Most connected products fail in the field for boring reasons: a connector that loosens after a few hundred temperature swings, a battery that refuses to charge in a cold van, condensation that creeps under a gasket overnight. Environmental testing is how you find those failures before your customers do, but a full lab campaign is hard to justify when you are building ten prototypes rather than ten thousand units.

This post is for founders and engineers in exactly that position. It covers how we decide which stresses matter, what can be done credibly on a bench, and where paying for chamber time at an accredited lab is the cheaper option in the end.

What environmental testing is for at prototype stage

At prototype stage the goal is not a certificate. It is information: which part of the design breaks first, under which condition, and how much margin you have. A pass/fail result on three units tells you very little statistically, but a failure on one unit tells you a great deal about the design.

That changes how you plan. Instead of reproducing a full qualification programme, you run shorter, harsher, well-instrumented tests aimed at the weak points you already suspect. Formal qualification comes later, on production-intent hardware, once the design has stopped moving.

Pick the stresses from the use case, not the standard

The IEC 60068-2 series is the usual reference for environmental test methods, and it is worth knowing the parts that come up most often:

Cold and dry heat: IEC 60068-2-1 (Test A) and IEC 60068-2-2 (Test B).
Change of temperature: IEC 60068-2-14 (Test N), the basis for most thermal cycling and shock profiles.
Damp heat: IEC 60068-2-78 for steady state, often run at 40 °C and 93% RH, and IEC 60068-2-30 for cyclic damp heat, which drives condensation.
Mechanical: IEC 60068-2-6 (sinusoidal vibration), IEC 60068-2-64 (random vibration) and IEC 60068-2-27 (shock).

The standards define methods, not which ones your product needs. That comes from the use case. A sensor bolted inside an unheated agricultural building sees daily thermal cycles and high humidity; a handheld that lives in a pocket sees drops and sweat; a unit on a vehicle sees vibration for its whole life. Write down the realistic worst case for each stress, then add margin, rather than copying the most severe profile in the standard.

Two limits are worth checking early because they often set the real envelope. Many components are only rated for the commercial range of 0 to 70 °C, while industrial parts are typically rated from −40 to 85 °C. And lithium-ion cells usually allow charging only within a narrower window, often around 0 to 45 °C, so check the cell datasheet before you promise anything about winter use.

What we run in-house

A surprising amount of useful environmental testing needs no chamber at all. On a small-batch budget we lean on simple equipment and good instrumentation:

Cold soaks in a freezer, which typically holds around −18 °C. It will not reach −40 °C, but it reliably exposes cold-start problems, sluggish displays and batteries that will not charge.
Warm soaks in a temperature-controlled oven with thermocouples on the board, to find the hottest components and check regulator and radio behaviour near the upper limit.
Crude thermal cycling by moving units between cold and warm environments. It is not a controlled ramp, but repeated cycles still find cracked solder joints, loose connectors and enclosure clips that creep.
Drop and handling tests from realistic heights onto a hard floor, in several orientations, with the unit powered and logging.
Water ingress spot checks, such as a 1 m immersion for 30 minutes as in the IPX7 test of IEC 60529, when the enclosure is meant to meet it. This is a sanity check, not a rating.

The instrumentation matters more than the equipment. The firmware logs supply voltage, temperature, reset causes and radio errors throughout the test, so a failure leaves a trace rather than a mystery. A thermal camera and a handful of type K thermocouples cover most of the rest.

One trap to respect: taking a cold unit straight into warm, humid air causes condensation on and inside it. That is a useful stress if you want it and a misleading failure if you do not, so decide in advance and bag the unit if needed.

What we send to a lab

Some environmental testing cannot be faked on a bench. Controlled humidity, fast ramp rates, calibrated vibration and dust ingress all need proper equipment, and accredited labs will rent chamber time by the day or run a defined test for you.

Test	Bench-friendly?	Why
Cold and warm soak	Yes, within limits	Freezer and oven cover most functional checks
Thermal cycling with controlled ramps	Partly	Profile and dwell times need a chamber to be repeatable
Damp heat, steady or cyclic	No	Holding high relative humidity accurately needs a climatic chamber
Vibration and shock	No	Needs a shaker table and calibrated fixtures
Dust ingress (IP5X/IP6X)	No	Needs a dust chamber with controlled powder and vacuum
Salt mist	No	Needs a dedicated salt spray cabinet

To keep lab costs down, we arrive with everything ready: units pre-screened on the bench, logging firmware installed, fixtures and harnesses built, and a short written test plan. Lab time is billed whether or not your units boot, so the hour spent finding a flat battery should happen at home.

Making a small sample size count

With a handful of prototypes, every unit is precious, so sequence the tests deliberately:

Non-destructive first. Soaks and functional checks before drops, immersion and vibration.
Combine stresses where it is honest to do so. Running the radio and logging during a thermal test costs nothing and doubles what you learn.
Keep one golden unit out of the abuse entirely, as a reference for comparing behaviour after testing.
Inspect after every stage. Photos, a quick X-ray or cross-section if something looks suspicious, and a note of anything that changed.

Treat every failure as a design input. A failure on one unit out of three is not a 33% failure rate. It is one confirmed failure mechanism, and finding mechanisms is the point. This is the same thinking behind the pre-certification reports we write as part of our testing and verification services: the fix list is worth more than the pass/fail column.

A checklist before you book chamber time

Write the realistic environment for the product: temperature range, humidity, condensation, vibration, drops and water exposure.
Check the narrowest temperature rating in the bill of materials, including the battery’s charging window.
Add logging to the firmware for supply voltage, temperature, resets and radio errors.
Run cold and warm soaks, drops and simple cycling in-house first, and fix what breaks.
Book lab time only for humidity, vibration, dust and salt, with a written plan and pre-screened units.
Keep one unit untouched as a reference.
Record each failure as a mechanism with a proposed fix, not just as a fail.

Done this way, environmental testing on a small batch becomes a cheap way to buy confidence before tooling and certification, which are the expensive steps. If you are weighing how much testing to plan into a project, our FAQ on cost and timelines is a good place to start.