How We Test POS Terminals Before Shipping: Six Benches and Their Pass Limits

Reliability testing is not a certificate. It is a search for the point where the design margin runs out — and each of the six benches looks for a different failure.

A reliability test report that says “passed” tells you almost nothing. What it should tell you is where the margin ran out first — because whatever failed at the margin is what will fail in the field, only later and only on the units you sold to your most demanding customer.

Six benches cover a POS terminal. Each one answers a different question, and only the first one runs on every unit. The other five are run on samples from each model and each lot, which is the difference between testing and quality control — a distinction worth keeping straight, and one this article will come back to.

Bench 1: the functional station, every unit, ninety seconds

Every terminal goes through a functional station where a fixture connects power, Ethernet, cash drawer, printer and scanner, and a script exercises the hardware in a fixed order. The takt is 90 seconds, which sets the throughput of the whole line: 400 units per shift per station, and a second station the moment you need more.

The checks are deliberately physical rather than software-driven. The touch panel is swept along all four edges and the centre, because a dead column at the border is the defect that survives an icon-tapping test. The printer is fired for a fixed pattern that exercises the full 80 mm width. The drawer kick is measured as a pulse, not as a click. The USB ports are enumerated one at a time and in combination, because two devices that each work alone and fail together is a power budget problem that only shows up with a scanner and a printer attached.

The failure that gets through

A unit that passes every test and reboots when the drawer kicks has a 12 V rail that sags under a simultaneous print and kick. That is a power design issue, not an assembly defect, and no amount of functional testing will catch it — which is why bench 2 exists.

Bench 2: burn-in, 48 hours at 45 °C

Burn-in runs the terminal at elevated ambient temperature with the screen on, the printer cycling and the radio associated, for 48 hours. The purpose is infant mortality: the small population of components that fail in the first hundred hours fail here, in a rack, instead of in a restaurant.

The temperature is chosen to accelerate without changing the failure mode. 45 °C is roughly 20 degrees above a typical indoor ambient, which roughly doubles the rate of most chemical failure mechanisms and does not soften adhesives or push a battery into a state it would never reach in service. Pushing to 60 °C to shorten the test is a common mistake: it produces failures that would never have happened, and hides ones that would.

The yield from burn-in is the most useful number a factory can report. Expect 0.5 to 2% to fail on a mature product, and expect the first three lots of a new model to be worse. A supplier quoting zero is either not running the rack or not counting what comes off it.

Bench 3: drop, at three different heights for three product classes

There is no single drop specification, and quoting one number across a product range is a sign that nobody wrote the test plan. The height follows the device, because the device follows the hand.

Product class Bare unit Packaged What fails first
Handheld with printer 1.2 m onto concrete, six faces, three edges, one corner ISTA 1A, 76 cm Paper compartment latch, then the screen corner
Countertop terminal 0.5 m, corner and back ISTA 1A, 76 cm Rear customer display hinge, then the drawer cable
Kitchen display / panel Not dropped ISTA 1A plus a 30 cm edge drop on the carton Corner of the panel, then the mounting bracket
Self-service kiosk Not dropped ISTA 2A with the pallet Pallet corner, then the door hinge

The orientation matters more than the height. A flat drop spreads the load across the whole back cover. A corner drop puts the entire energy of the fall into a few square millimetres of plastic, which is why the published figure — almost always flat onto a concrete slab — is the easiest case. Ten drops covering six faces, three edges and one corner is the sequence that finds the real weak point, and the weak point is nearly always a latch or a hinge rather than the electronics.

Bench 4: the thermal chamber, and the step that fails

The standard cycle is −10 to 55 °C, five cycles, four hours of dwell at each end and a ramp of about 1 °C per minute. The dwell is long enough for the whole mass to reach temperature; the ramp is where the interesting failures happen.

Three things go wrong in the chamber, and only one of them is about the electronics. The first is condensation: as the chamber warms, moisture condenses on any surface still below the dew point, and a terminal powered on in that state can see leakage across a connector that is dry at every other moment. The second is the display: a liquid crystal response slows markedly below 0 °C and a panel rated to 0 °C shows a visible smear at −10, which is cosmetic but alarming. The third, and the one that matters commercially, is the battery: a lithium pack cannot be charged below about 0 °C, so a handheld brought in from a cold delivery round will refuse to charge until it warms up. That is correct behaviour and it will still generate a support ticket every winter.

Bench 5: ESD, and what criterion A actually means

Electrostatic discharge testing follows IEC 61000-4-2. The contact discharge required for CE is ±4 kV and the air discharge ±8 kV; a terminal built for high-traffic retail is usually run at ±6 kV contact, because a customer reaching for a card reader in a dry January carries more than the minimum.

The result is graded, and the grade is where buyers get lost. Criterion A means the unit behaved normally throughout. Criterion B means it recovered by itself, which for a touch terminal can mean a visible flicker or a momentary loss of the touch link. Criterion C means it needed a restart, and criterion C is a fail.

