Inside Our Factory QC: How Every Flight Controller, ESC and Motor Is Tested Before It Ships

Every FPV buyer eventually learns the same lesson: the gap between a good board and a returned board is rarely the design — it's the testing. This article documents what we actually test, unit by unit, on the parts we ship under our own name. Use it as a checklist with any supplier, including us.

Per-unit testing vs batch sampling: the honest trade-off

Most FPV electronics is tested by batch sampling: pull 3–5 boards per production lot, test those, ship the rest on faith. It's cheaper, and for high-volume commodity parts the math can work. The failure mode is also obvious — sampling statistics tell you nothing about the specific board in your customer's quad.

We test every unit of the Factory Direct lines — flight controllers, ESCs, motors, VTX, receivers and cameras. It costs more per unit. It also means the DOA rate our retail customers and OEM partners see is a rounding error, not a support queue. For a brand shipping under its own name, we think that trade is obvious.

Flight controller: the 5-point outgoing check

Every Edge F7 goes through:

1. Power integrity. Board powered from a bench supply; inrush current, 5V rail and 9V rail verified under a resistive load for 60 seconds — not just "the LED lights".
2. Sensor calibration sweep. Gyro and accelerometer run through a calibration cycle on a leveled jig; outputs logged against tolerance. Boards that fail drift checks here are the ones that cause "random tilt on arming" tickets later.
3. OSD and flash. AT7456E OSD boot verified on a monitor; firmware version, UART mapping and VTX table confirmed against the shipping configuration.
4. Port-by-port I/O test. Every UART, the buzzer pad and LED pad exercised with a test harness. Solder bridges and cold joints are caught here, not by your customer.
5. Label and config match. The gyro variant on the board must match the variant on the box. Mis-labeled revisions are the most common error we've seen in third-party boards we've cross-checked.

ESC: load, not just boot

A 4-in-1 ESC that powers on tells you almost nothing. Our Bolt 55A check runs each of the four channels at calibrated load current while logging MOSFET temperature — a channel that runs hot at 30A will fail at 55A, and it fails in someone's frame. Firmware version (Blheli_32, DShot600), current sensor offset and telemetry output are verified per board.

Motor: every unit spins, no exceptions

Batch sampling motors is where most savings-and-regret stories come from. Ours are individually:

• Dynamically balanced and spin-tested through the full throttle range with RPM logged — this catches bent shafts, loose bells and winding faults in seconds.
• Resistance and KV spot-verified against the winding spec; a 1750KV unit measuring 1810KV goes in the bin, not the box.
• Bearing check by sound and freewheel time under a stethoscope mic — cheap but effective for catching contaminated bearings.

VTX: RF is measured, not assumed

Video transmitters are the one category where marketing numbers and reality diverge most. Every Pulse 5806 is power-swept on an RF bench: output power at each of the four settings verified against tolerance, harmonic and spurious emissions checked, thermal protection triggered on the bench before it can trigger in a customer's flight. For EU-bound OEM orders we can lock VTX tables to CE power limits at the firmware level — ask during quoting.

The paperwork you should demand

Testing claims are cheap. Paperwork is what makes them verifiable. For B2B and OEM orders we provide:

• Per-batch QC summary — test items, pass rate, sample photos of the tested configuration.
• Serial-mapped test logs for orders above 100 pcs, so a field return can be traced to its bench record.
• Motor thrust bench data with test conditions stated — voltage under load, propeller, current, temperature. A thrust number without conditions is marketing, not data.
• RMA workflow agreed before shipment: warranty period, failure classification, replacement vs credit.

Frequently asked