One USB To Rule Them All: USB Type-C Connector Specification Finalized

Status
Not open for further replies.

pbrigido

Distinguished
Mar 6, 2008
529
0
18,980
I'm most excited about the ability to plug it in, in any orientation. If there are 2 ways to plug in a USB cable, I always find the opposite one first.
 

dstarr3

Distinguished
I'm most excited about the ability to plug it in, in any orientation. If there are 2 ways to plug in a USB cable, I always find the opposite one first.

Usually takes me three tries before I get it right. Don't know how, but that's how it goes for me.
 

SirKnobsworth

Reputable
Mar 31, 2014
43
0
4,530
"this spec is designed to replace them all"
Wasn't this the intent of USB when it was first rolled out?

Yes, but they might actually succeed this time. USB Type C is now small enough for mobile devices and they've eliminated the need for separate host/device connectors. It's also got a lot of room for expansion - there are 6 unused pins beyond the current USB signaling specs. The USB signaling pins are even reconfigurable in specific use cases. This could allow things like running PCIe or display signals over the same cable without having to go through Thunderbolt, or simply speeding up USB at some point in the future by adding another superspeed link.
 

nthreem

Honorable
Mar 23, 2012
28
0
10,540


Heat generation through cabling and devices is mainly due to resistance. Resistance of wiring will decrease proportionally with increase of cable thickness. Higher power demand will mean additional current through the wire.

You're right to think that if they increased power delivery to 100W without beefing up the cable, it'd get pretty hot. Imagine having a USB cable 10 times the thickness of your existing cable designed for 10W @ 2Amps. That'd be ridiculous.


Fortunately, the new USB 3.1 spec increases operating voltage to 20V when delivering power up to 100W. There can still be 5V powered devices, but you'd be limited to only 10W, which most tablets, chargers, USB cables, are designed to handle already.

So you're talking about an increase to 5A, which is much more manageable from a cabling standpoint than 20A if they had maintained 5V operating voltage.
 

hannibal

Distinguished
The ability to plug this without checking orientation is good, but the durability is something that may be not so stellar. I almost hope that there would be USB C and USB mini connectors...
 

Osmin

Distinguished
Jan 19, 2012
132
0
18,680
I would have been more excited if it included an optional fiber cable down the center to be able to pass Thunderbolt 3+ speeds in the future. This would eliminate the need for other special cables and even replace the display port cables. One cable to truly rule them all. I am looking forward to ditch all the different USB cables for just one standard.
 

none12345

Distinguished
Apr 27, 2013
431
2
18,785
While a reversable connector is good....its still dumb. They should have taken the opportunity to do a surface contact connector with a magenetic attachment.

While ive never broken a usb connector(standard, micro, or otherwise), the first thing i think of when i look at this thing is its going to break off.

They should have taken the opportunity to just solve the issue for good. But guess there will be a type d cable at some point to correct that mistake.
 

ozicom

Distinguished
Jun 21, 2012
51
0
18,640
it's not one usb, it's again two usbs. of course it'll be better from recent one but it's not still one :) Why didn't they think this at first? Because they want to make us spend lots of money to recent ones. Ok where do i have to pay for these new ones?
 

InvalidError

Titan
Moderator

I cannot find the pinout anywhere yet but I would be surprised if there were that many unused pins: Type-C carries all the same signals as USB 3.0 Type-A and to carry its 5A high-power spec with its tiny pins, it likely has a bunch of extra power/ground pins.

Looking at the plug model, there are four "long finger" type connections and those are always either power or ground so it looks like they at least doubled the number of power/ground pins.

That leaves only one pin unknown since the back pins are cross-wired to the front pins.
 

SirKnobsworth

Reputable
Mar 31, 2014
43
0
4,530
I cannot find the pinout anywhere yet but I would be surprised if there were that many unused pins: Type-C carries all the same signals as USB 3.0 Type-A and to carry its 5A high-power spec with its tiny pins, it likely has a bunch of extra power/ground pins.

Looking at the plug model, there are four "long finger" type connections and those are always either power or ground so it looks like they at least doubled the number of power/ground pins.

That leaves only one pin unknown since the back pins are cross-wired to the front pins.

The pinout is available in the official specs: http://www.usb.org/developers/docs/usb_31_081114.zip (page 18-19 of the USB type C specification).

You're right that there are only 12 pins per side, so with only basic reversibility that wouldn't really allow room for expansion. The trick (and I missed this too at first) is that if the other end is another USB type C port (as opposed to a passive legacy adapter), then it can auto-detect which side is which rather than just mirroring the pins. It's not all the pins of course - I count 16 unique pins. You're correct that the longer pins are the power delivery pins which connect before the rest.
 

vern72

Distinguished
Jul 15, 2012
306
47
18,810
"this spec is designed to replace them all"
Wasn't this the intent of USB when it was first rolled out?
Yes but the initially designed USB to be half-duplex, meaning that only one side can send data at any given time. With USB 3.0, they made it full-duplex so both sides can communicate at the same time. This is what they should have done in the first place so that we wouldn't need to redesign the connectors now.
 

InvalidError

Titan
Moderator

You need a "fully featured cable" for that, along with devices at both ends that support it. Normal Type-C cables will just be mirrored at both ends to avoid the six or so extra wires in the cable.

