Temperature, voltage, load, power cycles or repeated switching must match a credible failure mechanism.
Sensor quality assurance guide
What Sensor Burn-In Testing Reveals Before Shipment
Burn-in can expose early-life defects that appear only after a sensor has been powered, heated, loaded, cycled or repeatedly operated. But a warm chamber alone does not prove reliability. The test must stimulate the expected failure and record the right signal from every unit.
Short answer
A meaningful sensor burn-in screen can reveal intermittent connections, resets, abnormal current, output chatter, drift, weak output stages, communication faults and lost configuration. It cannot, by itself, prove service life, sealing, EMC, full-range accuracy, functional safety or suitability in your machine.
Laboratory image: Radarvector / Wikimedia Commons, CC BY-SA 4.0. Instruments illustrate monitored electrical testing; the exact fixture and acceptance limits remain sensor-specific.
A target, pressure, light path, communication command or other input may be needed to expose the fault.
Per-unit current, output, threshold, data and temperature reveal more than a chamber setpoint.
Contain the lot, preserve the first failure and feed the mechanism back into process control.
Is burn-in the right test?
Start with the decision you need to make. Several tests use heat, time or cycling, but they answer different questions. Choosing the wrong label often produces a long report with little purchasing value.
Screen production units for latent workmanship, component or assembly defects that create measurable symptoms.
Best match: shipment screenExplore design limits with severe stresses during development. These stresses are not copied directly into shipment screening.
Best match: design learningDemonstrate defined performance against temperature, vibration, ingress, EMC or another stated requirement.
Best match: requirement evidenceEstimate degradation over time or compare output with a traceable reference. Burn-in cannot replace either activity.
Best match: life or measurement proofBurn-in is a controlled early-life screen, not simply time in a hot chamber.
Practical definition
Sensor burn-in is a production screening process in which finished units are operated or electrically biased for a defined period under controlled conditions. Relevant sensor inputs, loads and power cycles may be applied while electrical and sensing behavior is measured before, during and after the exposure.
The aim is to make marginal defects fail inside a controlled factory process instead of after installation. A weak crimp may open as materials expand. A regulator may reset at a supply corner. A photoelectric threshold may wander as the emitter and receiver warm. A configuration value may disappear after repeated power cycles.
Time and temperature are therefore only part of the screen. If the unit is unpowered, never switches, has no representative load and is checked only once at the end, brief resets and output chatter may never be recorded.
Failure mode
Define the defect or field symptom the screen is expected to find.
Stress
Apply heat, voltage, load or cycling that activates that mechanism.
Stimulus
Present the target or process input needed to make the sensor respond.
Measurement
Log the electrical, sensing or communication symptom at useful speed.
Reaction
Define unit failure, lot action, analysis and corrective action in advance.
The official MIL-STD-883 Method 1015 describes burn-in as a method for screening marginal microelectronic devices. That principle is useful, but its device temperatures and profiles should not be copied blindly to a complete industrial sensor containing plastics, potting, optics, connectors and assembled electronics.
Buyer check: Ask which failure mechanisms the supplier expects the screen to reveal. “We test every unit for 24 hours” is not enough without the stress profile, monitored parameters, acceptance limits and failure reaction.
Use the right term
Similar test names can hide different evidence.
These activities can support one another, but they are not interchangeable.
| Activity | Main purpose | Typical population | What it does not prove alone |
|---|---|---|---|
| Burn-in | Screen early-life production defects under defined operation or bias. | Often 100% of selected production. | Full life, application fit or every environmental requirement. |
| Temperature soak | Hold units at a stable temperature for conditioning or observation. | Samples or production units. | Dynamic faults if units are not powered, stimulated and monitored. |
| Temperature cycling | Exercise expansion and contraction through repeated temperature changes. | Qualification samples or a production screen. | Long-term life unless a validated model and failure mechanism apply. |
| ESS / HASS | Use tailored environmental stresses to expose latent manufacturing defects. | Production, after development and validation. | That aggressive stress is harmless without screen-development evidence. |
| HALT | Find design margins and weak links with accelerated stresses. | Development samples. | Shipment quality or a directly reusable production profile. |
| Qualification | Verify representative units against stated product requirements. | Selected samples. | That every shipped unit is defect-free. |
| Calibration | Compare output with a reference and adjust or report error. | Samples or every unit, depending on product. | Intermittent defects, sealing, EMC or lifetime. |
Temperature-change testing is separately described by IEC 60068-2-14:2023. A production screen should name the actual method and purpose instead of using “burn-in” as a catch-all term.
