running WSJT-X and a reasonably sized yagi array, the same contacts happen routinely on evenings the Moon is up. The absurdity of it hasn't budged. Only the cost of admission has, which has quietly dropped to where a person with a normal yard can reasonably attempt it. Which, if you stop and think about what's actually happening, is not a small thing.
+
+</div>
+
+<script src="/js/moonbounce.js" defer></script>

content/blog/stacked-noise-figure.md

@@ -1,5 +1,5 @@
 +++
-title = "Noise figure, cascaded: why your LNA goes where it goes"
+title = "Noise figure, stacked: why your LNA goes where it goes"
 date = 2026-04-29
 draft = false
 +++