Design & architecture
Airborne data acquisition systems, explained
An airborne data acquisition system turns what happens on an aircraft — strains, temperatures, pressures, bus traffic, video — into time-stamped data that engineers can trust on the ground. Every flight-test campaign, equipment trial and certification flight starts with one. Here is what it is made of, how the parts fit together, and how to compare one system with another.
What it is, and what it is not
An airborne data acquisition system — DAS for short, its units DAUs — measures physical quantities and captures digital traffic on board, stamps every sample with time, and delivers the result to a recorder, a telemetry link or a network. It is an engineering tool: installed for a test campaign, configured for the parameters that campaign needs, and changed for the next one.
It is not a flight data recorder. The FDR records a parameter list set by regulation, is built to survive an accident, and cannot change without a certification process. The two often sit side by side, and the flight-test system may even listen to the FDR feed, but they answer different questions.
Inside a unit
scroll
- Analog inputs and conditioning: bridges for strain gauges, thermocouples with cold-junction compensation, IEPE accelerometers, pressure sensors, voltages and frequencies — each needs excitation, gain and anti-aliasing filtering before the converter.
- Digital interfaces: avionics buses such as MIL-STD-1553, ARINC 429, ARINC 717 and CAN, serial lines, Ethernet and discretes, plus video and audio from cameras and intercoms.
- Controller: schedules the sampling, time-stamps every sample and message, and formats the result — PCM frames, IRIG 106 Chapter 10 packets, IENA, or a manufacturer's own format.
- Time: an IRIG-B, GPS or PTP input that disciplines the unit's clock, so that its data line up with every other unit, the video and the ground.
- Outputs: an onboard recorder, a telemetry transmitter, and an Ethernet stream to other units, displays or a ground station.
Sampling: the decision that cannot be undone
Each channel is sampled at a rate chosen for what it must show: a few samples per second for a temperature, hundreds for a control surface position, thousands for a vibration. Too slow, and a fast event is missed or, worse, folded into a false low-frequency signal; too fast, and the recorder and the telemetry link fill with data nobody needs. The anti-aliasing filter in front of each converter must match the rate, and both are set in the unit: no processing on the ground recovers what was not sampled correctly.
One unit or several
A small installation fits in one unit: every sensor wired back to it, one clock, one recording. Larger aircraft and longer campaigns usually distribute several units close to the sensors, linked by Ethernet, with time carried across the network. Many installations mix the two: a central unit for the cockpit and the buses, remote units in the wings or the tail.
Recording, telemetry and streaming
The three outputs answer three needs. The onboard recording is the reference: complete, at full rate, kept after the flight. Telemetry sends what the ground needs to follow the flight live and clear the next test point, within the capacity of the radio link. Streaming on the on-board network feeds displays, other recorders and, when a high-rate link exists, the ground.
The format matters as much as the rate. A recording in an open format such as IRIG 106 Chapter 10 can be read by any compliant tool for decades; a closed format ties the archive to one supplier's software. Ask what the system writes, and test the files with a tool you did not buy from the same supplier.
Comparing systems
scroll
| Criterion | What to ask | Why it matters |
|---|---|---|
| Channel types | Does every type on your list have a native interface? | An adapter is one more box, one more time stamp and one more failure point |
| Rates and resolution | Maximum sample rate per channel and in total, at the resolution you need | Datasheet maxima are rarely reachable on every channel at once |
| Time | Which references are accepted, with what accuracy, and what happens when the reference is lost? | A unit a millisecond off invalidates every comparison with another |
| Format | Which format is recorded and streamed, and is it open? | Your archive must outlive the supplier's software |
| Expandability | Can channels be added, removed or swapped between campaigns, and how fast? | The next campaign never has the same list |
| Configuration | How is a setup prepared, checked and loaded, and can it be exported? | A configuration error grounds a flight as surely as a hardware fault |
| Environment | Qualification standard, temperature range, vibration category, EMC | The installation zone decides, not the cabin |
| Size, mass and power | Per channel, cabling and cooling included | On a small aircraft, this budget is often the limit |
| Data security | Encryption, erasure and access to the removable media, where the data require it | Some campaigns cannot fly without them |
The mistakes that cost the most
- Choosing the unit before the channel list is frozen, then fitting the list to the unit.
- Sizing for the average flight, and saturating the recorder, the link or a network segment on the worst one.
- Treating time as a detail: two units without a common reference cannot be compared, and the error is invisible in each file taken alone.
- Leaving the ground tools to the end: the first flight produces data that nobody can open, or open correctly.
- Never testing the whole chain, from sensor to report, before the first flight.