5 ms·

Why does USB-C have such terrible failure modes? I'm sure that the people behind it are bright engineers, so why do I have a fear unlike anything I've ever felt

Why does USB-C have such terrible failure modes? I'm sure that the people behind it are bright engineers, so why do I have a fear unlike anything I've ever felt when buying a USB-C cable or charger or device? Is this just a "tragedy of the commons" where every manufacturer expects the others to follow the spec so that they can skimp, or is there some kind of fundamental flaw in the USB-C spec that is making it so seemingly dangerous and difficult to use correctly?

_pmf_ 9y ago
It's almost as if hard USB A and USB B ends where there for a reason. I've heard good things about the Cypress EZ-PD[0] to handle power delivery issues (you'll need it in addition to the USB core). [0] http://www.cypress.com/products/usb-type-c-and-power-delivery http://www.cypress.com/products/usb-type-c-and-power-deliver...
fjsolwmv 9y ago
Because USB-C has a million features that require complex logical controls and hardware.
delecti 9y ago
Prior to USB-PD, USB maxed out at 4.5W, and that was a later addition, originally it maxed out at 2.5W. The USB-PW spec allows for 100W power transfer. There are far more modes in USB-C/PW, and they're much higher energy. If all you ever need to deal with is 5V and 0.1-0.9A then there's a much narrower range of failure states.
- spookthesunset 9y ago
  What voltage is the 100W coming through at? If it is low voltage, that is gonna be a really beefy cable... At 12V that is more than 8 amps. 5v is 20 amps. That is a lot of current to send over a typical USB cable...
  - AstralStorm 9y ago
    Yes The cable can get warm. That said, now we use two tiny power wires from PCIe spec to deliver north of 12V 200W.
  - CNJ7654 9y ago
    Either 15 or 20V I think
  - bri3d 9y ago
    Profile 5 is 20V@5A
    - samhain 9y ago
      They are getting rid of explicit profiles btw...
    - gargravarr 9y ago
      Suffice to say, this makes me very hesitant to connect a USB-C phone (which I don't have, thankfully) to a laptop USB-C charger, because even though they /should/ negotiate the correct voltage, the idea of a bug in the implementation causing the charger to fry my phone, perhaps even setting it on fire in the process, isn't worth the risk. As others note, the poor implementations wind up causing us to use only the adapters shipped with the device in the first place, which is actually a worse situation than with standard USB - right now, I know that I can plug a device into any USB charger and get 5 volts. Amperage may vary considerably, but unless the adapter is a real lemon (and yes, I know those exist), it's not going to kill my phone or burn my house down if I don't get the planetary alignment of adapter, cable and device right.
  - Dylan16807 9y ago
    The limit for a normal cable is 3A, the limit for a specialized charging cable is 5A.
  - bdonlan 9y ago
    In order to reach 100W, you need to use a special cable that is capable of actively participating in the negotiation process in order to advertise its ability to handle 5A@20V.
formula1 9y ago
This seems to a problem with the spec then. Its about the customer. Amoung the customers are - the consumer - who buys the product - the product - who buys from the manufacturer - the manufacturer If the manufacturer wants to only implement a subsect of the full spec, they should be allowed to do that. This has additional be benefits with regard to minimizing the resources used. Granted, its a cool spec with great potential and i enjoy seeing it in action with regard to sbcs. But the manufacturer is also a customer. If it doesn't work for them then there probably needs to be some changes
- digi_owl 9y ago
  What people usually refer to as USB-C are 3 independent specs. First there is the C plug and cable, including provisions for converting between C and the various A and B sizes. This is the spec that introduce various resistors to signal if the cable is a converter or (a very big or that has created much problems) if it can handle various watts at 5V. then there is the power delivery spec that on paper can be used with any USB plug (yes, even your old A and B formats), and allows current to go in either direction at up to 20V. And thirdly there is a continuation of the 3.0 data spec, 3.1, that include a provision for using various wires in the C cable in an alternate mode. This mode allows anything from digital video to PCI bus traffic to travel over the same cable, if both ends support the protocol traveling over the alternate mode wires. outside of alternate mode the 3.1 data spec can also be used with 3.0 A and B ports. So even if your device have a C port, it may not be able to handle more than 5V at 0.5A and 1.0 data speeds...
ksk 9y ago
It depends on what you consider normal I suppose. Its the 'devil you know' .. or maybe 'glass houses'. HTML, JS, CSS, etc, are all so buggy to where we have not managed to create a reference implementation for any of them (AFAIK). Imagine 100 browsers implementing the HTML/JS/CSS spec. Yeah..

