Live data from Hacker News

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

philip.greenspun.com

341–350 of 511 posts

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

#341
post #335

Earlier quoted context omitted.

Just tighten one of them, its what i do half the time

And other half of the time you tighten another?

Yes :P. I’m a coward and never put it in without tightening at least one or both

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

#342

I think the main reason is cost. A proper usb-c port should be able to be used for storage, display and powering devices with high power requirements. Even 15W is kinda high power unless you're buying a top end space heater from Intel. Now multiply this circuitry by 6-10 ports...

There is no need to make every port fully featured. Even MacBooks doesn't have that for the kind of price they demand

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

#343
post #125
post #121

Earlier quoted context omitted.

something that might be useful to you for that; you can get m.2 slot network adapters which you could use on (some) ITX boards. there's at least 1 & 10gbe ones available.

Sure. But if you want to actually have two m.2 drives to either mirror your root filesystem or use one as a secondary? It gets tangly quickly. inb4: CPUs as well as commonly used chipsets have enough lanes, it's not that.

Add 4- or 8-port GbE NIC to PCIe, add 64GB of RAM, add USB stick for router operating system, and you should be able to handle edge traffic for a small office.

What's your goal here??? Multiple NIC in ITX form factor, for what?

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

#344

I think the main reason is cost. A proper usb-c port should be able to be used for storage, display and powering devices with high power requirements. Even 15W is kinda high power unless you're buying a top end space heater from Intel. Now multiply this circuitry by 6-10 ports...

There is no need to make every port fully featured. Even MacBooks doesn't have that for the kind of price they demand

Having different ports that are mechanically compatible but not feature-wise is a UX failure. The blue/black color labeling made that somewhat acceptable for USB-A, but now that motherboards have started to use blue/red color labeling for ports that are feature compatible (just connected differently internally), even that color labeling is becoming a UX failure.

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

#345

My experience has shown that USB C connectors are far more fragile than A, and every time I plug one in, for some reason the fit just feels vague and unsatisfying. An A plug, when you get it the right way, slides in solidly. As others have mentioned, C seems to have been designed to be the smallest possible, which makes sense for mobile devices, but not for a desktop and its peripherals.

I got my HTC vive.out of the closet today and as usual hate USB micro and love USB c.

Usb-c just fits.

Do you mean the real USB plug the bigger one? Those are dependent on my device. The USB a port on my desktop case are okay, the ones at the back of my Mainboard way to tight.

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

#346
post #200

Earlier quoted context omitted.

It’s pretty crazy that an operating system needs that much at all. I mean, we all get used to these big numbers but 4GB is actually a lot of bits. A single GB can store more books than a human can read in their lifetime. Somehow, we have accepted that the most fundamental piece of software on a computer should require many lifetimes worth of books of RAM. We have come a long way from thinking we would never need more…

But you’re conflating data at rest vs data in memory. A pdf for example can be very small but then take up considerably more memory to display an embedded image once it’s done decompressing for render. Similarly text is more expensive just by virtue of higher resolution displays. What we ask of our machines has expanded greatly, and the immediacy with which people want things to load is more exacting too, so a lot of…

I used such a computer (4GB DDR3 ram and 2 GHz dual core Pentium) for some very basic work tasks on Windows 10 LTSC (the one without the bloat). It had Windows due to some weird hardware incompatibility, so I used it for about a month. The computer is used for browser with a couple of Google Spreadsheets opened tabs and a couple of messengers once in a while to reply to some messages.

It was almost unusable in that very scenario. I could handle a couple of tabs, but once I opened a couple of extra tabs (e.g. some search either through Google or any local shops) it literally hanged. I had to wait for it (to swap, I suppose) for a couple of minutes to unfreeze.

I went very angry over time and replaced the weird incompatible with Linux component (GPU) for another one and now it runs Fedora 39 with default Gnome 45 installed. It works perfectly well and it hangs only when I open relatively big number of tabs, 15 to 20, which is a lot for the use-case of that very machine. I thought of optimising the machine to run just swaywm (which eats like 300 MB compared to over 2GB of Gnome), but I never faced any issue for over a 3 months period.

So I would confirm that even 4 GB is plenty unless you need a big number of browser tabs opened. But not with Windows. Windows was usable on that machine if you don’t use browser heavily, but very very lightly.

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

#347
post #226

Earlier quoted context omitted.

Minimum is just that, minimum. Have you ever seen a broken standard USB port or connector? Me not once in my entire life, unlike micro-USB or USB-C.

