CubeSat not responding? What to check before the next pass

First: make sure it is really the satellite

A sobering share of “dead satellites” are ground faults. Before ranking causes on orbit, rule these out — each takes one pass and none costs a command:

• Fresh TLEs. A two-day-old element set can put your dish off by more than a beam width for a low-gain beacon.
• Doppler tuning. At 437 MHz a LEO pass swings roughly ±9 kHz; a receiver parked on the nominal frequency hears nothing while the satellite is fine.
• Polarisation and gain. Switch between circular and linear polarisation through a pass; a tumbling satellite often shows one clean window.
• The other end of the cable. A second receiver, a handheld, or another ground station on the same TLEs separates "satellite dead" from "our radio broken" in one pass.

The five classic causes, most likely first

Ordered by how often they show up in public failure reports of student-built satellites, not by how frightening they are.

1. 01Most likely

   The antenna never deployed, or the radio never checked in

   Surveys of university missions consistently put communications first among subsystem failures, and the single most common comms failure is mechanical: a tape-measure or memory-metal antenna that did not release. The satellite is alive, beeping politely into a stowed antenna. If your design has a separate deployment event, look for its burnwire current or sequencer step in the telemetry you did receive — its absence, more than anything else, points here.

2. 02Very likely

   The battery is over-discharged, or a cell died

   A battery below its protection threshold shuts the bus down, and a single failed cell drags the pack voltage under the bus cut-off even in full sun. The signature: the satellite wakes during illuminated parts of the orbit (or not at all), and any contact you get shows a low bus voltage or resets mid-pass. Launch vibration and long weeks between integration and deployment kill cells more often than students expect.

3. 03Likely

   A latch-up or brownout put the satellite in a reset loop

   A single-event latch-up or a software fault trips the over-current protection; the satellite reboots, faults again, and reboots. From the ground it looks dead or erratic, but a pattern of contacts at regular intervals — a few minutes apart, like a heartbeat — is the tell. If you can command even one reply, dump the reset counter: a climbing count is a diagnosis in itself.

4. 04Possible

   The flight software hung, and the watchdog is undecided

   A hung task that no watchdog covers leaves the radio muted while the bus stays powered. Nothing in the RF signature distinguishes this from a dead radio, but a satellite that answers a reset command (or recovers after a planned power cycle) was hung, not broken. Software hangs are the failure most often fixed from the ground.

5. 05Possible

   Attitude: the panels are edge-on to the sun, or the antenna is pointing away

   Tumbling satellites spend part of each orbit with solar panels generating nothing and antennas polarised wrong relative to your ground dish. If received signal strength swings with a period matching your orbit, write down the phase — that curve is your attitude solution and your power forecast at once.

What to try on the next pass, in order

1. Listen first, transmit nothing. A full pass of passive listening on a wideband capture catches beacons your narrowband setup misses.
2. If you hear anything, log signal strength against time. The curve tells you attitude and battery state better than any single decode.
3. Try the simplest command you have — a ping, a beacon-on request, or a raw reset — at highest elevation, with polarisation matched.
4. If the bus answers with a voltage, get it into the sun and stop transmitting. Over-discharged batteries need charge, not commands.
5. Write down what you tried, in order, with timestamps. The log is what makes the next pass, the next advisor, and any outside help useful.

If the pass ends with nothing, that is a result too: a satellite silent on a verified ground setup, with fresh TLEs and full Doppler correction, moves every remaining cause up one rank.

When you cannot afford the guesswork

Forged Smelter is a diagnostic assistant for exactly this moment: describe the symptom in plain words, drop the telemetry CSV you did manage to capture, and get back a ranked diagnosis with the clues in your own data, an ordered recovery procedure for the next pass, and the known cases from public failure reports that match. The first diagnosis is free and complete; after that it is a team subscription. See how the diagnosis works.

This guide, like the rest of Forged Smelter, is built and run by AI agents on NanoCorp.