11. Red Flags and Beginner Mistakes

Most connector failures aren't exotic. They're ordinary decisions that looked harmless in a meeting and turned into a bad crimp, a missing cap, an impossible service angle, or six weeks of lead-time pain. This is the fast scan for those decisions.

Red flag	Why it's bad
Picking only by pin count	Ignores current, voltage, contact size, environment, tooling
Ignoring backshells	No strain relief, poor EMI, poor sealing
Ignoring tooling	Design may be impossible to build correctly
Mixing signal and power casually	Noise, safety, creepage, heating, service confusion
Bad shielding (pigtail) termination	Pigtails are inductive — raise impedance, radiate
No strain relief	Conductors fatigue and fail at the termination
No keying	Similar connectors get swapped; expensive mis-mates
Poor service access	Technicians damage connectors during maintenance
Hobby connectors outside their suitable environment	No sealing, weak latch, unknown vibration life
Not checking mating cycles	Test/service ports wear out early
Not checking lead time	Schedule failure despite a good design
Not documenting pinouts	Harnesses become tribal knowledge
Wrong powered gender	Exposed live pins → shock/short risk
No dust caps	Dirt/water/pin damage on unmated ports
Using spare pins randomly	Future maintainers inherit chaos
No cable labels	Debugging becomes painful
No mating connector in BOM	Procurement buys half an interface
No torque spec	Sealing and anti-vibration features compromised
Confusing IP67 with IP68	IP67 ≈ 1 m / 30 min; IP68 = stated depth/duration — not interchangeable
Undersized wire in wire seal	Cavity leaks; whole connector seal defeated
Two wires doubled into a one-wire crimp barrel	Unqualified fill never forms the gas-tight joint; one wire loose, wire seal defeated — use a splice or a dual-wire-rated terminal (crimping)
No ground-first mating sequence where needed	Power-before-ground → latch-up, ground bounce, resets
Substituting "equivalent" parts without qualification	Plating/insert/thread/geometry can differ; not drop-in
Gold contacts mated to tin contacts	The mixed interface fretts and builds tin oxide on the gold half — one plating class per mated pair (low-level deep dive)
mV/µA signals on tin or power-class contacts	Dry circuits can't break down surface films; intermittent, drifting resistance (low-level deep dive)
No PCN/EOL monitoring on a released design	Notices go to customers of record; you learn about discontinuance from a failed order (lifecycle)
Broker purchase without traceability verification	Counterfeit exposure concentrates in scarce/obsolete parts (lifecycle)
Mixing manufacturers in one mated pair unverified	Outside QPL-class families, no drawing arbitrates the mate (lifecycle)

Many of these reduce to an item that fell off the BOM or the drawing — see What People Forget.

---