The five types, one at a time
125 kHz proximity — the legacy layer
The beige-and-white workhorse of South African estates, office parks and older factories, retailing locally from around R10. It broadcasts a fixed serial number with no encryption whatsoever, which had an honest logic in 1995 and none today: handheld duplicators that clone one in under ten seconds are consumer products, and key-cutting kiosks copy them as a service. That does not make the technology useless — it makes it an identifier, fine for low-stakes convenience (a gym locker, a time-and-attendance tap where fraud has other controls). It is not a lock, and any door that matters deserves better.
MIFARE Classic — the broken middle
NXP's 13.56 MHz MIFARE Classic became the default "smart" card of the 2000s — transit systems, campuses, thousands of SA sites — because it added memory sectors and a proprietary cipher, Crypto1. That cipher was publicly broken in 2008 and attacks have only industrialised since; cloning tools are open-source. The practical read: a MIFARE Classic estate is not an emergency, but it is a decision already made for you — new issuance should not be Classic, and the migration path matters more than the panic. Classic cards still serve honourably wherever the card is a token rather than a secret: membership, print release, canteen balance backed by a server-side ledger.
MIFARE DESFire EV2/EV3 — the current answer
Same 13.56 MHz band, entirely different security: DESFire runs AES cryptography with mutual authentication — the reader must prove itself to the card before the card says anything meaningful — plus per-application keys, so one card can carry access, payment and identity applications that cannot read each other. There is no practical cloning path. For any new credential deployment where the card guards something, DESFire EV2 or EV3 is the boring, correct choice, and the cost difference per card has shrunk to the point where choosing Classic for a new system needs a justification.
NFC mobile credentials — the card becomes software
Phones carry credentials in a secure element and present them over the same 13.56 MHz interface. The security story is strong — cryptographic credentials, remote revocation the moment a phone is reported lost, no card printing pipeline to manage. The operational story needs honesty: staff without compatible phones, battery-dead phones at 05:00 shift start, and BYOD policy questions mean most sites run mobile alongside cards rather than instead of them. As a second credential for managers and contractors it is excellent; as the sole credential for a 400-person plant it is premature for most SA operations.
UHF cards and windscreen tags — the different question
A UHF "card" on a dashboard is not an access card that grew up; it is a different band (860–960 MHz, EPC Gen2) answering a different question — which vehicle is at my gate, from metres away, without stopping. It has no place on doors (the range that makes it right for vehicles makes it wrong for proving a person is present) and it pairs naturally with the short-range card estate: DESFire for the person at the turnstile, UHF for the vehicle at the boom. Our gate automation page covers that architecture, and our technology comparison maps the full frequency ladder.
The five types side by side
| 125 kHz prox | MIFARE Classic | DESFire EV2/EV3 | NFC mobile | UHF vehicle tag | |
|---|---|---|---|---|---|
| Frequency | 125 kHz (LF) | 13.56 MHz (HF) | 13.56 MHz (HF) | 13.56 MHz (HF) | 860–960 MHz (UHF) |
| Standard | Vendor formats (EM4100 class) | ISO 14443A | ISO 14443A-4 | ISO 14443 / NFC Forum | EPC Gen2 / ISO 18000-63 |
| Read range | ~2–10 cm | ~2–10 cm | ~2–10 cm | ~2–4 cm | 1–12 m |
| Memory | Fixed serial only | 1K / 4K, sectored | 2K–16K, per-app files | Secure element | 96-bit EPC + user memory |
| Security | None | Crypto1 — broken 2008 | AES-128 mutual auth, CC EAL5+ | Phone security stack, revocable | Password/TID options — an identifier, not a credential |
| Cloning risk | Trivial — seconds | Easy — open tools | No practical path | No practical path | Copyable unless TID-checked |
| Card cost tier | Lowest (from ~R10 retail) | Low | Moderate — the premium has shrunk | Licence/credential fees, no card stock | Low (windscreen) to moderate (rugged on-metal) |
| Right for | Low-stakes convenience only | Tokens backed by server-side checks | Any door or tap that matters | Second credential, managers/contractors | Vehicles at gates and booms |
Passive or active — what powers the card
Every card type above is passive RFID: the card has no battery and draws its power from the reader's field, which is why it works for decades, costs rands rather than hundreds of rands, and reads only at the range the physics of its band allows. That trade is the right one for credentials — a card that never dies in a drawer is exactly what an access system wants. Active RFID puts a battery-powered transmitter in the tag: read ranges stretch to tens or hundreds of metres, which is what electronic tolling transponders and some long-range vehicle and rail tags use — at the price of a battery that expires (typically 3–5 years), a much higher unit cost, and a replacement programme forever. Between them sits battery-assisted passive (BAP): the battery boosts the tag's reply for extra range, but the reader still initiates, mostly seen in sensor-carrying tags rather than credentials.
