A microchip cat flap runs an access control system in your kitchen door, and the credential it checks is a glass capsule the size of a rice grain that was never designed to be a key. The chip in your cat’s shoulder exists so a vet or shelter can identify a lost animal and phone the owner. Somewhere along the way it got promoted, and almost nobody noticed.
The arc from a hole in a board to a camera judging what is in a cat’s mouth covers six hundred years, three authentication schemes, and one Amazon product manager with 23,000 photographs of his own cat.
Table of Contents
- Before the Flap: A Hole, a Board, and a Cat
- The Magnet Era: A Door With One Bit of Memory
- The Microchip Cat Flap: Turning a Vet ID Into a Door Key
- The Physics That Limits Every Microchip Cat Flap
- Flo: The Cat Door That Judged What Was in the Mouth
- Identification Is Not Authentication
- Frequently Asked Questions
- The Quiet Part
Before the Flap: A Hole, a Board, and a Cat
The earliest documented cat hole in English is not an invention story. It is a plot device. In Chaucer’s Miller’s Tale, written in the late 1300s, a servant needs to spy on a locked room and finds his opening: “An hole he foond, ful lowe upon a bord / Ther as the cat was wont in for to crepe.” Chaucer never explains the hole to his audience, which is the interesting part. By 1390 a cat-sized gap at the bottom of a door needed no gloss.
British churches went further. Openings cut into church doors survive from the 16th century, though most were sized for the local dogs. The principle holds: a deliberate permanent aperture, because keeping animals out was never the point. Rodent control was, and rodent control walks on four legs.
What the Newton Story Actually Rests On
The legend says Isaac Newton cut two holes in his door at Trinity College, Cambridge, a large one for his cat and a small one for her kitten, because the animals kept spoiling his optics experiments by letting light into a darkened room. In the best version, the kitten ignores the small hole entirely.
The evidence is thin. No contemporary record says Newton kept a cat, and the tale does not reach print until long after his death in 1727, with David Brewster’s 1827 Memoirs of the Life, Writings, and Discoveries of Sir Isaac Newton among the early sources, roughly a century late. The mathematician John M. F. Wright hedged carefully: whether or not the account is true, he wrote, there are two plugged holes in the door to this day. Two plugged holes are evidence of two holes. They are not evidence of cats. Chaucer beat Newton by three centuries anyway, so what survived is the shape of the anecdote: a genius, a mundane problem, an obvious solution he supposedly missed the second time.
The Magnet Era: A Door With One Bit of Memory
The real engineering problem was never the hole. It was selectivity. A hole admits your cat, the neighbour’s cat, a fox, a draught, and occasionally a rat. A century of cat flap patents is one long argument about how to make a hole that discriminates.
The first workable answer put a token on the cat. British Patent 1567001 describes a magnetic responder connected to a latch, released when a bar magnet on the collar passes over it. Later filings are rude about how well that worked: the responder and latch had high inertia and did not always release in time, so the cat met a door that was technically unlocked and mechanically still stuck. Anyone who has watched a cat headbutt a flap twice and then sit down in protest knows the failure mode.
The fix was to stop asking a magnet to do mechanical work. GB-A-2223257 describes an electromagnetically controlled cat flap where a reed switch senses the collar magnet and a solenoid pulls the catch, turning the magnet into a sensor input. The Staywell No. 21 went further, with a passive inductive key element on the collar and a tuned circuit in the door. That is RFID in embryo: an unpowered resonant tag, a reader coil, a lock that opens on resonance.
All of these share one flaw. A magnet is not an identity. Any magnet of roughly the right strength opens the door, so on a street where three houses bought the same flap, all three cats hold the same key. The credential also lived outside the animal, which means it comes off eventually.
Infrared Keys and the Coded Collar
Infrared flaps upgraded the token from magnetic to optical. A detector above the opening watches for a small battery powered emitter on the collar, and because the emitter sends a code rather than a field, two cats on the same street stopped being the same key. A real gain in entropy, paid for with a battery on a cat’s neck.
