12 Environmental and Physical Testing
Thermal, Humidity, Vibration, EMC, Power, and Soak on the Real Assembly
IoT environmental testing, thermal cycling, humidity ingress, vibration shock, EMC, ESD, power cycling, soak testing, HALT HASS, retest trigger
12.1 Start With the Device in the Weather
Picture the first unit that leaves the bench: a sealed gateway mounted outside a pump room, a sensor strapped to a vibrating motor, or a node installed in a greenhouse where warm air condenses overnight. The question is not “does the code run?” It is “does this exact assembly still do the job after the physical world has had a chance to attack it?”
A lab bench is a benign place — stable power, room temperature, dry air, no vibration. The places IoT devices actually live are not: a freezing rooftop, a humid greenhouse, a vibrating machine, a dusty field, a car door slammed a thousand times. Environmental and physical testing exposes the real device — its enclosure, board, connectors, sensors, and mounting, not a bare development board — to controlled versions of those conditions and checks that it still behaves as intended afterward, and often during.
This is validation work no amount of software testing can replace, because the failure modes are physical: solder joints crack under vibration, seals let moisture in, plastics warp and connectors loosen with thermal cycling, and electrostatic discharge resets a microcontroller. The result of an environmental test is therefore bound tightly to what was tested — the exact assembly, configuration, and exposure profile. Change the gasket, the enclosure, or the mounting, and the old result may no longer hold.
For example, an outdoor gateway release might run one sealed assembly through -20 to 60 C cycling, a humid soak, vibration while powered, and repeated power recovery while logging joins, sensor readings, resets, and enclosure inspection notes. If the device survives but loses calibration during the cold-to-warm transition, the result is not a pass; it becomes a design or retest action tied to that exposure profile.
Start simple: name the environment, expose the real assembly, watch the behavior that matters, inspect the parts the stress could damage, and record the change that would reopen the result. That habit keeps the survival-is-not-function rule visible from the first page.
Test Tessa
“A demo proves it can work once; a test proves it keeps working — bring the field, not the bench.”
Through this chapter, Tessa keeps one board per stress: what the bench proved, what the field still owes, and the exact assembly the record is bound to.
If you only need the intuition, this layer is enough: the physical world breaks hardware in ways a unit test never sees, so expose the real assembly to controlled temperature, moisture, vibration, electrical, and endurance stress and confirm it still works. Keep the conclusion bound to the exact assembly and exposure you tested.
12.1.1 The Core Idea in One Minute
Real conditions break real hardware
Temperature, moisture, vibration, and electrical stress cause faults software tests cannot reproduce.
Test the assembly, not the board
Use the real enclosure, seals, connectors, sensors, and mounting, in the state the field will see.
Bound to the build
A result holds for the assembly and exposure tested; a material or design change can reopen it.
12.1.2 Everyday Signals of the Risk
- A sensor node works on the bench but resets in winter; a thermal-cycling test reproduces the fault before shipping.
- A sealed enclosure passes a quick look, but a humidity-and-ingress test finds condensation entering at a cable gland.
- “It survived with no visible damage” is not “it still measured correctly throughout” — only monitored behavior shows the latter.
12.1.3 Overview Knowledge Check
If you can say why physical conditions need their own tests, you have the core idea. Continue to Practitioner for the stress categories and how to run them.
12.2 Choose the Stress and the Evidence
Environmental testing is a family of controlled stresses, each targeting a different physical failure mode. The exact profiles depend on where the device will live; the categories below are the common ones, and the decisive detail is the device’s state during exposure — powered, asleep, transmitting, sensing, charging, or cycling — because a fault may only appear in one of them.
Tessa’s Field Check
- Bench says: -20 to 60 C cycling, a humid soak, vibration while powered, and repeated power recovery, logged against joins and resets.
- Field adds: the device’s state during exposure — powered, asleep, transmitting, or charging — because a fault may only appear in one of them.
- Stamp when: the record monitors behavior and inspects the physical assembly — a pass label with neither is weak.
12.2.1 Measure Behavior, Then Inspect the Body
A strong environmental test does two things: it monitors behavior during or immediately after exposure — logs, counters, sensor readings, connectivity, reset records — and it inspects the physical assembly afterward at the parts the stress would affect. A pass label with neither is weak. The record should name the assembly and configuration tested, the exposure profile and why it was chosen, the function under review, and any condition assumed rather than measured.
