A commercial POS terminal does not usually die. It becomes unsupportable. The distinction matters because unsupportable arrives years before broken, and it arrives as a procurement event rather than a service call: a panel that is no longer manufactured, a battery in a shape nobody makes any more, a firmware branch that stopped receiving patches.
The lifecycle write-ups that describe twelve stages from concept to end-of-life spend eleven of them on development. This one is about the twelfth, because the twelfth is the only stage that costs the buyer money after the purchase order is closed.
What stops first
Five things end a terminal’s life, and they do not end it at the same time. The order below is the one that holds in practice, and it is not the order most buyers plan for.
| Asset | Typical support horizon | What the buyer notices |
|---|---|---|
| Display panel | 3–5 years from design | The model is discontinued because the panel is gone, not because the terminal failed |
| Battery pack (handheld) | 2–4 years | Packs become unavailable while the units are still working, and a handheld without a pack is scrap |
| SoC board support package | 4–6 years | No further Android versions, then no security patches |
| Certification | Until the standard is revised | A new production run needs a retest against a newer edition |
| Plastic tooling | 300,000–500,000 shots | A repair housing cannot be moulded, so a cracked bezel writes off the unit |
Read the first line twice. The most common reason a POS model disappears from a catalogue is not that sales were poor. It is that the panel went end-of-life and the replacement does not fit.
The panel is the first part to disappear
A 15.6-inch panel is not a commodity that a factory can buy from anyone. It is a specific part with a specific interface — usually 30 or 40-pin eDP — a specific backlight current, a specific outline and a specific mounting hole pattern. When the panel maker ends that part, the mechanically compatible alternative is usually a different backlight voltage and a different connector position.
The consequence is not a swap. New backlight current means a driver change on the mainboard or a different backlight board. A moved connector means a new flex and often a new bracket. Both mean a new firmware build to set the panel timing. That is a redesign of the display assembly, and it is why a terminal maker discontinues a model rather than fighting to keep it.
The question that predicts discontinuation
Ask for the panel part number and its lifecycle status at the time of purchase. If the panel is already marked end-of-life or not-recommended-for-new-designs, the terminal has a shorter commercial life than the warranty suggests — and the vendor knows it.
The battery is the second, and it is worse for handhelds
A handheld pack is a pouch cell in a custom shape, often with a moulded housing and a specific connector. A 5,000 mAh pack measuring roughly 90 by 60 by 8 mm is not a catalogue item in year four; it was a tooling decision in year zero.
This makes the battery the single most consequential spare part in a handheld fleet. The cells inside it are standard, but the assembly, the protection board and the plastic carrier are not, and a pack supplier will not run 200 pieces of a discontinued shape at a sensible price. The practical result is that handhelds get retired while working perfectly, because the pack cannot be bought.
Two clauses reduce the damage, and both have to be agreed before the first order. A stated minimum pack production horizon — five years from the last unit shipped, not from the first — and a commitment that the pack remains available to the buyer after the terminal itself is discontinued. The second clause is the one that gets forgotten and the one that matters.
Firmware stops before the hardware does
The firmware horizon is set by the chip vendor, not by the terminal maker. Android builds for a given system-on-chip depend on that vendor’s board support package; when the chip moves to end-of-life, the BSP freezes at whatever Android version it last supported, and no amount of work produces a version the silicon was never validated against.
What breaks, in order: the certificate store stops updating and TLS connections to payment gateways begin to fail one root at a time; the payment application raises its minimum API level and the new release will not install; and the device management agent drops the platform, so the terminal vanishes from remote monitoring. None of these is a hardware fault, and all three end the deployment.
A useful question at purchase is not “which Android version” but “which SoC, what is its lifecycle status, and what is the committed patch horizon”. A terminal shipping Android 11 on a chip that stopped receiving BSP updates in 2024 has a shorter life than its warranty.
Certification expires by revision, not by date
A test report does not expire, but the standard it was written against gets revised. A unit already in the field can generally keep being sold under the edition it was tested to, for a transition period. A new production run, or a meaningful design change, needs a test against the current edition.
There is also a definition problem worth knowing about. CE is a self-declaration: the lab issues a test report, and the legal document is the Declaration of Conformity with a technical file behind it. A supplier who sends a folder containing a lab report and calls it a certificate has not provided the document a customs officer or a customer’s compliance team will ask for. Ask for the DoC by name.