| Passive | Battery-assisted (BAP) | Active | |
|---|---|---|---|
| Power source | Reader's field only | Battery boosts the reply; reader still initiates | Battery-powered transmitter |
| Typical range | cm (LF/HF) to 1–12 m (UHF) | Tens of metres | Tens to hundreds of metres |
| Tag life | Decades — nothing to expire | Battery-limited | 3–5 years per battery |
| Unit cost | Rands to tens of rands | Higher | Hundreds of rands up |
| Maintenance | None | Battery programme | Battery programme, forever |
| Where it belongs | All personnel credentials; vehicles at engineered gates | Sensor-carrying tags (temperature loggers) | Open-road tolling; wide-area yard location |
The buying rule is simple: people carry passive, and almost always should — there is no access-control scenario where a personnel card needs a battery. Vehicles at a gate also do fine on passive: a well-designed read point reads a passive on-metal or windscreen tag at 5–8 m, which covers a boom lane without active-tag cost or maintenance. Active earns its place where range genuinely cannot be engineered at the reader side — open-road tolling at highway speed, wide-area yard location — and if a quote proposes active tags for an ordinary gate or door, ask what the battery programme costs in year four.
Form factors and where each one fits
The same chips ship in different bodies, and the body is chosen by the scenario:
- ISO cards — the printable wallet-sized standard. Right wherever the credential doubles as visible ID: staff badges, student and patient cards, membership cards. Printable face, lanyard hole, dies in a back pocket wash.
- Key fobs — the same chip in a moulded teardrop for a keyring. Right for estates, gyms and residential access where nobody prints photos on credentials; harder-wearing than cards and harder to lend with a straight face.
- Wristbands — silicone (reusable: gyms, waterparks, hospitals, lockers) or Tyvek/paper (single-event: festivals, conferences). The scenario logic is hands-free convenience plus wet environments where a card fails; cashless payment at events runs on exactly these.
- Stickers and discs — NFC labels on equipment for tap-to-log inspection rounds, epoxy discs on returnable crates. The credential becomes the object's, not a person's.
- Windscreen tags and hard vehicle tags — UHF, read at metres: residential estates, office parks, weighbridges and automated yard gates. Tamper-evident windscreen versions void themselves if peeled off and moved to another car.
- Wearables and implants for industry — gloves-compatible wristbands and helmet tags where plant rules make pocket cards impractical; the same reasoning as our mining proximity wearables.