The Microchip Cat Flap: Turning a Vet ID Into a Door Key
The idea that ends the collar problem is to move the credential inside the animal, and by the mid 2000s millions of cats already carried one. Nick Hill, a Cambridge physicist, had the ordinary version of the problem: neighbourhood cats coming through the flap to eat his cat Flipper’s food and intimidate him. After about three years of development, the first SureFlap launched in 2008 with a purpose built RFID front end the company called OnTune, tuned for better read range at a given battery budget. The pitch was elegant. The key is already in your cat, and someone else paid for it.
What the flap reads is an ISO standard transponder. ISO 11784 defines the structure of the identification code, ISO 11785 defines how it is transmitted, and the dominant implanted format is FDX-B at 134.2 kHz. The object is a glass capsule of roughly 1.4 by 8 mm, usually implanted between the shoulder blades, holding a coil and a small integrated circuit. No battery, no clock. It wakes only when a reader’s field induces a current in its coil, then modulates that field to spell out a 15 digit number.
Why 134.2 kHz and Not Something Faster
Animal identification sits at low frequency for a physical reason. A cat is mostly salt water, and salt water is hostile to the higher bands logistics tags use. At 134.2 kHz reader and tag couple inductively through a near field instead of communicating by radiated waves, so tissue, fur and moisture attenuate the link far less. The standard was written to be read through an animal, not across a warehouse. The bill comes due in distance: low frequency plus a tiny antenna inside the capsule puts a hard ceiling on range.
The Physics That Limits Every Microchip Cat Flap
A passive transponder harvests all of its energy from the reader’s field, and near field coupling falls off brutally with distance. Handheld veterinary scanners are typically specified around 8 to 15 cm for glass implants, and a technician holds those against the animal. A door has to hit a comparable budget on batteries, for months. Three consequences follow.
- The tunnel is an antenna housing. Every microchip flap makes the cat push its head and shoulders into a short tube before anything unlocks. The tube exists to put the implant within centimetres of the reader coil. Door shape follows field strength, not cats.
- Chip migration breaks the model. Implants sometimes travel from between the shoulder blades towards a leg or the flank. A vet with a handheld scanner sweeps the animal and finds it. A fixed coil in a door cannot sweep anything, so a migrated chip becomes an intermittent cat.
- Latency is a behavioural problem. Detect the field disturbance, read the code, compare it against a stored list, drive a solenoid. The delay is short but not zero, and cats notice. One bumped by a slow latch a few times starts approaching with visible suspicion, which is where a cat’s paw-first investigation habit collides with human engineering.
And a fourth that no antenna fixes: once the latch releases, the door is open to whatever follows the authorised cat through it.
Flo: The Cat Door That Judged What Was in the Mouth
Every system so far answers one question: who are you. In 2019 Ben Hamm, a product manager at Amazon, built a door that answered a harder one: what are you doing.
His cat, Metric, kept delivering prey through the flap. Hamm mounted an Amazon DeepLens camera above the door, wired an Arduino to a lock, and collected over 23,000 images of Metric coming and going to train a classifier on Amazon SageMaker. The system runs a cascade of three questions: is there a cat in frame, is it arriving or leaving, and if arriving, is it carrying something. The decision takes under two seconds.
On a positive detection the door locked for 15 minutes, sent Hamm a text, and donated money to the Audubon Society, making Flo the only known cat flap with a built in guilt tax. Hamm presented the project at Ignite Seattle in 2019. Across five weeks of live operation, Metric entered innocent 180 times and was locked out six times.
Those numbers are the whole story of vision-based flaps. Six events in five weeks means the interesting behaviour is rare, the dataset is permanently imbalanced, and every false positive lands on a cat that did nothing wrong. The machine is inferring intent from silhouette and posture, at night, through fur, on a subject that does not cooperate. Set against how cats themselves see the world, the camera looks slow.