The clauses that cover it
Six clauses in a purchase agreement do most of the work. They are cheap to ask for at the start of a relationship and almost impossible to add later.
| Clause | What it should say | Why it matters |
|---|---|---|
| Spare parts horizon | Five years from the last shipment of the model, with a published price list | Without a price list, availability is theoretical |
| Security patches | Three years of security updates after the last unit ships | Payment software will require them |
| End-of-life notice | Twelve months written notice before discontinuation | Enough time to run a last-time buy |
| Last-time buy right | The buyer may order any quantity at the last published price | Prevents a discontinuation surcharge |
| Tooling ownership | On a custom build, the tool is the buyer’s and is released on request | Otherwise moving supplier means paying for the mold twice |
| Panel and SoC disclosure | Part numbers stated at purchase, with lifecycle status | The only forward-looking signal available |
Two of these are routinely refused and are still worth asking for. Firmware source escrow is refused by almost every terminal maker, because the source contains third-party licensed code; a build-and-hold agreement, where the vendor keeps a compiled image and the toolchain for five years, is the achievable version of the same idea.
Last-time buy: the arithmetic that decides it
When a discontinuation notice arrives, the decision is how many units to buy and hold. The arithmetic is simple and the storage cost is not.
Take 300 installed terminals, a 3% annual failure rate and a five-year horizon: that is nine replacements a year, forty-five over the horizon, plus a buffer for the failure rate rising as the fleet ages — call it sixty units. Sixty terminals is capital tied up in a store room, and it is also sixty batteries quietly ageing on a shelf, because a lithium pack stored at room temperature for four years will not hold a shift when it is finally used.
The practical resolution is to buy the units and not the packs, and to buy packs in two tranches with a late delivery date. It is also worth checking whether the same mainboard exists in a different housing: often the electronics outlive the enclosure, and a service replacement can be a different model that shares the platform.
Refurbishment: what can be replaced, and what cannot
A refurbished unit is worth doing when the replaceable parts are the ones that wore out, and worth avoiding when the worn part is structural. On a terminal the line is unusually clear.
| Part | Replaceable in the field | Cost relative to a new unit | Verdict |
|---|---|---|---|
| Display assembly | Yes, as a unit | A quarter to a third | Worth it if the panel is still available |
| Battery pack | Yes, if it is a catalogue part | Under a tenth | Always worth it — and the reason to check the pack is a catalogue part before buying |
| Print head and platen | Yes | Under a tenth | Wear item, replace on schedule |
| Mainboard | Yes, but it carries the serial and the licence | A third to a half | Only if the same board revision exists |
| Housing and bezel | Only while the tool exists | Low, if moulded; impossible after EOL | The part that ends refurbishment programmes |
| Touch panel alone | No, on a fully laminated assembly | Whole assembly | Not a field repair |
The pattern is that refurbishment is viable exactly as long as the panel and the housing can both be sourced, which is another way of saying it is viable for as long as the model is in production plus the spare parts horizon. Once the tool is retired, a cracked bezel writes off a working mainboard, and the fleet turns into a parts donor programme by accident.
What to do at year three
Run an audit of the installed base by serial number, not by model name, and mark every unit against three dates: panel lifecycle status, battery availability, and firmware patch horizon. Any unit with less than twelve months on any of the three is a replacement candidate, and the sites with the highest revenue should be replaced first.
The second action is standardisation. A fleet of six models across forty sites multiplies every spare parts problem by six; two models per site class, bought in a single order, gets better pricing and a single set of spares. The cost of variety in POS hardware is paid in year four, not on the invoice.
FAQ
How long should a POS terminal last?
Seven to ten years of service is achievable for a countertop terminal, and five to seven is realistic for a handheld because of the battery. The constraint is rarely the electronics; it is whether the panel and the pack can still be bought in year six.
Is a discontinued model a bad buy?
Not necessarily, if the price reflects it and the spare parts horizon is stated. A discontinued model bought without a parts commitment is a bet that nothing breaks.
Can firmware be updated after the vendor stops?
Only with the source and the signing keys, which almost no vendor releases. The realistic protection is a stated patch horizon and a plan to replace the fleet before it expires.
Where to go next
Buy the support horizon, not just the unit. The models and the platform decisions behind them:
DCA-155A 15.6-inch Android terminal DCA-156A aluminium bezel DCA-H62 handheld, removable pack the four numbers behind lifespan where a new model starts, and what gates it service and spare parts · the footprint cost of an all-in-one · mini PC boxes for factory automation