As the goto electronics guy I have swen broken USB A but not broken USB-C (yet). USB-C is sturdy, the weakness is either going to be the cable or the bonding of the copper with the PCB. Both is basically a matter of go much money the manufacturers are willing to spend.

My experiences differ. I've had plenty of USB-C devices of which the port becomes very loose after a while of normal use. I'm glad my current phone has wireless charging, not because of its convenience but because that way I don't wear out the USB-C port.

USB-A is so simple, two holes and some leaf springs make for a satisfying clunk that doesn't wear out so quickly.

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

#348
post #329

Earlier quoted context omitted.

Sure. But if I remember correctly the cable ends are both male, so it was never the device that got bent. In some sense weakness on the right part is a feature.

every meeting room had a long semi-permanent cable wired through the walls / floors. bent pins on those were not fun.

Once you’ve been involved in repairing them, you start to insist on the sane approach - cable in the wall, connected to a faceplate which has a plug on the back and a plug on the front. That way any damage can be repaired without re running the cable.

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

#349

Earlier quoted context omitted.

Personally I think these numbers are fabricated horseshit, drafted by some engineer behind a desk running endless simulations. The USB industrial complex made the same claims with micro-USB vs mini-USB; and are now doing it again with Type-C. My empirical evidence: I've now had to replace two laptop motherboards due to a (mechanically) failed USB-C port. Guess what I haven't had to do in 25 years? Replace a failed US…

Exactly. There is no way a significantly smaller connector with way more pins can be as robust as a large one with less pins. It is 100% horseshit. My anecdote to add, but this time comparing USB-c power delivery with the lowly ~8mm barrel jack. I recently got a laptop shipped to me by a remote client to use to connect to their network. I pull it out of the box set it up and I notice the usb c power cable has the sli…

Since 2018 I've charged, operated, and booted a variety of USB-C devices over a variety of USB-C PD power sources.

e.g MacBooks of the 60W and 90W variety I've operated on 90, 60, 45, 30, 20, 15, and 5W chargers, car adapters, power banks, displays, powered hubs, or other USB PD devices. Of course on 5W they can't charge while in use (even idling, where they ~maintain charge) and booting requires them to have enough of a battery charge to support the boot power surge, but it has always worked: PD has always negotiated to the best power the charger could do. IOW when manufacturers do their job, it works.

> A barrel jack wouldn't care in the slightest about such kink.

Oh boy I would disagree. I've had decades of terrible experience with barrel jacks, many of which were completely invisible to the eye. I would not be surprised the small kink would have hidden much worse internal damage or defect. If it failed to negotiate power that's because of an abundance of safety, and I'd rather have that than a fire hazard silently being accepted and internally arcing under load.

> The PSU I had was 90W while the other laptop required 120W.

What would be the non USB PD alternative? Proprietary PSUs, with either one of: non interchangeable because of a different plug, the same plug but the most power hungry computer still refusing to boot from some proprietary charger id protocol or hardware, the same plug but different voltage and either doesn't work of fries the computer's power circuit (I've had all of those back in the day).

So your current USB-C situation isn't any worse than it was before USB PD existed. In fact it's better WRT failure modes because USB-C PD compliance means it will negotiate a safe voltage (including taking the cable into account). Of course if you plug in a non-compliant charger all bets are off but that's no worse than before when there was no power delivery standard whatsoever.

That said, as demonstrated by the MacBook example, a manufacturer doing their PD job would not lock the computer out at the bios level - instead waiting to have a safe charge on the battery -, a kind of locking that also happened back in the day of proprietary plugs among the same manufacturer.

Therefore I do not see any reason to blame that on USB PD "horseshit" instead of that particular manufacturer.

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

#350
post #272

Earlier quoted context omitted.

Still we live with power sockets and lamp sockets that hark back to Edison times, more than a century ago, and live relatively happily. Backwards compatibility with millions (and hundreds of millions) of existing devices is a powerful force. While I'm in favor of USB-C connector eventually replacing USB-A connector completely, I think it's best done gradually, with new devices still offering the occasional USB-A sock…

New mobos still have PS/2, especially the higher-end ones. Edit: don't believe me? https://www.gigabyte.com/Motherboard/Z790-AORUS-TACHYON-rev-... 1 x PS/2 keyboard port 1 x PS/2 mouse port

And long may it last, there is no way to truly replicate a ps/2 port/header with adapters etc

Hopefully if they ever go away for good someone makes a pcie ps/2 card with proper interrupts (if that is even possible behind abstractions like MSI) rather than converting ps/2 to usb

Post reply on HN