Observable failure symptoms
What a well-designed screen can reveal before shipment
The useful result is not “the chamber completed.” It is a measured symptom connected to a credible defect mechanism.
Intermittent wiring, crimps or solder joints
Heat and switching can expose opens, voltage drops, contact resistance and output chatter that disappear at room temperature.
Watch: terminal voltage and event countResets and abnormal current
A weak regulator, protection component or damaged assembly may show delayed startup, current spikes, thermal shutdown or repeated resets.
Watch: current, voltage and boot eventsLeakage, heating or load weakness
Representative output load and switching can expose excessive residual voltage, leakage, slow transitions or a stuck output.
Watch: load current and output voltageThreshold, hysteresis and repeatability changes
The sensor can begin switching too early, too late or inconsistently as its electronics and sensing head reach operating temperature.
Watch: switch points in both directionsZero, span, noise or settling drift
Pre-test and post-test averages can hide a temporary warm-up shift. A time series shows whether the output stabilizes or continues moving.
Watch: continuous output against referenceCommunication and configuration faults
Power cycling and temperature can reveal dropped messages, checksum errors, boot loops or settings that are not retained correctly.
Watch: error log, state and serial identityIncomplete cure or contamination effects
Potting, adhesive or optical contamination can create drift, leakage, outgassing or reduced optical margin after heating.
Watch: drift trend and optical marginUnexpected self-heating
A unit may run hotter than its neighbors because of excess current, poor thermal contact, output loading or assembly variation.
Watch: unit temperature, not chamber onlyFixture or chamber problems
Position-based failures may come from airflow, a bad socket, harness resistance, a drifting target or a weak data channel.
Watch: reference channels and position map
Do not overclaim the result
Burn-in cannot prove the complete application.
A passed screen means the unit met the defined limits during that defined exposure. It does not automatically demonstrate every condition the sensor will see in service.
- It does not guarantee years of operating life.
- It does not prove full-range accuracy or calibration uncertainty.
- It does not demonstrate IP sealing, humidity, chemical resistance, vibration or shock unless tested.
- It does not replace EMC, functional-safety or application validation.
- It cannot correct the wrong target, sensing distance, wiring, mounting or PLC interface.
Production screen workflow
A meaningful burn-in process has nine connected steps.
The sequence begins before the chamber door closes and continues after a failure is found.
Define the objective
Name the field symptom, suspected mechanism, product population and evidence needed.
Record the baseline
Capture model, revision, serial, configuration, visual condition and electrical or sensing values.
Build the fixture
Control supply, load, target, communication, grounding, cable route and mounting position.
Map the environment
Check chamber gradients, airflow, fixture loading and actual product temperature.
Apply controlled stress
Use defined ramps, dwell, voltage, load, target cycles and recovery without overstressing good units.
Monitor every unit
Log signals fast enough to capture the shortest fault that matters to the machine.
Recover and retest
Compare pre-, in- and post-test behavior before recalibration or repair changes the evidence.
Analyze failures
Preserve the first failure, confirm it, localize the mechanism and determine lot impact.
Control the screen
Trend fallout, fixture faults and drift; revalidate after important product or process changes.
The test system matters
A chamber is only one part of the burn-in rack.
The screen needs a controlled electrical path, a repeatable sensor stimulus and a data path that identifies each unit. The measurement system must also distinguish a failed sensor from a failed socket, harness or channel.
- Measure supply at the device terminals, not only at the power source.
- Use a representative PLC input or DC load when output-stage behavior matters.
- Keep target position, optical path or process reference stable and traceable.
- Add reference channels, fixture self-checks and a channel-swap rule.
- Synchronize temperature, electrical data and event time.
