Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Why aren't motherboards mostly USB-C by now? (2021)

philip.greenspun.com

Post reply on HN
Print view View original

81–90 of 511 posts

Re: Why aren't motherboards mostly USB-C by now? (2021)

#81

Post by crote »

anakaine wrote: post #34
korginator wrote: post #27

USB-A should be relegated to the rubbish-heap of poorly designed interfaces for the simple reason that it has no physical indication (e.g., RS-232C) or clear visual indicator for orientation, and you get it wrong 50% of the time. It's way beyond time the USB-A interface is eliminated. USB-C can't get here fast enough. My MacBook pro only has USB-C ports, they work well for data, video out, as well as power (charging)…

This said, USB-C needs to have its standards updated, and standardised. Eg all cables must support a minimum of USB 3.1 with power and data, with appropriate shielding. Stop yielding so many concessions. I'm already in a spot where I have a number of USB-C to C cable varieties that all look basically the same, but have very different characteristics. One cable basically refuses to charge anything. Another has no data…

A big part of this is that a lot of manufacturers are just completely ignoring the standards.

The standards specify a rather limited set of cables, and it specifies clear ways to label them which make it extremely obvious to consumers what the cable is capable of. But manufacturers just... can't be bothered to actually follow the standard.

The main reason why there isn't one cable is that it has mutually-incompatible goals. A high-speed data cable needs expensive, thick, shielded data wires - and is limited to about 80 cm in length due to signal integrity issues. All completely fine for connecting a laptop to a docking station.

On the other hand, a smartphone charging cable needs to be cheap, long, and flexible - but it doesn't need to carry high-speed data. The aforementioned docking station cable would technically work, but it'd be $30 rather than $3 and provide a far worse user experience.

Re: Why aren't motherboards mostly USB-C by now? (2021)

#82

Post by lxgr »

cesarb wrote: post #74
gweinberg wrote: post #37

Earlier quoted context omitted.

Maybe it's a chicken and egg thing, but why do we still have usb-a keyboards and mice and thumb drives? wouldn't it make more sense for them all to be usb-c?

Having them be USB C would mean they could only be used on USB C ports (adapters from USB C socket to USB A plug are forbidden by the USB standard, because they would allow creating the forbidden USB A to USB A cable), while having them be USB A allows them to be used on both USB A ports and (with a simple passive adapter) USB C ports.

That’s a very good point. It doesn’t stop some manufacturers from still shipping the “forbidden” adapter type for just that use case, though.

I prefer USB-A for FIDO authenticators for that reason (and because the plug is more robust for USB-A and basically indestructible; C plugs can and do get bent on a keychain).

Re: Why aren't motherboards mostly USB-C by now? (2021)

#83

Post by tom_ »

I feel dumber after reading anything by Greenspun. Look at me, actually saying this out loud, wasting all your time with the sort of pointless comment that 5 minutes ago I'd have laughed at the very idea of posting. That's the power of the man's content. What a guy!

Re: Why aren't motherboards mostly USB-C by now? (2021)

#84

Post by crote »

eternityforest wrote: post #56
B1FF_PSUVM wrote: post #12

Earlier quoted context omitted.

USB-C is the bees knees if it works right. Which does not seem to always happen. Plus, "active cable" is close enough to an oxymoron to make me afraid we overreached. (I do like that a small pocket kit with an USB C cable and a couple of adapters seems to cover most needs.)

The one thing I don't quite like about active cables is that they didn't use Dallas One-Wire or i2c or uart. Otherwise, it seems to work fine. Hobby level stuff would be a lot simpler if any old microcontroller could talk to a PPS supply, interrogate cables, etc.

I2C or uart would require an additional wire, so that'd be cost-prohibitive. But I do agree that something like One-Wire would be better from a hobbyist perspective.

Although probably easy to implement in silicon, USB PD is an absolute nightmare to DIY. Pretty much your only option is to get a dedicated PHY like the FUSB302B to do the actual electronic part of PD comms for you.

Re: Why aren't motherboards mostly USB-C by now? (2021)

#85

Post by gattis »

lxgr wrote: post #80
gattis wrote: post #79

any device that doesnt need a 20gbps or higher link is better off connected to your motherboard via a usb-a to usb-c cable. the "a" connector has not been deprecated. in some ways it is the superior connector, higher mating cycles, less fragile, less likely to get ripped loose, much more torque required to bend, and so on. the "c" connector is more useful on the opposite end where you will be plugging in and out more…

> higher mating cycles It’s the other way around. USB-C sockets are much more robust than their A counterparts, since the spring is in the plug for C, as opposed to in the socket for A.

perhaps. but thats why you have the C at the other end because it doesnt make sense to be frequently plugging things in and out of the motherboard and and wearing down those connectors.

but i doubt you will get more cycles out of C in real life. that might be what the spec says but given the pitch of the pins on the C connector compared to the A, you have a world of greater tolerance to work with

Re: Why aren't motherboards mostly USB-C by now? (2021)

#86

Post by lxgr »

chungy wrote: post #51
cassianoleal wrote: post #50

Earlier quoted context omitted.

More likely they mean micro-USB. USB-B is anything but micro.

Micro-USB is a variant of the USB-B connector. The whole point of A and B was that A ports were used on the "host" system, and B were used on the peripheral. If you remember USB On-the-Go, that was an attempt to reverse the trend and allow a USB-B device (including micro ports) to act as a host. USB-C erased that distinction in favor of a full duplex network connection between two hosts.

USB-OTG ports are technically mini-AB and micro-AB ports that can fit both A and B plugs.

And USB-C is neither full-duplex (at least not for USB 2) nor host-to-host; there is a protocol negotiation and some devices can never act as hosts, although some can indeed assume both host and device roles.

Even in a “host-to-host connection”, only one side will act as the host.

Re: Why aren't motherboards mostly USB-C by now? (2021)

#87

Post by p1mrx »

crote wrote: post #70
p1mrx wrote: post #8

USB-C was designed for highly-integrated mobile devices. Motherboard vendors don't want to deal with power delivery and DisplayPort, and users will be confused when that stuff doesn't work. When a USB-C port is out in space, rather than flat on a desk, C-to-A adapters create a lever arm that increases the risk of damage. Though this can be solved by using adapters with ~100mm of wire in the middle.

Motherboard vendors don't need to deal with PD and DisplayPort. USB-C will work just fine without them - it'll just behave like any regular old USB port.

USB-C is like a regular old USB port, with twice the data lanes (or a mux[1]) and a transistor to turn on the power when requested.

In other words, it costs more.

[1] https://richardg867.wordpress.com/2020/02/29/usb-c-done-chea...

Re: Why aren't motherboards mostly USB-C by now? (2021)

#89

Post by lxgr »

crote wrote: post #67
korginator wrote: post #30

Earlier quoted context omitted.

USB-C is simply more convenient to use, and you get it right 100% of the time, instead of your orientation being wrong 50% of the time with USB-A.

Just don't accidentally stick USB-C into a USB-A port. It will fit and it will short the port, causing the machine to crash. Speaking from experience...

Sometimes I look at USB-C plugs and US power outlets and wonder if there’s a shock hazard there…

Re: Why aren't motherboards mostly USB-C by now? (2021)

#90

Post by amelius »

No matter what devices I use, USB seems to give me a bit error every few terabytes or so, over cheap cables >1ft. This makes me wonder: how do I find good (certified) cables, and why isn't there forward error correction in the USB protocol such that cheap cables simply work?
Post reply on HN