Testing and qualification

Why test

The first hundred CubeSats taught the community that cheap, fast development has its own failure statistics: a large share of early missions died within days, many from problems a system-level test would have caught[1]. Later surveys connected failures to the power and communication subsystems and to insufficient testing, and argued for more functional and environmental verification at system level[2]. Testing is the cheapest insurance a small team can buy.

Model philosophy

Agencies separate the roles of the hardware a project builds[3]:

• Engineering model (EM). Representative in form and function; used to debug interfaces and software.
• Qualification model (QM). Identical to flight; tested at qualification levels and durations, with margin above the expected environment, and never flown.
• Protoflight model (PFM). Flight hardware tested at qualification levels for acceptance durations; the usual choice of small missions that cannot afford a separate QM.
• Flight model (FM). Tested at acceptance levels to find workmanship defects.

Most PocketQube and CubeSat teams fly a protoflight model after an engineering model, which is a sound compromise when the levels and durations are respected.

Test as you fly

The most useful principle is to test the spacecraft in the configuration and sequence it will fly: deployables stowed, separation switches pressed, timers running, the flight software in its launch mode. Grizu-263A followed a test-as-you-fly sequence of bench tests, engineering-model tests, thermal-vacuum tests and vibration tests before delivery[4].

flowchart LR
  BENCH["Bench and board tests"] --> EM["Engineering model: integration and software"]
  EM --> FUNC["Functional and end-to-end radio tests"]
  FUNC --> VIB["Vibration: resonance survey, random"]
  VIB --> TVAC["Thermal vacuum and cycling"]
  TVAC --> BAKE["Bake-out"]
  BAKE --> FINAL["Final functional test, deployer integration"]

Vibration

Launch loads reach a deployed small satellite through the deployer as random vibration, sine sweeps and shock. NASA’s CubeSat Launch Initiative asks for systems that survive the GEVS random-vibration environment of about 10 Grms[5][6]; the actual levels and durations come from the user’s guide of each launch provider, and the CubeSat Design Specification leaves them to the provider[7]. A typical axis sequence is a low-level resonance survey, the random run, and a second resonance survey: a shift of a resonant frequency between the two surveys reveals loosened or damaged structure.

Grizu-263A on the shaker

The flight model of Grizu-263A went through protoqualification vibration on 21 September 2020 at the Vibration and Acoustics Laboratory of the Faculty of Mechanical Engineering of Istanbul Technical University, mounted inside its flight pod[4]. A TIRA TV59327/AIT shaker controlled with Vibration VIEW software drove the tests; a Brüel & Kjær PULSE system acquired the data, with a control accelerometer on the slip table and response accelerometers on the pod and a reference point, from which transmissibility functions were derived. Each axis, the two horizontal axes first and then the vertical one, received a resonance survey and a random run, and inspections after every run showed no damage. The satellite was integrated into an AlbaPod in October 2020 and its kill switch released for flight.

Thermal vacuum and cycling

Thermal-vacuum tests prove that the spacecraft works at its hot and cold extremes in vacuum and survives the transitions; thermal cycling, often at ambient pressure, screens workmanship such as solder joints. Delfi-PQ was qualified with four thermal-vacuum cycles between −20 and +50 °C[8]. For PocketQubes the cycling deserves extra weight: a 1P sees about 5500 eclipses a year, with bulk swings of about 30 K and panel swings far larger, so qualification should cover the predicted range with margin[9].

Bake-out and outgassing

Materials are screened by heating samples to 125 °C for 24 h in vacuum[10]; the integrated spacecraft is then baked to drive out water and solvents before delivery. Delfi-PQ was baked at 50 °C and 10⁻⁵ mbar for 24 h[9]. The CubeSat Design Specification requires a thermal-vacuum bake-out of every CubeSat[7].

Functional, deployment and radio tests

Environmental tests only matter if a full functional test follows each of them. Deployment tests of antennas and arrays should be repeated, including in the cold. End-to-end radio tests through the real ground-station chain, with the flight frame format and a calibrated attenuator standing in for the path loss, catch errors in line coding and framing; on Grizu-263A, early community decoders that assumed a different line coding failed until the framing was clarified[4].

Magnetic and attitude tests

For magnetic attitude designs, the polarity of every sensor and actuator axis should be checked, the magnet moment measured at temperature extremes and the residual dipole of the integrated spacecraft measured, for example in a Helmholtz cage[9]. These tests are cheap and catch the errors that make a spacecraft spin up instead of down.

What the flight record says

Delfi-PQ, one of the best-documented PocketQubes, reported its in-orbit performance and the lessons of its development in detail, including the limitations of its battery protection[11]. Grizu-263A met its primary communication objective and operated for 628 days, well beyond its expected year[4]. Both missions point to the same lesson: the interfaces with the outside world (the deployer, the radio community, the launch schedule) need as much testing as the boards.

A minimum campaign for a small team

NASA’s CubeSat 101 guide describes how these tests fit into the launch-provider process in more detail[12].