Question Question about USB ports that never got answered in that article or comments

[Videogamer555]

I read over on https://www.tomshardware.com/news/usb-3-2-explained about all the different USB version names, and then there was an interesting comment in the comments section on that article that someone posted. I wanted to ask a followup question about that comment so I headed over to the associated forum page https://forums.tomshardware.com/thr...lained-whats-gen-1-gen-2-and-gen-2x2.3778787/ only to find that that page is locked from new posts. So I've started my own post here to ask the question.

Here's the question:
@Nikolay Mihaylov had commented that

All USB3+ connectors introduce separate data pins in addition to the USB2.0 data pins. A USB2.0 cable only makes contact with the legacy USB2.0 pins so everything works as if USB3 is not there at all.

Obviously, you want USB3+ for anything data intensive. Apart from the speed, it has the added advantage that it's full duplex, so data can flow simultaneously in both directions, like on Ethernet networks or PCIe.

BTW, 10Gbps USB3 introduces a new channel encoding which is more efficient. So while the 5Gbps variant translates to at most ~450MB/s real data (after accounting for channel encoding and protocol overhead), with 10Gbps you get ~1100MB/s. Consequently, Gen1x2 is slower (2x450 = 900MB/s) than Gen2x1 (1100MB/s). But it might work over longer distances.

I'd always assumed that the 5Gbps referred to the real data throughput rate (and it gets even higher when counting the overhead), as the real throughput is the only thing the average user cares about, so would be reported in the specs as such (nobody other than engineers cares about what the data rate is when also counting the overhead). Are you sure that for 5Gbps USB 3 the real data rate is actually only 450Mbps? 450Mbps is actually SLOWER than USB 2 (with a real data rate of 480Mbps).

Also you mentioned Gen1x2 in your comment. I've never heard of Gen1x2 before, and it wasn't even mentioned in the article you were commenting on. The only "x" mentioned anywhere in this article are Gen2x1 and Gen2x2. There's basically only 3 types of USB 3 ports: USB 3 Gen1, USB3 Gen2 (aka USB 3 Gen2x1), and USB 3 Gen2x2. I don't know where you got the idea of USB 3 Gen1x2 from. Can you please explain it?

And something else the article didn't really go into is the number of wires in the cable. USB 3.x cables have more wires in them than USB 2.x or 1.x cables. All USB cables before 3.x had only 4 wires. While older USB cables will work with older devices on USB 3 ports, USB 3 cables are needed to use USB 3 devices on USB 3 ports. This is because there's more than the standard 4 wires in a USB 3 device, cable, or port. But how many wires more? I know USB 3 Gen1 still had additional wires, but does USB 3 Gen2 have more wires than Gen1? Or does it use the same number of wires as Gen1 but the electronics just process data faster? How about Gen2x2? I know it has a different connector (USB-C) that has 2 sets of the same wires (to allow for reversing the cable), but if you ignore the duplicate wires, how many wires are used in Gen2x2? Is it the same as previous generations of USB 3? Or is it still more wires?

And just what exactly are the extra wires used for? Are they used to compliment the data on legacy wires (legacy wires used to send data in one direction, while new wires used to send data in the other direction) for full duplex? Or do USB 3 devices ignore the legacy wires, and do full duplex on the new wires only (are there enough new wires to actually do full duplex communications without also using the legacy wires)?

 

[Eximo]

Most of your answers are here:

https://en.wikipedia.org/wiki/USB

USB 3.2 Gen 1×2 is a thing, essentially USB is to blame for renaming USB so often.

Pins and wires would be equivalent from what you mentioned. I may be wrong, but I think all the pins in USB Type C are used (if implemented) in either orientation, they are just designed to be to used in either direction so that when you flip it, the same pin/wire is connected (they are mirrored essentially from top to bottom)

Power is a lot of the USB wiring today in addition to the extra RX/TX lines.

 

[Paperdoc]

To your first question on Gen1 speed, you and MANY others are confused by a small detail in the labels. USB2 speed is 480 Mb/s or 0.48 Gb/s. USB3.2 Gen1 is 5 Gb/s, or about 10 times faster. Since the actual data encoding means that one BYTE of data needs about 10 BITS of data transmitted, that can be re-specified as 0.5 GB/s. NOTICE that the Bits per second way is "Gb/s", and the BYTES per second is "GB/s". So 0.5 GB/s is still ten times faster than 0.5 Gb/s.

NOTE also right here that all these speed specs are the max data transmission rate possible by the data communication system involved, and the whole design philosophy of these systems is that the communications subsystem should always be FASTER than the actual data access speed of the device using it so the the communication subsystem is never limiting the device performance. For example, a "Laptop Hard Drive" for USB3 systems will use a USB3.2 Gen1 label and communication subsystem capable of moving data up to 5 Gb/s. But the HDD unit inside can access its data for reads and writes at a rate of maybe 1.5 Gb/s (the old SATA I speed) because it is designed for slower performance to use less power from the USB port. Even if you are using, instead, an enclosure with a full-performance SATA II or SATA III desktop HDD and a proper power supply unit, the data speed of that drive may get up to 3 Gb/s, maybe a bit more, but never to 5 Gb/s. IF instead you put the fastest new SSD unit into that enclosure it might be able to access data faster than 5 Gb/s, and the enclosure really should be one with a USB3.2 Gen2 spec of up to 10 Gb/s. So the actual performance speed of any device is NOT the speed of the USB spec - it is limited by the real speed of the device inside.