12.2.2 Practitioner Knowledge Check
If you can choose the stress, set the device state, and capture both behavior and inspection, you can stop here. Continue to Under the Hood for accelerated stress, combined exposure, and why a material change reopens a result.
12.3 Know What the Result Really Covers
The deeper layer is about getting more signal from physical tests and being honest about what they bound: measuring behavior rather than survival, finding margins with accelerated stress, testing coupled conditions together, and treating the physical assembly as part of the result. In review shorthand, this is where survival is not function, qualification versus accelerated stress, and combined exposure stop being slogans and become evidence checks.
12.3.1 A Survived Case Can Still Fail
The recurring trap is accepting “no visible damage” as a pass. A device can emerge from a thermal or vibration run intact yet drift out of calibration, drop packets, or fail to rejoin the network. Qualification needs the monitored behavior that proves the function under review held, not just that the case survived.
Tessa’s Field Check
- Bench says: no visible damage — a device can emerge from a thermal run intact yet still drift or drop packets.
- Field adds: the split between qualification, meeting the rating, and accelerated stress such as HALT and HASS, finding the margin.
- Stamp when: monitored behavior proves the function held — not that the case merely survived.
12.3.2 Separate a Pass From a Margin Search
Two different goals use stress differently. Qualification confirms the device meets its requirements within its rated conditions. Accelerated stress — testing beyond the rated range, as in highly accelerated life testing (HALT) and stress screening (HASS) — deliberately pushes past spec to discover the margins where the design actually breaks and to screen out early-life failures. They answer different questions: one says “it meets the rating,” the other says “here is how much headroom there is and where the weak point is.” Keep which one a result came from explicit, because a margin-finding failure beyond spec is not the same as a qualification failure.
12.3.3 Retest When the Physical Boundary Changes
Some faults appear only when conditions combine — vibration while hot, moisture under enclosure strain — so combined exposure is worth running when a failure is suspected to be coupled, with the rationale recorded. And every environmental result is bound to the physical boundary that produced it: the enclosure, gasket, adhesive, connector, sensor opening, and mounting. When one of those changes — a new gasket material after a passed moisture test, for instance — the old result no longer covers the new boundary, and the test must be rerun or its limits restated. That is what a retest trigger records.
12.3.4 Common Review Findings
- Survival mistaken for function. “No visible damage” is accepted as a pass with no monitored behavior.
- Wrong device state. The unit was idle during exposure when the field state is transmitting.
- Unstated assembly. A bare-board result stands in for a sealed unit because the enclosure or connector state was not recorded.
- Wrong inspection. The post-exposure check looked at the wrong parts for the stress applied.
- Stale after a change. A material or assembly change left the old result in place with no retest trigger.
12.3.5 Under-the-Hood Knowledge Check
At this depth, environmental testing is physical validation bound to a physical configuration: exercise the right stress in the right device state, measure behavior rather than survival, separate qualification from margin-finding, and rerun when the assembly that earned the result changes. A trustworthy review names the assembly, the exposure, and the change that reopens the conclusion.
12.4 Summary
- Environmental and physical testing exposes the real assembly to controlled field conditions and confirms it still works, catching physical failure modes software tests cannot reproduce.
- Test the deployed assembly — enclosure, seals, connectors, sensors, mounting — in the state the field will see, not a bare development board.
- The common stress categories are temperature (including thermal cycling), humidity and ingress, vibration/shock/drop, electromagnetic compatibility and ESD, power disturbance and cycling, and soak/endurance.
- The device’s state during exposure is decisive; monitor behavior during or after, and inspect the parts the stress would affect.
- “No visible damage” is not a functional pass; qualification needs monitored behavior showing the function under review held.
- Qualification (meets the rating) is distinct from accelerated stress such as HALT/HASS (finds margins and screens early-life failures); keep which one produced a result explicit.
- Every result is bound to the physical boundary that produced it; a gasket, enclosure, connector, sensor-opening, or mounting change reopens it, which is what a retest trigger records.
Environmental testing connects real temperature, humidity, vibration, EMC, power, and endurance stress to measured device behavior on the actual assembly — not to a survived photo. Exercise the right device state, measure function rather than survival, know whether you are qualifying or finding margins, and rerun when the physical boundary that earned the result changes.
12.5 See Also
Testing and Validation for IoT Systems
See where environmental testing sits among the IoT test levels.
Integration Testing for IoT Systems
Check component interactions before physical exposure evidence is reviewed.
Carry controlled environmental evidence into real operating contexts.
Security Testing for IoT Devices
Review debug-boundary and physical-interface exposure alongside environmental stress.