Why do Modern Motherboards Still Have PS/2 COnnectors?

[DemonDragonJ]

Wow, another user revived this thread after more than an entire year had passed since its previous post: that is most impressive.

So, Konrad, shall I presume that you favor PS/2 over USB, and that you believe that PS/2 shall not be obsolete at any point in the near future? If you believe that, then I have another question: what potential does the PS/2 interface have to grow and improve? USB can continue to improve with each new version, but I have not noticed any revisions or upgrades to the PS/2 interface in some time, which suggests that it does not have any room for upgrading.

[Konrad]

I'll admit that PS/2 spec was never designed to be scalar, never designed to accept any other devices. It's really just another PC/AT component which worked so well it became the de-facto industry standard for many years. I don't even think any official documentation was ever published, although the tech is comparatively simple to figure out.

USB standards involve all sorts of compliance testing, engineering, and extensive intercompatibility testing, where even the most trivial details merit very serious consideration. All governed by a central "non-profit" sort of corporation which does a pretty good job of ensuring conformity and compliance. A vast number of el-cheapo USB products do exist, at least products with USB plugs which generally work well enough, but a careful observer will note the conspicuous lack of proper USB branding on these products.

The reason PS/2 is still integrated in PCs is legacy compatibility. PS/2-capable keyboards and mice are still being sold (although many are actually USB devices with some kind of internal PS/2 emulation or compatibility mode). Certainly, we are at (or soon will be at) the point where PS/2 connectors disappear from laptops and other devices, but I'm willing to bet they'll remain standard issue on desktop mobos for many years. Legacy junk may not be fancy, but it's widely available, dirt cheap, and basically always works even when the latest-greatest fancy new tech fails - and again, basic user interfaces (keyboard, mouse) are sort of critical requirements for fixing everything else, including added functionality on the basic user interfaces.

Microcontroller guys and modders/hackers/etc have developed all sorts of clever uses for PS/2 parts. Of course it would be foolish to expect major PC engineering/marketing to accommodate our puny minority if they ever decide to chop PS/2, and we're always able to modify or construct our own PS/2 interfaces when none are provided.

For the record, I deliberately chose PS/2 protocol on my "perfect" gaming keyboard, it is more responsive and more capable than USB/HID allows (I considered using a COM port approach for instant-response full NKRO, but decided I can tolerate PS/2 limitations instead, lol). But at the same time, my keyboard cable is actually a bundle with pass-through USB and audio functions, since it's just really convenient to have these functions on my keyboard ... and I also use a second keyboard (USB) for general typing/programming stuff. My uber gaming mouse and regular cheapo mouse are both USB, since PS/2 response is clearly inferior in this regard ... and the only pure PS/2 mouse I still own is an ugly biege ancient ball-mouse thingy without a scrollwheel.

The reality is that the keyboard limitations designed into USB spec already exceed anything most users will ever need or notice, unless perhaps doing some frantically extreme multikey-mashing gaming. I would expect that if it was ever perceived by the market/industry as a "serious" limitation then a faster HID standard would be officially engineered on subsequent USB specs (I think this will eventually really happen, but only when all USB keyboards force the need with ridiculously extensive features and macros and such) ... but we will not see any further changes in PS/2 spec.

Sorry about the thread necro (and walls of text haha). I think I fell for the spammer again.

[TLHarrell]

Konrad, all your posts are a huge wall of text. Of course, it's relevant and I always consider it a good read. You either have a massive electronics encyclopedia you're copy/pasting from, or you've got some serious EE chops.

I actually found it rather hilarious that when I built my current desktop system I needed a PS/2 keyboard to get into BIOS (simply because it wouldn't enable USB prior to BIOS entry in the boot cycle), as well and a floppy drive (to install RAID drivers). Of course, I keep the PS/2 keyboard around for when I occasionally work on PCs for others as it always works. Most of the time they only bring the box anyway.

[Konrad]

I think self-educated hobbyist/amateur EE people often know more (irrelevant details) about the topic than properly educated EEs. Avocational passion rather than just a plodding vocational money path.

