Okay, before we get into this, I want everyone to picture the object. It's a brick. It has ten ports on it. Ten. Four USB-A, six USB-C, three hundred watts total, and it lives on Daniel's desk doing the work of what used to be four separate chargers.
And one of those ports has met its match. A pocket speaker. Clips on your shirt. Charges fine from the old port, flashes a red light at him from the new one.
So Daniel sent us the puzzle. He's got this GaN block, and he did the thing most of us never do, which is actually read the printing on the underside of it. Four A ports, sixty watts each. Two C ports at a hundred watts, four C ports at thirty, and the whole C side rated from three point three volts up to twenty.
Which tells you something already.
It tells Daniel something, which is that every single port on this thing negotiates. Nothing here hands out a fixed voltage. And that's what's making him nervous. He says the fast charging standards make his head spin a bit, he's always a little uneasy plugging an older device into this thing, and he's heard overvoltage fries electronics.
Reasonable fear.
So he plugs the speaker into one of the C ports, C to C cable, nothing happens, except the charge light is doing a rapid red flash. He yanks the cable. Then he digs out an A to C cable, plugs into one of the A ports, and it just charges. No drama. And his read on it is that the A port saw no fast charging support on the other end, dropped to five volts, the speaker drew what it needed, everyone went home happy.
Which is half right.
And his actual question is the other half. Why didn't C to C do the same thing? What's different about how those two ports decide what to send, and did he hurt the speaker?
Two handshakes, Daniel. Same connector on one end, completely different etiquette on the other.
The thing to hold onto is that USB-A and USB-C aren't two versions of the same port. They're two different philosophies of who's in charge. A USB-A port is a dumb pipe. It is always the provider, it always starts at five volts, and it doesn't ask the device anything before it energizes the rail. It just hands out five volts, and if the device wants to negotiate up to nine or twelve, that's a proprietary trick on top of a connection that already works.
So the A port never refuses.
The A port essentially cannot refuse. There's no state in which it sits there holding power back. Power is on, that's the default, and the device takes what it takes. Which is why a cable that's A on one end and C on the other works with almost anything, because the A side has already made the decision, and the cable just presents a sensible face to the C device on the other end.
And the C port.
A C port with Power Delivery on it is the opposite. It starts from nothing. It will not put voltage on the rail until it has identified something on the other end as a legitimate sink. That's the whole design. Both sides have to introduce themselves before any current flows.
Introductions through what?
Through the Configuration Channel. When you look at a USB-C plug, there are two little pins in there that have nothing to do with power or data. They're called CC1 and CC2, and they carry the entire relationship between the two devices. That's the handshake that Daniel's C to C cable was trying to have and that his A to C cable skipped entirely.
So on the A side there's no handshake at all.
On the A side there is no handshake because there's no one to have it with. The A port has no CC pins. It can't ask. So the cable translates. An A to C cable presents the right resistor to the C device to satisfy it, and presents plain five volts to the A port, and the two ends never speak.
That's a strange sentence. The cable does the lying.
The cable does the translating. It's a diplomat. It tells the A port nothing and tells the C device everything it wants to hear.
So back to the C to C connection. Daniel plugs in the speaker, and the charger goes through this introduction.
And step one is physical, before a single digital message. The source, the charger, puts a pull-up resistor on those CC pins. The sink, the device, is expected to put a pull-down resistor to ground. When the charger sees that pull-down, it knows something is plugged in and it knows roughly what size that something is, because the resistance value tells it.
Tells it what?
The current tier. There are basically three default levels a device can advertise just with the resistor. Nine hundred milliamps, one and a half amps, or three amps. That's before any digital conversation at all. Just a resistor value saying, I'm a small device, I'm a medium one, I'm a big one.
That's the entire hardware cost of being a good USB-C citizen.
It is tiny. For a device that just wants to pull five volts at up to three amps and never wants anything fancier, the whole requirement is two resistors. Five point one kilo-ohms, one on each CC pin, to ground. That's it. No chip, no firmware, no power delivery controller. Two resistors that cost a fraction of a cent.
And that gets you fifteen watts.
That gets you fifteen watts, which is more than enough to charge a pocket speaker at a reasonable rate. Which is the part that should sting. The device that's failing here isn't failing because it's too old or too simple. It's failing because of two parts that cost less than the plastic in the connector housing.
So what happens when the charger looks at the CC pin and sees nothing.
Then as far as the charger is concerned, there is no device. The pin is sitting there with no pull-down on it. There's no advertisement of a current tier, there's no valid sink state, and the source does exactly what it's designed to do when it can't identify anything. It keeps the rail off.
No power at all.
That's the fail-safe, and it's deliberate. The whole system is built so that the failure mode of a failed handshake is darkness, not voltage.
So let's say the speaker is compliant, and it does present its resistors, and the handshake proceeds. What does the charger do then?
