[Kelbit]

As an electrical engineer - I think incidents like these are a pretty solid argument for incorporating overvoltage protection into 5V USB peripherals.

Even something as simple as a zener clamp with a polyfuse to make a dead-simple crowbar circuit will save a device and won't contribute a whole lot to BOM cost.

[joshvm]

It's a bit odd, most older Macbooks have current limiting ICs on the USB ports. I've found this out when tinkering with devboards and accidentally shorting things. In fact most motherboards have some kind of protection for overcurrent conditions.

But USB-C isn't limited to 5V, power delivery is at 20V. That might explain what's going on here (since the user reports 20V on the output) - it thinks that the peripheral is a power hungry device and it's trying to charge it. That's a problem, but it could be that the peripheral is poorly designed and is mistakenly asking for power that it can't actually handle.

Edit: in this case the peripheral seems to be the Macbook charger... and plugging it in causes 20V on all the other outputs with only a dongle plugged in. Oops. Yeah not good.

I wonder what happens if you actually load the port? Perhaps it'll drop down to 5V? Or maybe it'll fry things.

That said, my comment above still applies: USB-C relies on both devices to be compliant with the spec. Otherwise you can get into situations where one device fries the other, or tries to charge things it shouldn't, etc.

[nameless912]

> most older Macbooks have current limiting ICs on the USB ports.

And thank goodness for this, early in engineering school I was doing a project on an arduino and because I was young and stupid I kept accidentally shorting power to ground. Killed at least 5 or 6 ATMega328's but the MacBook just helpfully chirped that I was drawing too much power and it was shutting off the port. Saved my ass at least a dozen times over.

[elsonrodriguez]

This feature has been on motherboards going back to the 90s. I remember that some DSL modem was notorious for requiring a reboot of the computer because the USB ports stopped working.

The reality was that the modem was drawing too much power and the motherboard just disabled every port on the bus.

[nielsole]

I accidentally shortened my USB port on a PC in a long chain of USB peripherals and only realized when I almost burned my finger on the most remote (wrongly connected) peripheral. I am really happy my mainboard didn't throw in the towel then.

[Kelbit]

> But USB-C isn't limited to 5V, power delivery is at 20V. That might explain what's going on here (since the user reports 20V on the output) - it thinks that the peripheral is a power hungry device and it's trying to charge it. That's a problem, but it could be that the peripheral is poorly designed and is mistakenly asking for power that it can't actually handle.

This was my interpretation, the host sets the output voltage to 20V incorrectly (due to a software bug or a hardware failure). Putting a crowbar circuit defensively on the peripheral side would save the peripheral.

[joshvm]

Ah ok, I misunderstood - yeah it baffles me why manufacturers don't do this. I make it a point to fuse absolutely everything at work, it's a minor expense to save a ton of time later.

Last week I blew up a $100 CO2 sensor because I wired it in backwards. Entirely my fault for not checking the pinout, but a 10 cent diode or a fuse would have saved me.

This is the same issue with telescope controllers. Plenty are rated for 6V absolute maximum, whereas typical power supplies in astro, e.g. the power tanks, or LiPos are 9V or 12V. Boom. Another $100 when a zener would have saved the day.

[zaarn]

A zener costs atleast 2 cents a piece when ordered in bulk. That's 2 ENTIRE CENTS LESS PROFIT!1!

[jowsie]

You laugh, but that could be a significant part of the profit for some Chinese peripherals.

[hinkley]

Hmm. Ever since I got the new macbook, my USB devices (especially my keyboard with built-in hub) often don't register properly when plugged into the Anker usb-c hub. I wonder if an overvoltage is happening and the keyboard has a polyfuse in it...

[koala_man]

If you were a business major, you would instead put that circuitry in a USB surge protector and sell them for $4.99

[mark212]

this is Apple after all: you're missing a digit -- $34.99

[hyperdimension]

Apple doesn't do cents!

It would be 49$.

[ansible]

> I think incidents like these are a pretty solid argument for incorporating overvoltage protection into 5V USB peripherals.

