[AlexandrB]

> It seems like in the new USB-C world, I'd only have to connect one cable.

That's something I've never seen properly explained. Can you charge at 100W at the same time as using Thunderbolt 3 connectivity? It strikes me as unlikely since the power delivery and the signaling would have to share the same pins/wires.

Edit: I stand corrected after seeing that new LG monitor. Now I'm just curious how it works. Even PoE can't deliver 100w and Thunderbolt 3 uses much higher speeds while having to inject power.

[jsjohnst]

There's a couple reasons for PoE not supporting over 36W but USB-C supporting PD at up to 100W.

1) PoE only uses two pairs of wires for transmitting current, USB-C PD uses four pairs.

2) Ethernet cables are rated for much longer lengths, show me a 100m USB cable.

3) Ethernet is typically run in walls and more often than not is grouped into bundles, so heat dissipation is an issue.

4) PoE delivers power at ~50v, USB-C PD does 20v.

Also, piggybacking on the other commenter's reply, PoE voltage is carried on the same wires that are transmitting data (for 1000Base-T anyway)

[jessaustin]

There's no physical reason that power and signal can't share lines. Any particular device or protocol might fail to implement that, but PLC and even POTS show that it's possible and not even new.

[WildUtah]

The principle of power and signal separation has been used to send multiple signals over a wire since the duplex telegraph in 1872. It's one of the few things in this world older than the headphone jack standard that we've been using since 1878.

[jessaustin]

You may be surprised when you stick a multimeter into a standard POTS jack and read 48V. (Admittedly, the voltage is lower when the line is in use, but it's still there, which is why old-fashioned phones don't need separate power cords.)

[xenadu02]

The USB-C connector dedicates 4 pairs of pins (8 total) to power transmission (VBUS and GND). It only does that to give enough contact area and enough conductor circumference in the cable to support 100w transmission.

It also has two pairs of unshielded twisted pair for non-superspeed (USB 2.0) though only one pair is used in the cable apparently; which USB 2 channel you get depends on which way the cable is plugged in I guess.

It has a pair for sideband use. A pair for configuration and control.

The data primarily travels over the four high-speed differential signal pairs; two TX pairs and two RX pairs.

All USB 3 cables must have a chip in them. It communicates over the C&C channel to select how the high-speed pairs are used. When a display is connected, one TX/RX set are used for USB 3 and the other for DisplayPort.

USB 3 also supports "alternate" modes, so the C&C can negotiate that all high-speed pairs are used for Thunderbolt which is just the PCI-Express bus. Intel added the ability to interleave DisplayPort packets with the PCI-Express data. I assume that displays with USB 3 built-in are simply connecting a PCI-E USB hub to the Thunderbolt interface since in that mode there is no actual USB signal in the cable. USB 2 can use the dedicated signal pair for that and not interfere with DisplayPort or Thunderbolt traffic.

Thunderbolt is a nicer standard in some ways - interleaving means if you aren't using a 16bpp 5K display at max refresh rate you have more bandwidth available for other devices. With USB 3 you get half the available bandwidth if you connect a display. Thunderbolt is also old-school in the sense that it projects the CPU's bus to external peripherals... something all early computers used to do. Everything old is new again!

It is obvious USB-3 was very forward-looking. The C&C channel means some future version of the standard can drop USB-2 support and start connecting the unconnected pair if both ends negotiate for it. Now you have an extra TX/RX channel to give you +50% bandwidth. You can also imagine taking over the side channel pins. If you assume the next-gen standard can double USB-3's 10Gbps and add in double the data pairs that would be 40Gbps which matches Thunderbolt 3. Apply the same math to Thunderbolt and that's 160Gbps.