The most useful thing to understand about USB-C is that it is a connector, not a capability. Two cables that look identical, fit the same ports, and carry the same logo can differ by a factor of twenty in power and by a factor of twenty in data rate.
Nothing about the plug tells you which one you are holding, and the reason explains nearly every charging behaviour that appears inconsistent.
Twenty four pins and one that matters
The connector carries twenty four pins arranged so the two halves mirror each other, which is what makes it reversible. Power and ground are duplicated on both sides, there are four high speed differential pairs, and there are two configuration channel pins.
Almost everything interesting happens on those configuration pins. They establish which end supplies power and which consumes it, they reveal which way round the plug went in, they advertise how much current is available before any negotiation, and once a conversation is possible they carry it. A USB-C link with no configuration channel connectivity is an inert piece of copper.
The resistor conversation
Before any protocol runs, the two ends already know something about each other through resistance alone. A source presents a pull up resistor on the configuration pins, a sink presents a pull down, and the resulting voltage tells each side that something is attached and which role it is playing.
Because only one of the two pins ends up in that state, it also reveals the plug orientation, and the device routes its high speed lanes to the correct half.
The resistor value carries information
The value of the source pull up is itself information. Three values are defined, advertising the default USB current, 1.5 amperes, or 3 amperes. The sink reads the voltage, infers the limit, and may draw up to it. No packets are exchanged and no negotiation can fail, because there is nothing to negotiate.
This is the fallback layer, and it is why a simple charger and a simple device always do something sensible together. Throughout, the bus stays at 5 volts, which is deliberate. A device that understands nothing beyond the resistors is never exposed to a higher voltage.
Power delivery, a protocol on one wire
Above the resistors sits Power Delivery, a genuine packet protocol running over one configuration pin at a few hundred kilobits per second. The source advertises what it can provide as a list of discrete offers, each a voltage and a maximum current. The sink examines the list, selects exactly one, and requests it.
The source accepts or rejects, and if it accepts it transitions the bus voltage and then signals that the new supply is stable, at which point the sink begins drawing.
The menu of offers
| Voltage | Range | Ceiling at 5 A |
|---|---|---|
| 5 V | Standard | 25 W |
| 9 V | Standard | 45 W |
| 15 V | Standard | 75 W |
| 20 V | Standard | 100 W |
| 28 V | Extended | 140 W |
| 36 V | Extended | 180 W |
| 48 V | Extended | 240 W |
The standard fixed offers are 5, 9, 15, and 20 volts, with a ceiling of 100 watts at 20 volts and 5 amperes. The extended range adds 28, 36, and 48 volts and raises the ceiling to 240 watts, which is what lets one connector run equipment that previously needed a barrel jack.
The offers are discrete, and that detail matters. A sink cannot ask for 12 volts if the source does not offer it, and cannot ask for a voltage between two offers. It selects from a menu, and a mismatch between what a charger offers and what a device wants is one of the ordinary reasons a fast charger charges slowly.
The cable gets a vote
Current above 3 amperes places real demands on the conductors, and the connector cannot tell how good a cable is by looking at it. The specification resolves this by putting a chip inside the cable, which responds to a query from the source and reports what the cable is rated to carry along with its data capability.
The rule is strict. A source will not supply 5 amperes unless a cable has identified itself as capable of carrying it, and the extended power range imposes further requirements for the same reason. This is why a capable charger and a capable laptop settle on much lower power when joined by an unmarked cable, and why the cable is the component most worth buying carefully.
Data capability is separate again. A USB-C cable may contain only the pairs needed for the legacy data rate, which is fine for charging and useless for anything faster. Such a cable charges a laptop at full power and transfers files at a rate that suggests something is broken, and nothing about its appearance distinguishes it.
Programmable supply and where the heat goes
Fixed voltage offers force the receiving device to do the final conversion. A battery cell wants a rising voltage during constant current charging and a falling current afterwards, and if the input is pinned at 9 volts a regulator inside the device makes up the difference, dissipating it as heat inside a sealed handheld object next to the battery, at the moment the battery is least tolerant of heat.
The programmable supply mode lets the sink request a specific voltage in fine steps, adjusted continuously as charging proceeds, with a current limit it also sets. The charging profile is then executed by the wall adapter, which is larger, better ventilated, and not attached to the battery, so the same energy delivered produces considerably less heat where it matters. Most of what is marketed as advanced fast charging is this idea.
The same pins, carrying video
The four high speed pairs are not permanently assigned to USB traffic. Using the same negotiation mechanism, the two ends can agree to repurpose some or all of them for another protocol, most commonly a display interface. The tradeoff is physical. There are four pairs, so handing all four to video leaves nothing for high speed USB and the port falls back to the legacy data rate, while handing over two preserves USB at the cost of video bandwidth.
That is why one cable driving a high resolution display and a fast external drive at once is frequently disappointing. The connector is not hiding capability, it has run out of wires.
Why it sometimes does not work
Nearly every confusing outcome traces to one of a few causes. The cable is not electronically marked, so current caps at 3 amperes. The cable carries only legacy data pairs, so transfers are slow while charging is fine. The charger's list of offers lacks the voltage the device wants, so both fall back.
One end does not implement the programmable mode, so conversion loss stays inside the device and it charges slowly while running warm. Or the cable is out of specification, with wrong resistor values or missing configuration continuity, which is a hazard rather than an inconvenience, because a source misled about what is attached can apply a voltage the sink was never designed to see.
The specification is safe by construction as long as every element tells the truth about itself. The failure mode is a component that lies.
Note: the same connector spans from half a watt to 240 watts and from a legacy data rate to tens of gigabits per second. That range is the achievement, and it is also why no two cables can be assumed equivalent.