What is load modulation, and why does a tag never transmit?
A passive tag has no transmitter. To answer, it switches a resistance across its own coil, which changes how much energy it absorbs from the field. The reader detects that as a small fluctuation in the current its own antenna draws. This technique is called load modulation, and it is why a tag can reply while consuming almost nothing.
The problem load modulation solves
A tag drawing microwatts from a borrowed field cannot run a transmitter. Generating a radio signal costs orders of magnitude more energy than the tag has available. Yet it has to answer, and the answer has to reach a reader several centimetres away.
The solution is to stop thinking of the tag as a sender.
How the reader hears an answer nobody sent
The reader and the tag are magnetically coupled. That coupling works in both directions: energy flowing into the tag is energy the reader has to supply, and the reader can measure how much it is supplying.
So the tag switches a resistance across its antenna, on and off, following the bits it wants to send. Each time the resistance is connected the tag absorbs more energy, and the reader sees the current in its own coil rise slightly. Each time it is disconnected, the current falls back. The reader reads the message off its own circuit.
Nothing crosses the gap except the reader's own field, modulated by a load at the far end. The tag is closer to a switch than to a radio.
Why the answer rides on a subcarrier
The variation the reader has to detect is tiny compared with the field it is producing. Looking for it directly would mean spotting a ripple on top of a signal thousands of times larger.
Instead the tag switches its load at a fixed rate, well below the carrier frequency, and encodes its data onto that rhythm. The reader can then filter for that specific rate and ignore everything else, which pulls the answer out of a very noisy neighbourhood.
When the answer is too faint
Load modulation has an asymmetry that causes real problems. Powering the tag and hearing it are two different thresholds, and they do not coincide.
A tag can receive enough energy to boot and still produce a load variation too weak for the reader to resolve. From the outside this looks identical to a tag that is out of range, but the cause is different and so is the fix. Enlarging the tag antenna helps both. Moving the phone a millimetre often helps only one.
This is also why a tag mounted on a lossy surface can behave strangely: the surface absorbs part of the field, and what it absorbs is indistinguishable, from the reader's point of view, from what the tag absorbs.
Why any of this matters in practice
Three things follow, and they explain a lot of otherwise baffling behaviour.
A tag cannot initiate anything. It has no way to signal a reader that is not already powering it, which is why a tag can never notify you, wake your phone from across a table, or be tracked while it sits in a drawer.
Two tags in the same field modulate the same reader's current at once, and the reader has to separate them by protocol rather than by physics. That is what anticollision is for, and it is why phones generally deal with one tag at a time.
And a tag placed against another tag will very often break both. Each one absorbs field the other needs, and each one's load variation muddies the other's. Stacking tags is not a way to get two actions from one tap.
Frequently asked questions
Does the tag emit any radio signal at all?
No. It only changes how much of the reader's field it absorbs. Everything the reader detects, it detects in its own antenna circuit. This is the reason a tag cannot be picked up by a scanner that is not itself powering it.
Why can a phone not read two tags at once?
Both tags modulate the same field simultaneously, so their answers overlap. The standards define an anticollision procedure to single one out, but phone software generally selects a single tag and ignores the rest rather than presenting a choice.
Can load modulation be intercepted from a distance?
Eavesdropping on the exchange is far harder than reading a tag, because the interesting part is a small variation on a field that is already very short range. It is not impossible with specialised equipment at close quarters, but it is not a practical everyday risk.
Is this how contactless payment works too?
The physical layer is the same. A card answers a terminal by load modulation exactly as a tag answers a phone. What differs is everything above it: the payment exchange runs a cryptographic protocol on top of that same electrical mechanism.