Controlled input and stressPer-unit evidence
Programmable supplyVoltage, current limit, power cycles
Environmental chamberRamp, dwell, product temperature
Sensor fixtureSerial position, cable and mounting
Target or stimulusMetal, light, material or process input
PLC input / loadRepresentative electrical interface
DAQ and event logCurrent, output, data and alarms
Device temperature
Output switching
Supply current
Fault events
Measure the symptom
What should be monitored during burn-in?
Choose parameters from the expected failure mode. A one-minute sample interval will miss a 100 ms reset, so the logging rate, event trigger and missing-data rule must fit the shortest fault that matters.
Supply pathTerminal voltage, current, inrush and resets
Discrete outputTransitions, chatter, leakage and residual voltage
Analog outputZero, span, noise, settling and drift
Sensing behaviorThreshold, hysteresis, repeatability and response
Digital dataProcess values, diagnostics, messages and checksum
TemperatureChamber, fixture position and actual unit temperature
Preserve the as-found condition. Recalibration or repeated “test until pass” can erase the evidence needed to understand a drifting or intermittent unit.
Pre, during and post
A drift value needs a reference, a denominator and a time history.
Compare the same defined measurement before and after exposure. If a normalized value is reported, state whether the denominator is span, full scale, nominal output or another reference.
ΔX = Xafter − Xbefore
Use the same target, supply, load, temperature and measurement method for both values.
Normalized shift = ΔX ÷ stated reference
Never publish “percent drift” without naming the denominator.
A time series separates normal warm-up from continued movement. An end-only measurement can miss both a temporary excursion and an intermittent failure that recovered before inspection.
Conceptual output versus timeNot a product specification
BaselineStress exposureRecovery / retest
Stress from failure physics
There is no universal burn-in temperature or duration.
The profile must activate a relevant defect without consuming useful life or damaging good sensors.
Build the profile from evidence
- Known field failure mechanisms and manufacturing defect history
- Rated product limits and internal component or material limits
- Actual product temperature, supply voltage and output loading
- Target, switching, communication and power-cycle conditions
- Screen-development trials, false-fail risk and correlation with failures
Avoid convenient but unsupported rules
- Copying a microcircuit profile to a potted complete sensor
- Converting arbitrary hot hours directly into years of service life
- Using chamber air temperature as the only thermal measurement
- Extending duration when the fixture cannot detect the field symptom
- Treating zero fallout as proof that the screen has good detectability
Arrhenius caution: The relationship
AF = exp[(Ea/k) × (1/Tuse − 1/Ttest)] can model temperature acceleration for a specified mechanism when the assumptions are valid. It is mechanism-specific, uses absolute temperature and does not turn any hot-chamber exposure into a universal life claim. See the Texas Instruments reliability note.
One screen does not fit every sensor
Monitoring must follow the sensing principle.
Electrical continuity alone cannot confirm that the sensing function remained stable. Each family needs a representative stimulus and family-specific measurements.
Inductive
Track oscillator behavior, sensing distance, switch points, hysteresis, output drop, leakage and current against a defined metal target.
Capacitive
Track threshold drift, false switching, target sensitivity, adjustment or configuration and output behavior with the intended material.
Photoelectric
Track emitter and receiver behavior, gain margin, threshold, alignment, contamination sensitivity, output and communication.
Analog process
Track zero, span, linearity, noise, settling and reference stability across the required process points.
Safety-related
Use the product’s safety architecture, diagnostics and applicable safety validation process. Ordinary burn-in is not a safety certification.
Population strategy
Should burn-in cover 100% of sensors or a sample?
The answer depends on defect risk, screen detectability, field cost and whether the test is nondestructive.
The screen can economically catch an important early-life risk.
- Historical defects show a meaningful early-life pattern.
- Field replacement, downtime or customer qualification cost is high.
- The validated screen detects the failure without damaging good units.
- Contract, product strategy or risk control requires every unit to pass.
The purpose is qualification or process monitoring, not individual screening.
- The process is mature and early-life risk is low and stable.
- The test is destructive, consumes significant life or has limited capacity.