_pmf_ 9y ago

It's almost as if hard USB A and USB B ends where there for a reason. I've heard good things about the Cypress EZ-PD[0] to handle power delivery issues (you'll need it in addition to the USB core). [0] http://www.cypress.com/products/usb-type-c-and-power-delivery http://www.cypress.com/products/usb-type-c-and-power-deliver...

fjsolwmv 9y ago

Because USB-C has a million features that require complex logical controls and hardware.

delecti 9y ago

Prior to USB-PD, USB maxed out at 4.5W, and that was a later addition, originally it maxed out at 2.5W. The USB-PW spec allows for 100W power transfer. There are far more modes in USB-C/PW, and they're much higher energy. If all you ever need to deal with is 5V and 0.1-0.9A then there's a much narrower range of failure states.

spookthesunset 9y ago
What voltage is the 100W coming through at? If it is low voltage, that is gonna be a really beefy cable... At 12V that is more than 8 amps. 5v is 20 amps. That is a lot of current to send over a typical USB cable...
- AstralStorm 9y ago
  Yes The cable can get warm. That said, now we use two tiny power wires from PCIe spec to deliver north of 12V 200W.
- CNJ7654 9y ago
  Either 15 or 20V I think
- bri3d 9y ago
  Profile 5 is 20V@5A
  - samhain 9y ago
    They are getting rid of explicit profiles btw...
  - gargravarr 9y ago
    Suffice to say, this makes me very hesitant to connect a USB-C phone (which I don't have, thankfully) to a laptop USB-C charger, because even though they /should/ negotiate the correct voltage, the idea of a bug in the implementation causing the charger to fry my phone, perhaps even setting it on fire in the process, isn't worth the risk. As others note, the poor implementations wind up causing us to use only the adapters shipped with the device in the first place, which is actually a worse situation than with standard USB - right now, I know that I can plug a device into any USB charger and get 5 volts. Amperage may vary considerably, but unless the adapter is a real lemon (and yes, I know those exist), it's not going to kill my phone or burn my house down if I don't get the planetary alignment of adapter, cable and device right.
- Dylan16807 9y ago
  The limit for a normal cable is 3A, the limit for a specialized charging cable is 5A.
- bdonlan 9y ago
  In order to reach 100W, you need to use a special cable that is capable of actively participating in the negotiation process in order to advertise its ability to handle 5A@20V.

formula1 9y ago

This seems to a problem with the spec then. Its about the customer. Amoung the customers are - the consumer - who buys the product - the product - who buys from the manufacturer - the manufacturer If the manufacturer wants to only implement a subsect of the full spec, they should be allowed to do that. This has additional be benefits with regard to minimizing the resources used. Granted, its a cool spec with great potential and i enjoy seeing it in action with regard to sbcs. But the manufacturer is also a customer. If it doesn't work for them then there probably needs to be some changes

digi_owl 9y ago
What people usually refer to as USB-C are 3 independent specs. First there is the C plug and cable, including provisions for converting between C and the various A and B sizes. This is the spec that introduce various resistors to signal if the cable is a converter or (a very big or that has created much problems) if it can handle various watts at 5V. then there is the power delivery spec that on paper can be used with any USB plug (yes, even your old A and B formats), and allows current to go in either direction at up to 20V. And thirdly there is a continuation of the 3.0 data spec, 3.1, that include a provision for using various wires in the C cable in an alternate mode. This mode allows anything from digital video to PCI bus traffic to travel over the same cable, if both ends support the protocol traveling over the alternate mode wires. outside of alternate mode the 3.1 data spec can also be used with 3.0 A and B ports. So even if your device have a C port, it may not be able to handle more than 5V at 0.5A and 1.0 data speeds...

ksk 9y ago

It depends on what you consider normal I suppose. Its the 'devil you know' .. or maybe 'glass houses'. HTML, JS, CSS, etc, are all so buggy to where we have not managed to create a reference implementation for any of them (AFAIK). Imagine 100 browsers implementing the HTML/JS/CSS spec. Yeah..