| Form factor | Printable ID face | Wet / rough use | Usual chip | Best-fit scenarios |
|---|---|---|---|---|
| ISO card (PVC) | Yes — the point of it | Poor | DESFire (new) / Classic (legacy) | Staff ID badges, access control cards, time-and-attendance cards, student & campus cards, patient ID, hotel key cards, membership & loyalty cards, canteen & cashless payment cards, library cards, visitor cards, photocopier / print-release cards |
| Clamshell card (thick) | Sticker only | Good | Legacy 125 kHz | Older estate & complex access, parking garage cards, staff parking — the thick rigid card old long-range prox readers were built around |
| Dual-frequency / combi card | Yes | Poor | 125 kHz + DESFire, or HF + UHF | Migration projects (old and new readers, one card), sites mixing door access with long-range parking or boom-gate reading |
| Key fob / keyring tag | No | Good | DESFire / legacy 125 kHz | Residential estate access, complex & apartment entry, gym membership, gate remotes' RFID companion, alarm arm/disarm tags, community scheme visitors |
| Silicone wristband | No | Excellent | DESFire / NFC | Gym & pool access, waterparks & resorts, locker keys, hospital & care-facility patient bands, spa & wellness cashless payment, school aftercare pickup |
| Tyvek / paper band | Print-on-issue | Single event | NFC / UHF | Festival & concert ticketing, conference & expo badges, day visitor passes, stadium access, race-event timing bands |
| NFC sticker / label / disc | Small label | Epoxy versions good | NFC (NTAG class) | Guard patrol & inspection checkpoints, equipment check-in/out, fire-extinguisher & safety-equipment service tags, returnable crates & kegs, product authentication, smart posters & table ordering |
| Windscreen / vehicle tag | No | Outdoor-rated | UHF EPC Gen2 | Estate & office-park vehicle access, boom gates & parking, weighbridge identification, truck & trailer yard automation, fuel-dispensing authorisation, e-toll-style lane reading |
| Rugged on-metal tag | No | Built for it | UHF EPC Gen2 | Tool tracking, gas cylinders & kegs, skips & bins, containers, racking & returnable steel assets, plant & machinery registers |
| Laundry / textile tag | No | Wash-rated | UHF (sewn-in) | Hotel & hospital linen tracking, uniform & workwear issue, laundry billing verification — see our linen tracking guide |
| Animal ear tag / bolus | Printed number | Built for it | LF (ISO 11784/85) / UHF | Cattle, sheep & goat identification, stud & auction records, theft recovery — see our livestock costing guide |
| Industrial wearable | No | Built for it | Per system (incl. proximity tags) | Mine & plant personnel — gloves-on access, helmet tags, lamp-room issue, underground presence & proximity detection |
Scenario examples that recur in South African operations: time and attendance (card or fob tap creating a payroll record — server-side verification matters more than card cryptography), cashless canteens and vending (DESFire balance or, better, a server-side ledger keyed to the card), hotel keycards (short-lived credentials, cheap cards, revocation at checkout), gym and locker access (wristbands beat cards in wet areas), contractor and visitor management (day-expiry fobs or printed cards), and checkpoint or inspection rounds (NFC stickers on the equipment, the tap proving presence — a scenario that feeds directly into the telemetry layer we build).
How to read your own site
Three quick diagnostics. Look at the frequency: if the card works near old long-read "prox" readers and has no visible chip contact plate, it is likely 125 kHz; a phone with NFC will detect 13.56 MHz cards and often name the type. Ask what the card protects: identifiers are fine for convenience functions, but every security-relevant door on a cloneable card type is carrying risk someone accepted without writing it down. Check the coupling to billing or payroll: where a tap creates money (canteen, time and attendance, fuel), server-side verification matters more than card cryptography — a cloneable card plus a good backend beats a secure card plus blind trust in the tap.
Migrations, for the record, are rarely rip-and-replace: multi-technology readers accept old and new cards simultaneously, so an estate moves over one reissue cycle at ordinary badge-printing cost. The mistake is not running old cards — it is buying new systems on old card types because the quote was marginally cheaper.
Where we sit in this: cards and readers enter our scope where they meet telemetry and identification systems — vehicle tags at automated gates, checkpoint cards in inspection rounds, credential taps that feed time or billing records we instrument. This article exists because choosing card technology badly is expensive and the good information is scattered; bring the access-control specifics to your security integrator, and bring the identification and data questions to us.