Safety, Reliability, and Derating
Reliable hardware keeps working under expected stress and fails safely under credible faults. Safety and reliability are designed through margins, derating, protection, diagnostics, documentation, and verification, not added at the end.
Learning Objectives
By the end of this lesson, you should be able to apply derating, estimate temperature rise, identify critical failure modes, plan protection, and review hardware for safe startup and fault response.
Derating
Derating means using parts below their absolute limits. It reduces stress from temperature, tolerance, transients, aging, and manufacturing variation.
Common derating rules of thumb:
- capacitors: voltage rating at least
1.5xto2xworking voltage when practical; - resistors: operate below rated power, often below
50%; - MOSFETs: voltage rating above supply plus transients;
- connectors: current rating above maximum continuous current;
- regulators: thermal margin at maximum ambient.
Derating targets vary by industry and risk level; document the rule used.
Thermal Reliability
Temperature rise:
$$
\Delta T = P\theta_{JA}
$$
where P is dissipated power and thetaJA is thermal resistance from junction to ambient.
For an LDO dropping 12 V to 5 V at 100 mA:
$$
P = (12 - 5) \times 0.1 = 0.7 W
$$
If thetaJA = 80 degC/W:
$$
\Delta T = 0.7 \times 80 = 56 degC
$$
At 50 degC ambient, the junction could reach 106 degC, leaving little margin.
Fault Modes
Think through:
- input overvoltage;
- reverse polarity;
- shorted output;
- open sensor;
- stuck actuator;
- stalled motor;
- overheated regulator;
- clock failure;
- memory corruption;
- firmware lockup.
This is the hardware side of FMEA thinking.
Protection Layers
Use layered protection:
- fuse or current limit at power entry;
- reverse-polarity protection;
- TVS for surge or ESD;
- thermal shutdown or temperature sensor;
- watchdog and reset supervisor;
- brownout detector;
- safe-state pull resistors;
- driver fault outputs.
One protection mechanism should not create another hazard. For example, a fuse must interrupt safely at available fault current.
Reliability of Connections
Connectors and cables fail often. Check:
- latch or retention requirement;
- mating cycle rating;
- vibration environment;
- current per pin;
- creepage and clearance;
- keyed orientation;
- serviceability;
- strain relief.
For field wiring, include labels, pin numbering, and protection against miswiring.
Safe Startup and Reset
Before firmware configures pins, hardware defaults matter:
- actuators should remain off;
- boot pins should be deterministic;
- regulators should sequence safely;
- external memories should not be corrupted;
- outputs should not glitch hazardous loads;
- watchdog and reset supervisor should recover brownout states.
Use pull resistors, gate pulldowns, and enable sequencing to make safe behavior independent of software.
Practical Checks
- Review voltage, current, and power derating for every stressed part.
- Estimate temperature rise at maximum ambient.
- Check single-point failures for unsafe output behavior.
- Verify brownout and watchdog behavior.
- Confirm fuses and protection parts are rated for fault energy.
- Test connector misplug and open-circuit sensor cases where safe.
- Document production tests for safety-critical paths.
Common Mistakes
- Confusing absolute maximum ratings with operating conditions.
- Calculating thermal rise at room temperature only.
- Letting firmware be the only thing keeping a MOSFET off.
- Using a fuse without checking interrupt rating.
- No diagnostic path for sensor open or short.
- No derating record for high-stress parts.
Summary
Safety and reliability require margins, safe defaults, protection, diagnostics, and verification. Derating reduces stress, thermal calculations reveal hidden failures, and fault-mode reviews make the design robust before hardware reaches users.
Further Reading
- IEC 61508 and ISO 26262 overview material for functional safety concepts.
- IPC-9592 power conversion derating guidance.
- NASA and military derating handbooks for conservative examples.
- Manufacturer reliability reports and FIT-rate application notes.