That isn't going to fly for product companies in general. The issue? Cost, as usual.

Adding 5 cents of cost to a $15 USD (retail price) mouse is a non-starter, because the actual production cost is closer to $3. So even 5 cents is a lot.

[_bxg1]

Between this and the Nintendo Switch fiasco, I'm worried about the future of USB-C. It sounds like trying to combine primary power supply and all data transfer into a single port is more dangerous than was anticipated.

[crysin]

Nintendo Switch isn't using USB-C, officially. They're implementation is not spec compliant unfortunately so the blame should be 100% on Nintendo in this case for using a universal port with proprietary implementation.

[stordoff]

FWIW, both Nintendo JP and UK refer to it as a USB Type-C connector

https://www.nintendo.co.uk/Nintendo-Switch/Specifications/Sp...

https://www.nintendo.co.jp/hardware/switch/specs/

[e_proxus]

There was a submission earlier on HN about how the Switch is not USB-C compliant: https://news.ycombinator.com/item?id=16706803

[Gaelan]

I wonder whether it would (would have been?) possible to trademark the USB-C shape to enforce compatibility.

[coldtea]

Or, you know, in the future, vendors will be more demanding from USB-C components and parts.

[_bxg1]

One would hope, but these aren't shady Chinese knockoffs having these problems, these are Nintendo and Apple products. Part of the problem, though, is that the worst-case scenario is so much worse than before. If your regular USB accessory is shoddy, it just... doesn't work. But since USB-C has the capability to transfer tons of power, having something go wrong can have much bigger consequences. Also, as mentioned in another comment, you can't have things like a hardware fail-safe to prevent a certain amount of charge from going through a port, because they're intentionally left under the control of software, because sometimes maybe you'd want that much power going through there.

[wyldfire]

Doesn't the trade group that owns the USB trademark require certification or conformance testing?

[ksec]

Nope. Not officially, one reason why I am against iPhone dropping lighting for USB-C. That was few years ago when no one see or believe this mess is coming.

[_bxg1]

I wonder if that's something they can start requiring after-the-fact

[squarefoot]

"I think incidents like these are a pretty solid argument for incorporating overvoltage protection into 5V USB peripherals."

Agreed, though for some peripherals effective overvoltage protection that goes up to the maximum port output could cost more than the device itself. I would rather get the chance and redesign the connector so that there won't be any risks of exposing old devices to higher voltages, then introduce it slowly to new laptops (say one port this year, two the next two years, etc) and possibly provide small external adapters that fit on the new connector on one side and export the old one (maybe more if working as hubs) with adapted voltages for both signals and supply lines. To me having multiple voltages driven by software on the same connector where one could insert devices that can't protect themselves in case of failure is asking for troubles. One day someone will connect a cheap LiPo charger to one of these ports and if the port firmware flips nasty things can happen.

[mirimir]

Yes! For example, Raspberry Pi have a resettable fuse on the USB power port.

[21]

But where do you stop? If a device supplies 20V instead of 5V, surely it might as well supply 230V.

[wlesieutre]

That would require a catastrophic failure in the power brick which has significant separation designed in to the board layout to keep the primary line voltage side away from the secondary low voltage.

See this comparison teardown between a genuine Apple magsafe charger and a counterfeit, especially the "What's wrong with this charger" section: http://www.righto.com/2016/03/counterfeit-macbook-charger-te...

Related, Apple ran a discounted trade-in program where people could swap potentially dangerous phone chargers for safe ones after a counterfeit charger killed somebody: http://www.cbc.ca/news/technology/apple-replacing-fake-iphon...

[hinkley]

With DC current you want to stay below 48 volts. There's a dog leg in how much damage you can do to a human being with electricity.

[zaarn]

Well, from experience you'll be mostly fine with 100 VDC too. 100VAC is a different story though. (AC flows better into the body while DC remains on the skin IIRC)

[mark-r]

There are legal limits, I don't think you can supply over 48V without entering a whole different regulatory structure.