[TLHarrell]

I have a very good understanding of a great many things, electronics included. But my knowledge is not as encyclopedic as yours seems to be. Certainly paves the way for more time spent fooling around with stuff, less time spent looking for the correct information and getting calculations right.

You should certainly look into gracing us with some EE type articles. The site could always use some content, especially good, relevant and exclusive content.

[Konrad]

This particular specific topic is one I just happen to know a great deal about after building my own "perfect" keyboard controller. lol, I wouldn't participate in discussions about topics I don't know anything about.

[Xpirate]

I actually found it rather hilarious that when I built my current desktop system I needed a PS/2 keyboard to get into BIOS

This is the reason why I still have at least one PS/2 keyboard in my small repository of computer junk.

[Konrad]

I've actually "retromodded" USB keyboards to PS/2. Real PS/2, with the old scansets and controller interfaces, not the hybrid (USB native, PS/2 emulated) types which have been standard issue for at least a decade.

I constantly need these keyboards to interface with non-PC platforms. Older peripheral tech seems to dominate industrial control panels and such. Sometimes (though with diminishing frequency), my microcontroller projects need a quick keyboard interface but don't merit the effort and complexity of integrating USB libraries.

[Azrael5]

Simple, because PS/2 connectors are better than USB for resources consumption and time responsivity, and simultaneously keys typing.

So PS/2 motherobards are better than without this conncetors.

REasons why PS/2 is preferred to USB:

1) PS/2 is hardware interrupt-based, while USB port is polling-based. When you press a key on a PS2 keyboard, it generates a hardware interrupt immediately, while USB polls your keyboard many times per second (125 Hz by default, up to 1,000 Hz in "gaming" keyboards) to see if any keys a pressed. This means that PS2 keyboards will have the lowest latency - although in all likelihood, you will never notice a difference. Perhaps more importantly, polling is more CPU-intensive, especially if high polling rate (like 1,000 Hz) is used.

2) PS2 supports full n-key rollover. USB keyboards are generally limited to 6KRO (able to recognize up to six simultaneous keystrokes). Of course, it also depends on the keyboard: most cheap keyboards are just 2KRO, and it does not matter whether you connect such a peripheral via USB or PS2 port: it will still register only 2 keys at once. Full NKRO is a feature usually found in gaming keyboards.

3) PS2 keyboards are more compatible with older hardware and software, which might be important for some users. To be perfectly fair though, turning on "Legacy USB" in BIOS means your USB keyboard will work in those cases as well, which almost makes this a moot point.

So Ps/2 input devices get a excellent latency in confront of USB.

[Konrad]

1) USB polling is based more on bandwidth than timing. ~125Hz is a fair assumption, but ~1000Hz is - at best - a marketing department's unrealistically optimistic spin of "the best case scenario" on an otherwise unoccupied USB bus caught in an unlikely series of moments of perfect timing. Keyboards advertising "1KHz" - even if/when they live up to this spec - are not realistically polled any faster than "slow" USB keyboards, and may in fact perform quite poorly (vs their PS/2 counterparts) when sharing a bus crowded by many USB devices (or even if caught sharing with just a single USB device which rudely hogs the bus).

2) PS/2 "full" NKRO is functionally limited to 8KRO, or even less on the vast majority of 8051 (compatible/emulated) keyboard controller parts. There's an "8051" in every keyboard and (somewhere in the SuperI/O chip) on every PC/AT-compatible mobo, and the slower one bottlenecks actual keyboard throughput. Kind of a moot point on most keyboards anyhow, since engineering tradeoffs and cheapest-possible matrix layouts means they're functionally limited to usually 2KRO or 3KRO (even the higher-end gaming keyboards which advertise "anti-ghosting" and such stuff are usually just cheapy 2-3KRO circuitry with WASD-optimized clusters). Another limitation imposed on PS/2 NKRO bandwidth is the theoretical limit of 112.5wpm max typing speed, while no actual limit is (theoretically) imposed by USB specs - although, again, the same cheap keyboards/parts/thinking would probably dictate much lower thresholds.

The wiki at geekhack.org is pretty definitive, although I personally find the forum leaders and their "keyboard science" are really more like stuffy "keyboard dogma". Still, if ye can't say anything nice ... then come to TBCS!