Hacker News new | ask | show | jobs
by rektide 1108 days ago
The usb-c connector is rated to 5 Amps (A). Usb-c went up to 20V already. 5A * 20V = 100W (where we are).

We didn't want to change the connector. So we are stuck at 5A. But we can raise voltage & still have some margin. So voltage was raised to 48V. 5A * 48V = 240W.

As for raising the voltage more, it's likely very technically feasible, but probably not, for safety reasons. There are various levels of what is considered safe voltages. Higher voltages have higher ability to injure humans, to cause shock or potentially spark. Voltage is a unit of electrical potential, essentially how bad some juice is desperate to get out. 50V is the "Safe Extra Low Voltage" (SELV) limit for AC. It's actually 120V for DC but 48V/50V has a certain dislodgable-for-now mindshare. https://en.wikipedia.org/wiki/Extra-low_voltage

Connectors also have limits to their voltage isolation. If you have 1000V that electric potential really wants to get out & will arc across gaps. If there's debris or a worn connector, the threat of say 200V causing some short is much higher.

We probably could go higher. Usb-pd is super safe. When plugged in, it's in a very limited 5V state. This is a high risk moment in the connection life cycle, as pins are sliding into place, but partially connected & likely not aligned fully yet; not having any real voltage applied at this phase is a colossal design win against so so so many power delivery mechanisms in the world. Device & source _after being connected_ then have to negotiate power, on the secured connection. They ongoingly perform bounds checking to make sure the power is flowing in an expected way & will disconnect if out of spec. The connector has good isolation. 120v * 5a would be 600W and is still SELV. Add some thermal protection to the ports to make sure everything's OK there and it's probably going to just work. But any mishaps (ex: puncturing the cable, debris inside the device causing a short) would be more severe/scary. I'd love for some hackers to find out how high our connectors can go. Maybe in lab conditions we can get a usb-c connector doing a couple hundred volts fine. I would be unsurprised. Being safe for the real world though, after wear, and what happens when there is a fault is why we have SELV. But again, SELV says we could go up to 600W on this connector maybe.

One non-concern should be any additional heat, such as on cables & connectors. Since amps are still only 5A, the power loss is actually the same. Pretty great.

2 comments

> When plugged in, it's in a very limited 5V state.

Depending on what's on the other end, it's actually even better than that would imply most of the time.

If both sides of the cable are in spec-compliant USB-C ports, then the power pins will be left at 0v for the initial connection, and only the CC pins will have a minuscule amount of power coming from the source (charger) - just enough to detect specific levels of resistance on the sink (phone/laptop/etc.) for 5v power and/or negotiate a higher voltage for USB PD.

Once they've established that one side is a source and the other side is a sink, then the source will provide power on the power pins of the cable.

If a USB-C cable is used to connect one charger to another, for example, then the negotiation will fail and neither charger will provide power on the main power pins.

USB-A ports, on the other hand, always provide 5v, so USB-A to USB-C cables always provide 5v on the main power pins.

The reason some cheep chineese electronics with USB-C connectors can only charge with USB-A to USB-C cables, not with USB-C to USB-C cables, is that they skipped the CC pins entirely. I actually modded one device by adding in two half-cent resistors to make it able to charge from a USB-C port: https://www.nfriedly.com/techblog/2021-10-10-v90-usb-c/

> The reason some cheep chineese electronics with USB-C connectors can only charge with USB-A to USB-C cables, not with USB-C to USB-C cables is that they skipped the CC pins entirely.

If anyone has a usb-c pass through adapter that can negotiate for 5v out, please please please share. Ideally for me it'd be a like 3 inch cable that that a male & female connector.

Extra credit if it uses usb-pd to try to ask for 5v 5a.

(these would be incredibly out of spec & could cause damage to systems, but mercy they'd be useful! I find this problem to happen on a bunch of devices, alas not just cheap/rare ones)

You can plug an A male to C male cable into an A female to C male cable, that gets you a (directional) C-C cable with guaranteed 5V output.
I'm so close to having zero usb-a in my everyday carry. But this is still a good suggestion & I often can do this already.
Negotiating voltage after being connected avoids the problem of sparking while the connector is being plugged in, but I don't think it helps with avoiding sparking in the process of unplugging the connector? Unless one of the leads is shorter so that is disconnects first and the circuitry can detect that and turn off the power super fast?
You are absolutely correct, and the USB-C specification even has an entire appendix dedicated to the problem. The trick is to reduce the voltage difference until the plug is far enough that a spark is no longer an issue, which means placing an appropriately large capacitor at the Sink side.