4 ms·

This isn’t specific enough for some situations. Let’s say you have a “USB-C 40G 45W” port. For power, your port could supply either 15V 3A, or 20V 2.25A, or bot

This isn’t specific enough for some situations. Let’s say you have a “USB-C 40G 45W” port. For power, your port could supply either 15V 3A, or 20V 2.25A, or both. If I have a device that wants 45W but only supports 20V, and your port only supports 15V via PD, it’s going to fall back to 5V and you’ll get a max of 15W out of the port. Similarly, on the data side, can your “USB-C 40G 45W” port be used to directly connect to a display via a USB-C to DisplayPort cable? Will it support Thunderbolt 4, e.g., can it connect to a TB4 dock and drive two 4K 60hz displays from a single cable? USB is a mess. Your suggestion is admirable, but given what the spec already supports, I don’t think there’s a way to convey all the info that should be conveyed in a short, easy to remember way.

aappleby 4y ago
I would say that you can't use the simplified cable markings unless your cable supports the full set of features implied by the marking - if your cable says "20G", it must support 4k60 * 1. If it says "40G", it must support 4k60 * 2. "100w" -> must support all USB-PD volt*amp combos <= 100W. Would require compliance testing beyond what is currently done for cables, but would also remove ambiguity for the user.
- cmatthias 4y ago
  Yeah, if we could totally re-do everything, something like this would be ideal. But the problem is, under the current spec it’s possible, and quite common (at least on the PC side of things) to have a 20gbps USB-C port that doesn’t support displays at all. I’m not giving Apple a pass here either; the product info pages for all of the M1 (non-Pro) Macs all make reference to “Thunderbolt / USB 4” ports, which is imprecise at best, and intentionally deceptive at worst.
  - tuatoru 4y ago
    I think they're suggesting a few suffixes: - USB40.100W - USB only, 40G, 100W PD USB20H.100W - USB with HDMI alternate mode and 100W PD (HDMI alternate mode exists in theory only at present though) USB40T.60W - USB with thunderbolt alternate mode USB40DT.40W - both thunderbolt and displayport alternate modes supported USB40DHMTV.60W - USB with all alternate modes (so far: displayport, HDMI, MHL, Thunderbolt, VirtualLink). Being allowed to color-code cables would be nice too. The USB-C spec forbids that IIRC. Edit: imagine if the USB consortium had authority in the 1900s - 1920s. "Let's have mains power, phones, and radio antennas all use the same connector!"
    - cmatthias 4y ago
      The poster didn't suggest suffixes like this. I took their point to be the opposite: let's have short, easy-to-read, meaningful names for each possible combination of port. But regardless, this is what I was getting at when I said "I don’t think there’s a way to convey all the info that should be conveyed in a short, easy to remember way." The spec is far too overloaded with functionality to have any kind of naming system that is both sane and unambiguous. For example, USB40DHMTV.60W (rolls right off the tongue!) doesn't specify what PD voltages are supported, or whether the "T" component is Thunderbolt 3 or Thunderbolt 4.
2Gkashmiri 4y ago
why are we stopping at USB4 + xx? why not do USB4, USB5, USB6, USB7...USBn? each version is an exact spec so we could go to like USB500 and the cables would still be different because it could be like a SKU?
- cmatthias 4y ago
  OK let’s say we do that. For the sake of argument, let’s define USB14 as what we call “Thunderbolt 4” today (basically USB4 plus a few extra things that are optional under the USB4 spec), with 100W of PD. And let’s define USB63 as plain old USB 3.1, at 10gbps with no alternate modes, with 30W of PD. I buy my mom (who is not technically savvy) a computer with 2 USB14 ports and 2 USB63 ports. She wants to buy a cable that will charge her USB34 (yet another version of USB corresponding to USB-C at 2.0 speeds with no PD) headphones. How does she know what cable to buy, and what port to plug it into? If you think about this some more, you might agree with me that it’s likely worse than the current naming scheme.
  - 2Gkashmiri 4y ago
    why should the ports be different? why not standardize the ports? even then, we have color coding on USB 3, why not take that further? like USB14 has RED on both USB port and cable end, different for USB 63? in the example, the USB63 port would have USB63 written on the top of port itself so user, even a non-technical user can go to the market and find USB63 cable..... edit: like a TV can have USB65 port and the documentation states it would support cables USB40-66 so you could go and buy any of them. the documentation would list, USB41 only does 4K 60, USB44 4K 120, USB50 8K 60, USB51 8K 60x2. on the device side, your laptop would tell you, "your laptop support cables USB 50-60 so you can decide to NOT buy any cable from USB40-49 because your laptop wont support it but your TV can so if you wanted them to work together, not buy them....
    - cmatthias 4y ago
      Again, this sounds like a nightmare and more confusing than the current naming scheme. I agree that in theory the USB working group should just standardize the ports, but if they’re doing that right, there’s no need for 500 different versions at all. Just mandate that for your device/port/cable to be called USB4, it must support all alternate modes, all PD voltages, etc.
ddejohn 4y ago
What about something like this for connector/port labelling: `USB-<connector> <protocol>.<speed>.<max_voltage>.<max_amperage>` Either with 1-1 mapping of "version" numbers to values, or just "revision" numbers which would need to be translated. E.g. * `USB-C 3.20.15.3` for 20Gbps (although this doesn't tell you how many layers, and what encoding), 15V, 3A or `USB-C 3.5.4.1` where the `.5` maps to the data transfer speed of 20Gbps over 2 lanes with 128b/132b encoding, the `.4` maps to max voltage of, say, 20V, and the `.1` maps to 2.25A (this way you don't have to deal with fractional amperages) Slap an `x` anywhere that the spec isn't fully utilized, e.g., for a charging cable, the `speed` would be `x`, so the label would read `USB-C 3.x.4.2` or whatever.
- cmatthias 4y ago
  There are still holes in this scheme which would require more fields and/or just leaving data out. Ports/cables can support multiple (but not necessarily all!) voltages for PD. This scheme as written doesn’t capture Thunderbolt compatibility at all, or alternate modes for DisplayPort. My point is, while it’s certainly possible to come up with an unwieldy naming scheme that covers everything, given the current state of what USB supports, it’s not possible to have a scheme that’s simple and unambiguous. Major changes would be required to the protocol/certification requirements to make that happen.