As the orgiginal writer said, Gen1x2 exists but is rarely used - other versions are better.

Wire count. Old USB2 cables have 4 wires. Two are the + and - power supply. The other two (ideally) are a twisted pair of data wires used for half-duplex communication. That is, data can flow from device 1 to device 2, OR the opposite direction, but not in both directions at the same time. The USB3.2 cables all have NINE wires. The first four are the same as USB2 and the contacts in the connectors on the cable ends are the same for these. IF the communication prototcol being used on that cable has been negotiated by the devices as a USB 2 type, then only these 4 wires are in use. Of the five added wires, one is an extra Ground, and the other four are TWO PAIRS of shielded twisted wires. Each pair is used in a differential signal transmission mode (this plus the shield reduces noise a lot to allow faster data rates), and each is a one-directional channel. So one pair sends data from device 1 to device 2, and the other pair sends in the opposite direction. This accomplishes simultaneous full duplex data transfer. A cable being used in any USB3.2 mode will use these two data channels plus the two power lines of the USB2 wires, and will not use the two USB2 data wires. Which mode is used - USB2 or USB3.2 Genn - is negotiated when the devices first connect, and preference is given to the fastest rate possible.

All USB3.2 cables have this structure. The difference between them is in subtle details of insulation, capacitance, and shields that impact the maximum data rate that can be used on each of those pairs of data lines; they do not differ in wire count. Those details determine whether the data rate can be 5 or 10 Gb/s. To get 20 Gb/s data rate, complex data encoding is used that requires all 4 wires.

The older-looking Type A CONNECTORS on USB3.2 cables are different from the old ones used for USB2 although they look similar. Both types have a plastic insert that supports four flat metal contacts used for the USB2 lines. The new USB3.2 cables have five additional contacts as round dots on the end of the plastic support (or in the rear) for the added lines. Typically the plastic insert in a USB2 cable is Black, while the one in a USB3.2 Gen1 cable is Blue. For Gen2 it may be Blue also, or perhaps Red. Type A is never used for Gen 2x2.

The new Type C connector has many more contacts - 12 on each side of a central plastic support tongue. They are arranged so one side duplicates the other such that the cable may be plugged in either way "up" and the connections are the same. This is a convenience and a help. The contacts are small and that can limit max current capacity, but basically each wire in the cable feeds to TWO identical contacts in the connector to distribute the load. Size and spacing has impacts on capacitance and inductance in the connector. The result is that this small configuration makes higher data transfer rates possible, so it MUST be used for the Gen 2x2 versions. Further, Gen2 speed of max10 Gb/s may or may not be achieved using a Type A connector and cable, but should always be achievable using a cable with Type C connectors.

USB ports may have other uses, very commonly for charging devices. The USB3.2 system greatly expanded the options for charging and added systems for negotiation between devices of charging rates, similar to the negotiation of data transfer rates. Most ports for high charging rates use the wires and contacts in the cable and Type C connector for different purposes. For that reason, the CHARGING PORTS on a powered USB3 Hub normally do NOT include any data transfer ability. Further, CHARGING CABLES may have odd wires and connections so they are not suitable for use as "normal" USB3 data cables.

 

[Nikolay Mihaylov]

I'd always assumed that the 5Gbps referred to the real data throughput rate (and it gets even higher when counting the overhead), as the real throughput is the only thing the average user cares about

Well, the USB Implementers Forum doesn't care about what the users care about. Like with so many other things, they rely on users being dumb in the bigger-is-better sense. Although, I would say that this should only matter when users have choice. Which other peripheral connectivity technology there is that uses the 5gbps moniker to implement higher actual data speeds than USB 3.2Gen1? But I digress.

Like several other technologies - SAS/SATA/PCIe 1&2, USB 5Gb use 10 physical bits to represent each 8 data bits. So the channel data rate expressed in useful bytes is 5000 / 10 = 500MB/s. Now, the datas tream is broken into packets that add some data to each packet so the actual data payload is around 90% of the data flowing through the wires. Which makes for around 450MB/s.

In a similar way, SATA3 offers 6Gb/s which translates to 600MB/s of which you can only use around 550MB/s for user data.

Are you sure that for 5Gbps USB 3 the real data rate is actually only 450Mbps? 450Mbps is actually SLOWER than USB 2 (with a real data rate of 480Mbps).

Please note the difference between bits in Gb/s and bytes in MB/s. It's customary to use lower case b for bits and upper case B for bytes.

Also you mentioned Gen1x2 in your comment. I've never heard of Gen1x2 before

According to the Wikipedia article for USB3 it is part of the 3.2 standard. I don't know if that's really the case since I haven't read the standard. And I think it's unlikely for any device to offer Gen1x2 only. It would either be Gen1x1, Gen2x1 or Gen2x2. But it might be useful as a degraded mode for Gen2x2 devices in case of longer distances/crappy cables/whatever.

But frankly, at this point I'm looking forward to USB4 or, as that's what USB4 is, Thunderbolt. Well, Thunderbolt docks and devices already exist but a wider adoption will bring the prices down. The difference in price for comparable feature sets (sans the data speed) is 1:3 to 1:4. It's ridiculous. But I digress again

Cheers,