Defense and unmanned systems electronics
Airframe electronics fail in three ways — vibration, the launch event, and a transmitter desensitising its own navigation receiver. All three are decided at layout, not in test.
Evidence
What we can substantiate
Everything below is work we have actually done, described at a level that respects the client’s confidentiality. If you need detail, ask under NDA.
- GNSS that still works with the video link running The most common airframe integration failure is a 1.2 or 1.3 GHz video transmitter, or a switching regulator harmonic, sitting close enough to L1 to desensitise the receiver. It is not visible on the bench because the transmitter is off. We plan antenna separation, filtering and the LNA position against the whole radiated environment, then verify with the link actually keyed.
- Surviving the launch event, not just the flight Tube-launched airframes see tens of g of setback in milliseconds. Tall electrolytics, unsupported connectors and heavy inductors are the parts that move. We orient and restrain by axis, prefer low-profile ceramics where the ripple budget allows, and treat the board as a structure rather than as a schematic realised in copper.
- Vibration is a connector problem before it is a board problem Airframe vibration works crimps loose, fatigues pigtails and fretting-corrodes contacts long before it cracks a solder joint. Positive-latch connectors, strain relief designed rather than assumed, service loops with a defined bend radius, and staking on anything with mass.
- Rigid-flex where the alternative is a harness that fails A flex interconnect removes the connector pairs that vibration attacks. Designed to IPC-2223 with the bend radius set by outer-fiber strain, not by what fits — and with the dynamic-versus-static distinction respected, because a joint that flexes in service needs a different stack-up.
- Power that tolerates a servo stall and a hot battery swap Servo and actuator current steps collapse a badly-decoupled rail and reset the autopilot mid-flight. Hot-plugging a battery rings the input LC hard enough to exceed capacitor ratings. Both are arithmetic before they are failures — we size against the worst realistic transient and verify on the bench.
- Screened supply chain, documented Component selection filtered on country of origin, corporate ownership and your program's exclusion list, with the screening basis recorded alongside the BOM and a second source identified for anything that becomes single-point.
- Redundancy that actually degrades gracefully Dual-source IMU footprints, independent power domains for flight-critical and payload, and a failure mode analyzed at the pin rather than assumed at the block diagram.
Common questions
Straight answers
What does "ITAR-free" actually mean here?
The designs contain no US-origin controlled content, so they carry no ITAR or EAR obligations that would restrict where you can sell or who you can employ on the program. For European primes building sovereign capability that removes an entire category of licensing friction. It is a constraint applied from part selection onward, not a label added at the end — and it is worth stating early, because retrofitting it means re-selecting silicon.
Why does my GNSS lose fix when the video transmitter keys up?
Almost always desensitisation rather than interference in the strict sense. A 1.2 GHz analogue video transmitter puts enormous power close to the receiver, and even well outside the GNSS band its noise floor and harmonics raise the effective noise at L1 — so the receiver sees fewer satellites and a worse fix, or none. The fixes are physical separation, a SAW filter ahead of the LNA, attention to where the LNA actually sits in the chain, and a hard look at whether a switching regulator harmonic lands in band. None of it shows on a bench test with the transmitter off, which is why it is found late.
Can you work to a client-imposed component exclusion list?
Yes, and it is a standing capability rather than a special request — we already design to exclusion lists. The output is a BOM with the screening basis documented per line, so your compliance function can audit the decision rather than take it on trust, plus a second source for anything that becomes a single point of failure under the constraint.
Why is the client not named?
Because they have not given permission, and in this sector discretion is a qualification rather than an inconvenience. Work is described generically — "fixed-wing UAV avionics carrier board for a European unmanned systems OEM" — and system parameters stay off the public internet entirely. You can expect the same treatment.
Are you set up for classified or export-controlled work?
We do unclassified design work, and we are not currently registered for the transfer of controlled technical data. Where a program falls under EU Regulation 2021/821, licensing has to be in place before controlled specifications change hands — tell us early and we will get our position confirmed rather than discover it mid-project. If you need cleared facilities or personnel, we will say so on the first call rather than three weeks in.
Projects
Work in this sector
Flight-controller carrier for a tube-launched UAV
Teardown, clean-room re-implementation, and a rigid-flex carrier that fits inside a launch tube.
Edge-compute architecture for an unmanned surface vessel
Three architectures compared on real bandwidth budgets, so the platform decision was made on evidence rather than TOPS figures.
Got a board to design — or one that won’t boot?
You talk to the engineer who would do the work. Reply within one business day, and we’ll sign your NDA before you go into detail.