aappleby 4y ago

I would say that you can't use the simplified cable markings unless your cable supports the full set of features implied by the marking - if your cable says "20G", it must support 4k60 * 1. If it says "40G", it must support 4k60 * 2. "100w" -> must support all USB-PD volt*amp combos <= 100W. Would require compliance testing beyond what is currently done for cables, but would also remove ambiguity for the user.

cmatthias 4y ago
Yeah, if we could totally re-do everything, something like this would be ideal. But the problem is, under the current spec it’s possible, and quite common (at least on the PC side of things) to have a 20gbps USB-C port that doesn’t support displays at all. I’m not giving Apple a pass here either; the product info pages for all of the M1 (non-Pro) Macs all make reference to “Thunderbolt / USB 4” ports, which is imprecise at best, and intentionally deceptive at worst.
- tuatoru 4y ago
  I think they're suggesting a few suffixes: - USB40.100W - USB only, 40G, 100W PD USB20H.100W - USB with HDMI alternate mode and 100W PD (HDMI alternate mode exists in theory only at present though) USB40T.60W - USB with thunderbolt alternate mode USB40DT.40W - both thunderbolt and displayport alternate modes supported USB40DHMTV.60W - USB with all alternate modes (so far: displayport, HDMI, MHL, Thunderbolt, VirtualLink). Being allowed to color-code cables would be nice too. The USB-C spec forbids that IIRC. Edit: imagine if the USB consortium had authority in the 1900s - 1920s. "Let's have mains power, phones, and radio antennas all use the same connector!"
  - cmatthias 4y ago
    The poster didn't suggest suffixes like this. I took their point to be the opposite: let's have short, easy-to-read, meaningful names for each possible combination of port. But regardless, this is what I was getting at when I said "I don’t think there’s a way to convey all the info that should be conveyed in a short, easy to remember way." The spec is far too overloaded with functionality to have any kind of naming system that is both sane and unambiguous. For example, USB40DHMTV.60W (rolls right off the tongue!) doesn't specify what PD voltages are supported, or whether the "T" component is Thunderbolt 3 or Thunderbolt 4.