- Samples are selected by a defined, representative plan.
- Other controls provide faster or more direct defect detection.
Observed burn-in fallout = confirmed failed units ÷ valid units entering the defined screen population
Report the product, lot, exposure, failure definition, false failures, retests and fixture-related invalid results with the percentage.
Screen, analyze, correct
Burn-in is a screen, not a cure.
A rising failure rate is evidence of a process problem. Removing failed units without learning from them leaves the cause in production.
Preserve the first failure
Save raw data, channel, position, serial, configuration and as-found symptoms before repair or recalibration.
Confirm the symptom
Repeat only under controlled conditions and separate sensor behavior from fixture, chamber and software faults.
Localize the mechanism
Use nondestructive analysis first, then destructive methods when needed and authorized.
Contain the risk
Identify affected lots, revisions, suppliers, component dates, fixtures or process windows.
Correct the process
Control the actual root cause in design, material, assembly, firmware, handling or test.
Verify effectiveness
Confirm that the failure no longer occurs and that the screen still detects the intended mechanism.
No test-until-pass: Repeated retesting without a defined rule can hide intermittent failures. Rework, rescreening and release require approved criteria and complete traceability.
Illustrative manufacturing case
Review field troubleshooting
Why an intermittent three-wire sensor passed the original chamber test
This example shows how test design changes the evidence. It is a diagnostic scenario, not a published product-failure claim.
Brief resets during machine operation
The PLC saw occasional missing transitions. The sensor worked normally when checked later at room temperature.
Warm chamber, fixed target, end-only check
All units shared a current measurement. Brief resets were invisible and the fixed target did not create switching events.
Representative load and per-unit event capture
Terminal voltage, current, output, device temperature, target cycles and power cycles were recorded by serial number.
Failures correlated with a regulator lot
Containment and supplier corrective action removed the mechanism. Extending the old chamber time would not have created the same learning.
Supplier audit
Ask for evidence that lets you reconstruct the screen.
A useful report connects the exact shipped product to the actual profile, actual monitored data and actual disposition.
Evidence a buyer can use
- Exact model, revision, serial or lot, quantity and test dates
- Purpose, method revision and intended failure mechanisms
- Actual temperature, ramp, dwell, recovery, supply, load and cycle profile
- Fixture, chamber, software, calibration, mapping and channel checks
- Per-unit parameters, limits, sample rate, alarms and raw-data retention
- Pre-, in- and post-test results, deltas, failures and invalid exposures
- Disposition, lot reaction, deviations, rework and release approval
Claims that need follow-up
- “100% burn-in tested” with no test profile or monitored signals
- Only the chamber setpoint, with no actual unit temperature
- Only a final pass label, with no pre-test or during-test data
- No separation of DUT failures from socket, harness or channel faults
- No defined retest, rework, lot reaction or failure-analysis rule
- A temperature/time profile copied from a different product construction
- A claim that burn-in alone proves years of life or application suitability
The active MIL-HDBK-344 provides a framework for planning, monitoring and controlling environmental stress screening. Measurement records should also reflect calibrated equipment, validated methods and uncertainty awareness; ISO/IEC 17025 describes competence and consistent operation for testing and calibration laboratories.
RFQ and audit checklist
Ten questions to ask before accepting “burn-in tested.”
| # | Question | Strong evidence |
|---|---|---|
| 1 | Which failure mechanisms is the screen designed to reveal? | Named defects, symptoms and validation evidence. |
| 2 | Is every unit screened or is a sample selected? | Defined population and sampling rationale. |
| 3 | Are sensors powered, loaded and actively sensing? | Fixture diagram and operating modes. |
| 4 | What is logged for each serial number? | Parameter list, rate, limits and retention period. |
| 5 | How are chamber gradients and self-heating controlled? | Mapping, airflow study and unit-temperature channels. |
| 6 | How was the stress and duration selected? | Failure-physics basis and no-damage validation. |
| 7 | What are the unit, lot, retest and rework rules? | Approved acceptance and reaction plan. |
| 8 | What do recent failures show? | Pareto, correlation, failure analysis and corrective action. |
| 9 | Which changes trigger screen revalidation? | Design, component, material, process, firmware and fixture triggers. |
| 10 | Which requirements are covered by other tests? | Qualification, calibration, EMC, ingress and application evidence. |
Need a sensor requirement that suppliers can actually verify?