Ironic how USB used to tout how its cables were so much simpler and cheaper than FireWire's six but now, fully-featured USB 3.1 cables are going to have about 16 wires.
 

SirKnobsworth

Reputable
Mar 31, 2014
43
0
4,530
 


That's because USB's requirements have skyrocketed since it's inception, it's a case of feature creep. Originally USB was designed to replace the typical Serial, Parallel, and / PS/2 ports that all computers had. Instead of having this myriad of devices each using a different connection technology, USB would unify them into a single standard for peripherals. This worked pretty well, even the IOMEGA ZIP drives, which used to use parallel ports, worked fine with USB speeds. Then cheap flash media was discovered and the concept of storing large amounts of date on cheap external devices happened. Our bandwidth requirements skyrocketed as drive density got higher and thus we started storing larger and larger amounts of data. Enter USB 2.0 which barely kept up as cheap cameras and video devices started to come out (professional ones were / are still firewire). Then we got USB 3 which finally got us the speeds we needed but now we've got a ton of different types of cables due to the explosion of mobile and extremely small devices that also want to charge battery's over the USB port. Now they are releasing a cable that takes care of everything while also providing for fast charging and even higher bandwidth.

USB itself isn't very good at bulk data transfer due to lack of native DMA support for attached devices, only the host controller can directly access memory and thus bulk devices need to negotiate with the host controller for bulk transfers. Technology's like Firewire and eSATA allow the device itself to directly write to memory and thus those device's internal controllers can just do a fast copy from their internal cache's directly into the host systems memory as part of the data transfer without requiring a CPU interrupt. We've since hacked around USB's limitation by implementing bulk transfer modes into the USB host controllers, it just acts like a fast lane to get the controllers attention and transfer the bulk data in one continuous sequential stream of USB packets, rather then needing to schedule the transfer individually for each packet. Still requires CPU interrupts and isn't as efficient but it's better then what we had before.
 


FW is very similar to a SCSI bus where each FW device needs to have a full featured controller on it. FW is not point-to-point but rather is a chained bus where at one point you have the host controller, and each attached device then communicates with the device in front of and behind it. So you get a FWHC <-> DeviceA <-> DeviceB <-> DeviceC (terminates bus). Each device can write directly to memory but has to communicate these requests across the bus from one to the next until it gets to the controller. Basically FW is more like PCI / SCSI while USB is a much simpler tree protocol where each device just commits it's commands to the host controller. This comes at a performance and efficiency cost though as the USB HCI has to micromanage everything attached to it and act as a traffic cop to get access to the system itself.
 

InvalidError

Titan
Moderator

That is not what section 4.1 of the Type-C doc says...

"SuperSpeed USB serial data interface defines 1 differential transmit pair and 1 differential receive pair. On a USB Type-C receptacle, two sets of SuperSpeed USB signal pins are defined to enable plug flipping feature"

The baseline Type-C spec is designed to allow passive adapters with USB3 Type-A and Type-B. Type-C to USB3 Type-A/B only requires the same eight wires as USB3 Type-A requires: 5V, GND, D+/-.TX+/- and RX+/-.

The "Fully Featured" cable (the ones fully wired from end to end) is only defined as an optional feature/extension (section 5: extensions / alternate modes) for Type-C to Type-C cables.
 

SirKnobsworth

Reputable
Mar 31, 2014
43
0
4,530
That is not what section 4.1 of the Type-C doc says...

"SuperSpeed USB serial data interface defines 1 differential transmit pair and 1 differential receive pair. On a USB Type-C receptacle, two sets of SuperSpeed USB signal pins are defined to enable plug flipping feature"

The baseline Type-C spec is designed to allow passive adapters with USB3 Type-A and Type-B. Type-C to USB3 Type-A/B only requires the same eight wires as USB3 Type-A requires: 5V, GND, D+/-.TX+/- and RX+/-.

The "Fully Featured" cable (the ones fully wired from end to end) is only defined as an optional feature/extension (section 5: extensions / alternate modes) for Type-C to Type-C cables.

I don't think any of that contradicts what I said. There are two modes in which the receptacle can operate: one which mirrors the pins and only allows basic USB 3.1 transmission and the orientation detecting mode which enables future additional functionality. Obviously there's no need for orientation detecting on a passive adapter. My point is that you won't find Type-C to Type-C cables which support USB 3+ but don't enable the additional pins because they are explicitly disallowed by the specification (first section of section 3.4). I'm not sure if all devices with Type C ports are required to have the orientation detection, but if you're trying to use the additional pins then they will by necessity and if you're not using them you don't care anyway.
 

InvalidError

Titan
Moderator

Heh, look at that. That is one totally lame restriction since Type-C supports USB3 with all other 3.x connector types. Guess they could not resist the urge to screw at least one thing up and force people to use thicker and more expensive cables than necessary.

Well, if someone wants to make cheaper and thinner cables with only eight wires, they can still do so by hard-wiring one end as host and the other as endpoint to make each end think it is connected to a USB3 Type-A/B device. This does mean people would have to pay attention to which end they plug in which device as if they were physically different.

Another thing I find a bit annoying about Type-C is they decided to use extra wires to figure out what is connected where when they could have simply used auto-detection like Ethernet has been doing for over a decade with MDI/MDI-X.
 
Status
Not open for further replies.