Downlink Budget: The Antenna Choice That Cost a Satellite Team 11 Days

Archi4 min read

satelliteshardwarespace

Contents
  1. The Link That Would Not Close
  2. What the Bench Actually Showed
  3. The Decision, Reconstructed
  4. Eleven Days of Recovery
  5. The Test We Should Have Run First
  6. What This Means for Smaller Operators
  7. A Short Checklist From the Bench

The test bench was a folding table in a leased office park outside Toulouse. On it: a software defined radio, a laptop running GNU Radio, and two antennas separated by a 40 meter run of coax. The team had a 3U cubesat in orbit with an S-band downlink at 2.2 GHz and a budget that had already been trimmed twice by the customer. They needed to pick a ground station antenna before the next launch window.

The decision looked trivial on paper. A 1.2 meter parabolic dish quoted at roughly 14,000 euros, or a 60 cm flat panel array at about 6,000. The dish had more gain. The panel was cheaper, lighter, and could be bolted to a rooftop without a structural survey. The team chose the panel. Eleven days later, they were still trying to make the link work.

What the Bench Actually Showed#

We reproduced their test setup over two afternoons. The first surprise came from the coax. At 2.2 GHz, 40 meters of RG-8 style cable loses several decibels, more than the panel's price advantage covered. The team had modeled the antenna in isolation, not the antenna plus feed line plus connector losses plus the receiver noise figure.

Here is the arithmetic they missed. Panel gain was quoted at 18 dBi. The dish was 29 dBi. That 11 dB gap is enormous in link budget terms. Cable loss ate another 3 to 4 dB on the panel setup before the signal ever reached the radio. The satellite transmitter put out 2 watts, and the slant range at low elevation angles stretched the path loss beyond what the panel could carry.

We ran the numbers at 20 degrees elevation, the point where passes usually start producing usable packets. The panel closed the link only above roughly 35 degrees elevation under clear skies. The dish, with shorter low-loss coax and a mast-mounted LNA, closed it at 12 degrees.

The Decision, Reconstructed#

In conversation with two of the engineers, the reasoning was clear and understandable. The panel was a known quantity in Wi-Fi and point-to-point work. It shipped in two weeks. The dish required a mount, a permit, and a civil engineer's sign-off on the roof. The cheaper option also meant more passes per day were theoretically possible because the antenna needed no repositioning.

That last assumption failed during testing. A fixed panel pointed at a specific azimuth covers a narrow slice of sky. The team's orbital inclination put most useful passes outside that slice. They ended up with a panel pointed at the wrong arc and a satellite transmitting into a null.

The panel was not a bad antenna. It was a bad fit for an orbit nobody had drawn on a whiteboard.

Eleven Days of Recovery#

The team kept the panel but added three things over the following week and a half. A mast-mounted low noise amplifier cut the effective noise figure by roughly 1.5 dB. They replaced 30 meters of the coax run with a shorter, lower-loss cable, recovering close to 2 dB. And they wrote a simple pass scheduler that logged every contact attempt with elevation, azimuth, and received signal strength.

By day 11, the link closed at 22 degrees elevation on a clear pass. Rain still killed it. The recovery work cost about 4,000 euros in parts and labor, which brought the total spend close to what the dish would have cost on day one.

The Test We Should Have Run First#

The team now runs a link budget spreadsheet before any hardware order, and they insist on measured antenna patterns rather than datasheet gain figures. Their bench notes list four numbers that matter: transmit power, cable loss, receiver noise figure, and minimum elevation angle. Everything else is secondary.

What This Means for Smaller Operators#

Ground station choices get treated as procurement details. They are not. For a cubesat operator with a tight schedule, the antenna decision sets the floor on how much data the mission can return and how quickly the team can recover from an anomaly. A cheaper antenna that forces a narrower operating envelope can end up costing more than the dish it replaced.

The lesson is not that everyone should buy a parabolic dish. It is that the link budget must be built from measured components, including the boring ones like coax and connectors, and it must account for the orbit the spacecraft actually flies. The Toulouse team learned this with 11 days of lost passes and a spreadsheet they now keep pinned above the test bench.

A Short Checklist From the Bench#

, Model the full receive chain, not the antenna alone., Use measured patterns at the frequencies you will actually transmit on., Check the orbit against the antenna's coverage area before ordering., Budget for the mount, the permit, and the cable, not just the hardware., Log every pass attempt so failures become data instead of guesses.

The dish eventually went on the roof six months later, after the mission's data volume targets rose. The panel now sits in a closet, labeled with the elevation angle it could not reach.

More articles