Share the sensor type, target, supply, output load, operating temperature, switching cycle and field concern. xsz sensor can help organize the electrical and sensing checks needed for model selection, sample evaluation or a supplier discussion.
Continue the evaluation
Related sensor quality and application guides
Frequently asked questions
Sensor burn-in testing FAQ
What defects can sensor burn-in testing detect?
It can detect early-life defects that create measurable symptoms under the chosen stress and operation. Examples include intermittent connections, resets, abnormal current, output chatter, warm-up drift, threshold changes, weak output stages, communication errors and lost configuration. The screen cannot detect a defect that its stress and measurement system never activate or observe.
How long should a sensor burn-in test run?
There is no universal duration. The time should come from known failure mechanisms, product limits, defect history, screen-development data and the time needed to observe the relevant symptom. A longer test is not automatically better and can waste capacity or damage good units if the stress is poorly chosen.
Does burn-in prove that a sensor will last for years?
No. Burn-in is mainly an early-life production screen. A service-life claim needs an appropriate reliability model, accelerated-life evidence, failure analysis and application assumptions. Passing burn-in does not by itself prove wear-out life.
Should every industrial sensor receive 100% burn-in?
Not always. One hundred percent screening is most useful when a validated, nondestructive screen catches an important early-life risk at acceptable cost. Sampling may be more suitable for qualification, process monitoring, mature low-risk production or tests that consume useful life.
What is the difference between burn-in and temperature cycling?
Burn-in usually operates or biases units for a defined period to screen early-life defects. Temperature cycling repeatedly changes temperature to exercise expansion and contraction. A production plan may combine them, but the mechanisms and evidence should be named separately.
Is burn-in the same as HALT or HASS?
No. HALT is a development activity used to discover design margins and weak links. HASS is a production screen derived from product knowledge and validated limits. Burn-in is a broader early-life screening term and may use less severe, more stable operating conditions.
Should a sensor be calibrated before or after burn-in?
Record a controlled baseline before the test and verify the sensor afterward. For products requiring calibration, the manufacturing sequence depends on the design and quality plan. Preserve as-found post-test values before recalibration so the adjustment does not hide drift.
Can burn-in damage a good sensor?
Yes, if the temperature, voltage, load, cycling or duration exceeds a valid screen window. Good screen development shows that the profile activates relevant defects without creating new damage or consuming unacceptable product life.
What should a sensor burn-in certificate or report include?
It should identify the exact product population, method revision, actual stress profile, fixture and chamber, monitored parameters and rates, limits, pre/during/post results, failures, invalid exposures, rework or retest rules, lot disposition and traceable approval.
What should happen when a sensor fails burn-in?
The supplier should preserve the first failure, confirm the symptom, separate fixture faults, analyze the mechanism, contain affected production, correct the root cause and verify effectiveness. Simply discarding the unit does not control the process.
Technical references
Sources used for this guide
- U.S. Defense Logistics Agency, MIL-STD-883 Method 1015, Burn-In Test
- U.S. Defense Logistics Agency, MIL-HDBK-344, Environmental Stress Screening of Electronic Equipment
- NASA Technical Memorandum, COTS electronic-part reliability and burn-in considerations
- IEC 60068-2-14:2023, Environmental testing - Tests - Change of temperature
- ISO/IEC 17025:2017, Competence of testing and calibration laboratories
- Texas Instruments, Reliability testing
- Texas Instruments, Calculating Useful Lifetimes of Embedded Processors
- Automotive Electronics Council, qualification and early-life failure-rate document registry
- Analog Devices, Reliability Handbook
Image credits: Hero laboratory image by Radarvector, cropped by Pittigrilli, licensed CC BY-SA 4.0 through Wikimedia Commons. Remaining application and sensor illustrations are xsz sensor website assets. External standards and handbooks are referenced for technical context; product-level acceptance limits must come from the exact approved specification.