The difference between "it works" and "it lasts"

A prototype running on a desk has proven exactly one thing: the principle is sound. It has not proven that the product survives a fall from a countertop, a summer inside a parked car, a thousand open-close cycles, or a cleaning wipe soaked in a harsh solvent. The gap between those two states is the main source of returns, bad reviews, and warranty cost - and it is closed by reliability testing.

Three named activities do most of the work. HALT - highly accelerated life testing - deliberately stresses units beyond specification to find design weaknesses fast. HASS - highly accelerated stress screening - applies a milder version of that stress to production units to catch manufacturing defects before shipment. Life testing estimates how long the product will actually last under representative use. They answer different questions, and a serious program uses all three at different points.

Start by defining what "reliable" means for your product

Reliability is meaningless without a mission profile: how many hours per day, how many cycles per year, over what expected service life, in what temperature and humidity range, handled by whom. Write it down. Every test condition and pass criterion derives from it.

Be careful with single-number reliability claims. A figure like mean time between failures is a statistical statement about a population under stated conditions, not a promise about one unit, and quoting one without its conditions is close to meaningless. What actually helps is a table of tests, conditions, sample sizes, and pass criteria - agreed before anyone touches a chamber.

Mechanical durability

  • Drop. From a realistic use height onto a hard surface, in every orientation. A handheld is dropped from waist height; a countertop appliance from counter height. Design intent for this is discussed in designing for drop test.
  • Static load and crush. What happens when someone leans on it, sits on it, or steps on it.
  • Vibration and shock. Relevant to anything that travels in a vehicle or ships by parcel carrier. A large share of failures that appear "on arrival" are caused in transit, not in use, and standardized transit test protocols exist specifically to reproduce that.

Environmental exposure

  • Temperature. High and low extremes, and especially rapid transitions, which drive condensation inside enclosures.
  • Humidity and water. Spray, drip, and immersion as applicable. The target level is expressed as a standard rating, and choosing it early drives seal, gasket, and vent design - see IP ratings explained.
  • UV and weathering. Unstabilized plastics yellow, chalk, and embrittle within months outdoors. Material choice decides this outcome, which is why it belongs in material selection rather than in a late fix.
  • Dust, sand, and salt. Essential for outdoor, marine, and jobsite products.
  • Chemical compatibility. Cleaning agents, sunscreen, sweat, and cooking oils crack plastics that pass every mechanical test.

Cycle and life testing

Anything that moves gets cycled: lids opened and closed, buttons pressed, connectors mated, hinges flexed, batteries charged and discharged. Cycle counts come from the mission profile multiplied by a safety factor. This is where fatigue and wear failures appear - the ones that never show up in a single-shot test. Where a failure looks structural, back the fix with analysis rather than another guess, as described in FEA simulation in product design.

Foreseeable misuse

Test what happens when a connector goes in backwards, the wrong charger is used, the device is submerged when it was only rated for splash, or someone operates it without reading anything. Safety standards explicitly address "reasonably foreseeable misuse", so this is not a theoretical exercise - it overlaps directly with the requirements outlined in product safety testing requirements.

HALT and HASS, specifically

HALT is a step-stress exercise on a very small number of units. You raise temperature, then vibration, then combine them, past the specification and onward until something breaks. You fix or note the weak link, then continue to the next one. The goal is not a pass certificate - it is a ranked list of the design's weakest points, obtained in days rather than months. Nothing else finds design margin so quickly or so cheaply.

HASS takes what HALT taught you and applies a reduced, non-damaging stress profile to every production unit or to a sampled fraction, to precipitate latent manufacturing defects - a cold solder joint, a marginal crimp - before the customer receives them. HASS is a production process, so its cost and cycle time have to be designed into the line, not bolted on.

Accelerated life testing and its two caveats

A product meant to last five years cannot be tested for five years. The standard answer is acceleration: elevated temperature, elevated humidity, faster cycling, to compress years of wear into weeks.

Two limits matter. First, acceleration is only valid while the failure mechanism stays the same. Overheating can create a failure that would never occur in normal use, sending the team off to fix an imaginary problem. Second, the output is an estimate, not a guarantee. The most reliable way to use accelerated results is comparatively - between two design revisions, two materials, or two suppliers - rather than as a precise life prediction. Combining it with real field data is what produces a complete picture.

When to test

The common mistake is testing once, at the end. The efficient sequence is progressive:

  • On early prototypes: crude, unscientific abuse. Drop it, squeeze it, leave it in the sun. Finds gross failures at almost zero cost.
  • Through the engineering build phases: structured testing on units made by increasingly production-representative methods. This maps onto the stage gates described in EVT, DVT, and PVT - design verification is where the formal reliability campaign belongs.
  • On the pilot run: the most informative point, because final material and final process behave differently from hand-built units.
  • At an accredited lab: only what certification actually requires, and only after the product has passed your internal testing. Paying lab rates to discover a problem you could have found on a bench is expensive.
  • In the field: units with real users and a structured reporting channel, which reveals usage patterns nobody predicted.

What to do with a failure

A failure at this stage is the point of the exercise. Ten units broken in a lab is vastly better than a thousand broken at customers. Each failure gets driven to root cause, corrected in the design, and re-tested under a controlled process - see the engineering change order process for how those changes should be handled so the fix actually reaches production.

Build a test plan that means something

Projects House writes reliability test plans with defined conditions, sample sizes, and pass criteria, runs the internal testing, and prepares products for accredited lab work - so failures happen on our bench instead of at your customers. More on the full development sequence is on our new product development hub. Send us your product and mission profile through the contact form and we will draft the plan.