Then it's a conversation, in a specific order. The charger sends what's called Source Capabilities, which is a list of Power Data Objects. Think of it as a menu. My charger's menu, in Daniel's case, would list five volts at three amps, nine volts, twelve volts, fifteen, twenty volts, plus a couple of programmable entries. Each item on that menu is a pairing of a voltage and a maximum current.
And the device picks from the menu.
The device picks, it sends back a Request, the charger sends Accept, then a Ready message, and only after that sequence does the voltage on the rail actually move from five volts to whatever was agreed. The order matters enormously, because the rail is still at the safe default until the very last message. Nothing jumps.
How fast is this conversation?
Milliseconds. The sink has about thirty milliseconds to respond to a message before the source starts worrying. The whole negotiation, when it goes well, is effectively instant from a human point of view. You plug in, the light comes on, you never knew any of it happened.
And the voltages in between. Daniel's charger says three point three to twenty volts, not a list of fixed steps.
That's PPS, Programmable Power Supply. Instead of the device asking for nine volts, it can ask for nine point one two, or eight point nine four, and the charger will deliver it in twenty millivolt steps. That's what makes some of the fast charging schemes work, because they can track what the battery actually wants in real time and adjust. That's what that printed range means. It's not a random voltage generator. It's a menu with very fine print.
Okay, so I'm looking at Daniel's speaker, and I want to be fair to it. Is it possible the thing is simply lazy rather than broken?
I think lazy is the right word, and I'd go further. The most likely explanation isn't that the speaker is broken at all. It's that the speaker's USB-C input was never built to be a proper USB-C sink. It's a charging jack that happens to be the right shape.
Meaning the two resistors aren't there.
Meaning the two resistors very likely aren't there, or aren't there in the right place, or the port is wired to a simple charging circuit that predates the standard. And this is not rare. A manufacturer building a cheap speaker has a choice: add two resistors and make the port universally compatible, or save the fraction of a cent and assume the customer will use whatever cable is in the box.
Which manufactures the exact experience Daniel had.
Except Daniel didn't use the cable in the box. He used a proper C to C cable, which is the honest thing to do, and the honest cable exposed the problem. That's the irony. The better cable made the device quit.
So to answer his question directly. Why did A work and C to C not?
The A path never asked the speaker to identify itself. Five volts were waiting before the speaker even arrived on the rail, and the speaker just drew current the way it was built to. Nothing was negotiated because nothing needed to be. Plug in, charge.
And the C path asked, and got silence, and did the only safe thing it could do.
Sat there with the rail off. And here's what I'd say to the worry underneath Daniel's question. He thinks something went wrong. Something didn't go wrong. The charger declined to start. That's a completely different event.
Which is a relief, and also slightly insulting. His charger looked at his speaker and didn't recognize it.
His charger looked at his speaker and saw an open circuit on the identifying pins. As far as it knows, there's nothing there. It's not judging the speaker. It can't even see it.
Fine. So now the light. Because the light isn't nothing. If the port were simply dead, the speaker would sit there dark and Daniel would have shrugged and moved on. Instead it's flashing at him.
And that flash is the speaker's side of the story, which is why it's interesting. A charger that refuses to talk gives you silence. A device that detects a fault gives you a signal. The speaker knew something was wrong.
So decode it.
Rapid flashing on a charging indicator is almost universally an error code. Not a status, an error. On just about any device you own, a slow pulsing light means normal work in progress, usually charging a depleted battery. A fast, frantic, irregular flash means the charging logic tried to start the cycle, found something it didn't like, and shut the port down.
Slow means filling. Fast means fault.
And the fault we're most likely looking at is the handshake failure itself. The speaker's charging controller goes through its own startup routine, waits for the input to stabilize at a voltage it recognizes, and it doesn't arrive. Depending on how the circuit is built, it may see a rail that never comes up, or it may see power flicker on briefly before the charger cuts it. Either way the controller concludes the cycle failed to initiate, and it signals that the only way it can. Red, fast.
So the flashing isn't damage.
The flashing is the opposite of damage. The flashing is a device announcing that it refused to proceed. It's a fault indication, and specifically a fault indication tied to the start of charging, not a health report about anything internal. The speaker was telling Daniel, I couldn't begin.
He yanked the cable.
He yanked the cable, which did nothing good and nothing bad. There was no dangerous condition to interrupt. The charger hadn't energized the rail, the speaker had already refused, and the two of them were sitting in a standoff. Pulling the cable ended a standoff.
I want to be careful here, because I've done exactly what Daniel did. Something flashes at me, I unplug it, and I feel like I've prevented a fire.
The instinct is right even when the reasoning is wrong. Yanking a cable when something looks wrong is basically always correct behavior. In this case the thing that looked wrong was a device that was already protecting itself.