Vision-Based Flaps Go Commercial
Five years later the hobby project had a price tag. Flappie, a Swiss startup founded by twin brothers Oliver and Denis Wilder, showed a prey-detecting cat door at CES 2024 in the Swisstech Pavilion at the Mandalay Bay Convention Center. It combines outward-facing motion sensors, a night vision camera, and a model trained on a proprietary dataset the company says was filmed across many cats, prey species, and lighting conditions. Flappie claims accuracy above 90 percent and also reads microchips, so the door can be restricted to specific animals. Announced pricing was 399 dollars outright, or 199 dollars with a two year commitment to an 8.90 dollar monthly subscription, launching in Switzerland and Germany.
Hold that 90 percent up to the light. A cat using a flap six times a day generates roughly 180 decisions a month, so a 10 percent error rate is 18 wrong answers, split between a cat locked out for holding a leaf and a cat waved through holding a vole. Fine for a consumer product, bad for anything you would call security. The behaviour is not a defect in the cat either. Prey delivery is instinctive and sits close to the chattering cats do at birds through glass, so the door has to be the thing that changes. With birdwatching booming across a whole generation, that market is not shrinking.
Identification Is Not Authentication
Here is the part the packaging skips. The pet microchip was specified for reunification: a vet or shelter scans a found animal, reads a number, looks it up in a registry, phones an owner. The threat model was a lost cat. There was no adversary in the room when the standard was written.
So FDX-B carries no cryptography. The transponder answers with its number to anyone who energises it, with no challenge and no rolling code. Security researcher Roald Nefs documented this in 2022, writing correctly formatted animal identification data onto blank and rewritable transponders for a kit cost of around 50 euros. Readers cannot distinguish a spoofed tag from an implanted one, because nothing in the protocol asks them to.
Anyone who has read about what a Flipper Zero actually does knows the shape of this: a lot of the world’s low frequency tags broadcast a static number and trust the environment. The modern answer, a credential that proves possession of a secret without transmitting it, is the reasoning behind end to end encryption key exchange and behind the migration to post-quantum cryptography. None of it applies to your cat, and none of it was ever meant to.
The honest framing is that a microchip cat flap is a convenience filter, not a security device. Nobody is cloning a transponder to burgle a house through a 15 cm hole. The realistic failures are mundane: a tailgating neighbour, a migrated chip, a battery that dies at 3 a.m., a stray who is cold and now has one fewer option. A door that knows only numbers cannot tell a hungry cat from a hostile one, and that is a design limit rather than a bug.
Frequently Asked Questions
Did Isaac Newton invent the cat flap?
No. Cat holes appear in Chaucer’s Miller’s Tale in the late 1300s, roughly three centuries before Newton, and church doors with animal openings survive from the 16th century. The Newton story does not appear in print until around a century after his death, and no contemporary evidence says he kept a cat.
Does a microchip cat flap work with every microchip?
Not automatically. ISO 11784/11785 FDX-B at 134.2 kHz is the international standard and the most widely implanted, but older and regional formats exist, including 125 kHz chips and protocols with different code lengths. Manufacturers publish compatibility lists, so match your cat’s actual chip number format against that list before buying.
Why does my cat have to push its head all the way in?
Because the implant has no battery. It draws power from the reader’s magnetic field, and that field weakens sharply with distance. Veterinary scanners are typically rated on the order of 8 to 15 cm against the animal, so a battery powered door needs the chip within a few centimetres of its coil. The tunnel shape exists to make that happen.
Do AI cat flaps actually stop cats bringing prey home?
Partially. Ben Hamm’s Flo locked his cat out six times across five weeks against 180 clean entries, and Flappie claims over 90 percent detection accuracy. Both numbers mean the system will sometimes lock out an innocent cat and sometimes admit a successful hunter, so treat it as a filter that reduces incidents rather than a guarantee.
The Quiet Part
Six hundred years of cat flap design is a compressed history of authentication. A hole trusts everyone. A magnet trusts anyone holding a magnet. An infrared fob adds a code that gets lost with the collar. A microchip moves the credential inside the animal and gives up cryptography to do it. A camera drops identity altogether and guesses at intent. Each step buys precision and pays in new failure modes, while the cat, who agreed to none of it, keeps testing the door with one paw first.
🐾 Visit the Pudgy Cat Shop for prints and cat-approved goodies, or find our illustrated books on Amazon.
Leave a Reply