2Gkashmiri 4y ago

why are we stopping at USB4 + xx? why not do USB4, USB5, USB6, USB7...USBn? each version is an exact spec so we could go to like USB500 and the cables would still be different because it could be like a SKU?

cmatthias 4y ago
OK let’s say we do that. For the sake of argument, let’s define USB14 as what we call “Thunderbolt 4” today (basically USB4 plus a few extra things that are optional under the USB4 spec), with 100W of PD. And let’s define USB63 as plain old USB 3.1, at 10gbps with no alternate modes, with 30W of PD. I buy my mom (who is not technically savvy) a computer with 2 USB14 ports and 2 USB63 ports. She wants to buy a cable that will charge her USB34 (yet another version of USB corresponding to USB-C at 2.0 speeds with no PD) headphones. How does she know what cable to buy, and what port to plug it into? If you think about this some more, you might agree with me that it’s likely worse than the current naming scheme.
- 2Gkashmiri 4y ago
  why should the ports be different? why not standardize the ports? even then, we have color coding on USB 3, why not take that further? like USB14 has RED on both USB port and cable end, different for USB 63? in the example, the USB63 port would have USB63 written on the top of port itself so user, even a non-technical user can go to the market and find USB63 cable..... edit: like a TV can have USB65 port and the documentation states it would support cables USB40-66 so you could go and buy any of them. the documentation would list, USB41 only does 4K 60, USB44 4K 120, USB50 8K 60, USB51 8K 60x2. on the device side, your laptop would tell you, "your laptop support cables USB 50-60 so you can decide to NOT buy any cable from USB40-49 because your laptop wont support it but your TV can so if you wanted them to work together, not buy them....
  - cmatthias 4y ago
    Again, this sounds like a nightmare and more confusing than the current naming scheme. I agree that in theory the USB working group should just standardize the ports, but if they’re doing that right, there’s no need for 500 different versions at all. Just mandate that for your device/port/cable to be called USB4, it must support all alternate modes, all PD voltages, etc.

ddejohn 4y ago

What about something like this for connector/port labelling: `USB-<connector> <protocol>.<speed>.<max_voltage>.<max_amperage>` Either with 1-1 mapping of "version" numbers to values, or just "revision" numbers which would need to be translated. E.g. * `USB-C 3.20.15.3` for 20Gbps (although this doesn't tell you how many layers, and what encoding), 15V, 3A or `USB-C 3.5.4.1` where the `.5` maps to the data transfer speed of 20Gbps over 2 lanes with 128b/132b encoding, the `.4` maps to max voltage of, say, 20V, and the `.1` maps to 2.25A (this way you don't have to deal with fractional amperages) Slap an `x` anywhere that the spec isn't fully utilized, e.g., for a charging cable, the `speed` would be `x`, so the label would read `USB-C 3.x.4.2` or whatever.

cmatthias 4y ago
There are still holes in this scheme which would require more fields and/or just leaving data out. Ports/cables can support multiple (but not necessarily all!) voltages for PD. This scheme as written doesn’t capture Thunderbolt compatibility at all, or alternate modes for DisplayPort. My point is, while it’s certainly possible to come up with an unwieldy naming scheme that covers everything, given the current state of what USB supports, it’s not possible to have a scheme that’s simple and unambiguous. Major changes would be required to the protocol/certification requirements to make that happen.