The specific failures are worth knowing because they are all recognisable in the field. The touch controller latches up and stops reporting until power is cycled. The USB host re-enumerates, so the scanner disappears from the application for two seconds. An HDMI link to a second display drops and does not come back. None of these destroy the unit, and all three generate a service call.

Bench 6: the touch panel does not wear out from tapping

This is where most touch specifications are dishonest. A capacitive panel has no moving parts in the sensing layer, so a rating of ten million touches describes a mechanism that does not exist. The number is copied from resistive panels, where tapping genuinely did wear out the spacer dots.

What does wear out is the surface. Three things, in order: the anti-fingerprint coating, which wears through in the centre of the screen where the cashier’s thumb lands; the cover glass itself, which scratches on a grit particle trapped under a cleaning cloth; and the printed black border, which lifts at the edge after repeated wiping with a strong cleaner.

So the meaningful tests are abrasion and chemical resistance, not tap count. An abrasion test runs an 8 mm pen tip under 250 g of load across the surface and checks the haze change. A chemical test wipes the panel five hundred times with 75% alcohol and checks that the border print and the coating survive. A terminal in a bar gets wiped with whatever is behind the counter, and that is usually a quaternary ammonium cleaner, not the mild glass cleaner in the datasheet.

Where the failures actually come from

Across a year of returns, the distribution is stubbornly consistent and it is not centred on the parts that get the most engineering attention.

Rank Failure Share of returns Bench that should catch it
1 Backlight connector / display Roughly a third Burn-in, thermal cycling
2 Battery swell or capacity loss A fifth to a quarter Not caught by any bench — it is a calendar effect
3 Printer mechanism and paper path Around a sixth Functional, plus a hundred-receipt run
4 Touch panel crack after a drop Around a tenth Drop, corner orientation
5 Charging port or cradle contacts Under a tenth Insertion cycle test, not a drop test

The battery line is the uncomfortable one. No bench in a factory catches calendar ageing, because the failure is a function of time at temperature and state of charge, and the accelerated version of that test takes months. It is also the single largest cause of handheld returns, which is why buying on cell grade and on charging policy is worth more than buying on capacity.

What a test report should contain

Ask for the report, not the certificate. A useful page lists, for each bench: the standard referenced, the level applied, the sample size, the number of failures, the failure mode and the disposition. A report with a pass stamp and no failure modes has never seen a production sample, because production samples always fail something.

The question to ask at the end of it is not “did it pass” but “what failed first, and what did you change”. The answer tells you whether the margin was used up on purpose or by accident.

FAQ

How long does reliability testing take for a new model?

Six to nine weeks if the chambers are free and the labs are booked in advance. Thermal cycling, drop and ESD run in parallel; the critical path is usually certification rather than the in-house benches.

Is a higher drop rating always better?

No. A terminal rated to 1.5 m usually gets there with a thick rubber overmould and a recessed screen, which costs bezel width, adds weight and makes the screen harder to clean. For a countertop unit that never leaves the counter, the extra rating buys nothing.

Why does my terminal fail after a power cut but not in the lab?

Brownouts and repeated short interruptions are a different test from a clean switch-off. Ask specifically whether the unit was tested with 200 interruptions of 200 ms at random phase, which is the test that exposes a filesystem that is not mounted safely.

Where to go next

The useful output of a test plan is a list of what broke first, not a stamp. If you are specifying a build, the same questions apply to the terminal and to the printer:

DCA-155A 15.6-inch Android terminal POS-8370 80 mm receipt printer DCA-H62 handheld the batch control that runs alongside these benches why receipt printers fail in kitchens drawer kick voltage and the 12/24 V mix-up · panel, printer and board in one chassis · what a thermal printer can and cannot do

Specifying the hardware? The hardware side is covered in the POS system, by counter position — where each part sits and which of them you can stop buying separately.

DcaPOS is a POS hardware and commercial display manufacturer.

Latest Post

Need Any Help?

Get Your Custom POS Hardware Quote

Request a Quote?

You May Also Like

Touch Screen POS Terminal: What the Panel Decides

Touch Screen POS Terminal: What the Panel Decides

Size, resolution and brightness are on the sheet. Aspect ratio, touch stack and backlight position are not, and they decide how many order lines fit, whether a glove works, and whether the screen survives a window.

Cash Drawer for Square: The Port You Cannot See

Cash Drawer for Square: The Port You Cannot See

A printer-driven drawer never appears in Square's hardware list, so there is no model, no voltage and no error message to read. The eight-row compatibility table admits as much, and it changes what you should order first.

Barcode Scanner for Square: HID Mode Decides It

Barcode Scanner for Square: HID Mode Decides It

The compatibility list has nine models. The reason yours is not on it is almost never the model. It is the interface mode, the report suffix and which of the three hub ports you have left.

DcaPOS-POS Hardware Manufacturer

+86 133 1295 3271
Furong Lantian Industrial Park, Xinqiao,Shajing Town,Shenzhen China.
Contact Us Now