Let's take the safety question head-on, because it's the one he actually asked. Is a three hundred watt charger dangerous for an old speaker.
Here's the part I find elegant. A compliant Power Delivery source is safer for that speaker than a dumb five volt brick. Not equally safe. Safer.
Explain that, because it sounds backwards.
The thing the whole protocol is built to prevent is a source forcing high voltage onto an unprepared device. So the source cannot simply put twenty volts on the rail. It can't put nine volts on the rail. It cannot move off the default until a valid sink has requested a specific higher voltage and the request has been accepted in sequence. The negotiation isn't a courtesy layered on top of the power. The negotiation gates the power.
And the brick.
The brick has no gate. A cheap fixed supply that puts out five volts and a decent amount of current will hand that straight to whatever is plugged into it, and if the device's input circuitry is bad, the brick will find out the hard way and so will the device. The brick is only safe because five volts happens to be the right answer. It's not safe by design. It's safe by coincidence.
Whereas the smart charger, which is the one Daniel's nervous about, will verify before it energizes.
Which is why he should relax about the three hundred watt number. That number describes a ceiling, not a behavior. The charger doesn't have three hundred watts in its hand waiting to dump it into the first thing that shows up. It has a menu and it waits. Two devices that can't talk to each other produce a dead rail. That's the whole outcome.
It's the least satisfying possible failure. You spent money on a charger that does everything and the result is that nothing happens.
Nothing happening is the system working.
Alright. But I want the exception, because there always is one, and I'd rather hear it from you now than find it in a comment.
The exception is that the negotiation is trust-based. When the charger sends its menu, the device is expected to be honest about what it can accept, and there's no verification in the base protocol that the menu itself is legitimate. Version two and version three of Power Delivery have no mandatory cryptographic check on what the source advertises. If the charger's firmware is compromised or badly written, it can advertise a voltage the device was never built to take and then deliver it.
BadPower.
That's the documented class of attack. It requires either a malicious charger or a buggy one. Version three point one added optional authentication, so a device can in principle verify the source before trusting it, but the adoption of that in consumer chips is limited. Most of what's on your desk today doesn't do it.
The theoretical fear is real but it requires a compromised source.
It requires a source that's lying. Which takes us back to the same distinction. The danger is never the number printed on the charger. The danger is a source that says one thing and then does another. A compliant charger doing three hundred watts is not the risk. A faulty charger doing sixty watts and lying about it is.
Which is a useful way to think about it. Judge the charger by whether it keeps its word, not by the size of the number.
There's a ripple effect here too, which is that the problem Daniel hit is going to get more common, not less. The A port is being retired. Once a charger is USB-C only, there's no bypass cable left. If your device can't say hello, there's nowhere to hide it.
The A-to-C cable is a dying escape hatch.
It's a compatibility crutch that only exists while A ports exist. When everything on your desk is C, a non-compliant device just stops charging, and the user has no idea why, because the cable fits and the charger works and nothing is broken. That's the trap. It looks exactly like a dead device.
What do you do, practically. Because I know the instinct is going to be, test the device with every cable in the drawer.
Test with A-to-C first if the device is old or cheap, because that tells you whether the device charges at all. If it charges from A and not from C, you've learned something real about the device, and it's the missing resistors. And then decide whether you care. If the speaker lives on your shirt and charges fine from the A port overnight, it doesn't matter.
If everything you own is C.
Then a short A-to-C adapter cable in the bag is the least elegant solution that works, and I'd rather have one than argue with a plastic port.
What about the actual question of damage. Can I say confidently that his speaker is fine.
You can say the mechanism suggests it, and that's not the same as a documented finding for this exact speaker. The charger withheld power, the speaker refused to start, so there's no obvious path by which energy reached the wrong place. But I can't inspect his circuit board. I'd say this: if the speaker charges normally from the A port now, at its normal rate, and the battery still holds what it used to, there's no story. If it had taken damage, he'd know by now.
Mark it as inference.
Mark it as inference, and a fairly confident one, because the design of the failure is friendly. Nothing in the sequence involves a high voltage being applied to a low voltage device. The whole fight was about whether to turn on at all.
Which is the opposite of what he was afraid of.
He was afraid the charger would overreach. What actually happened is the charger wouldn't reach at all.
One more thing before Hilbert weighs in. Daniel says the A port on this charger is also rated from five to twenty volts, sixty watts max. So the A port is doing some negotiation of its own.
It is, but only in the ways A ports with proprietary fast charging do. That's not Power Delivery. That's a vendor protocol layered on top, and it only engages if the device on the other end speaks it. If the device doesn't, the port sits at five volts and behaves like the dumb pipe we started with. That's why it worked for the speaker. The speaker doesn't speak it, so the port gave up and just handed over five volts.
The port tried to be clever and then gave up and was kind.
The A protocol fails open. The C protocol fails closed.
