• Please review our updated Terms and Rules here

Does PCI SuperIO/MultiIO board exist?

NobodyIsHere

Veteran Member
Joined
Dec 21, 2006
Messages
2,410
Hi! Does anyone know of a PCI multi IO or PCI Super IO board? I've seen various serial, parallel, USB, Firewire, etc PCI boards but that's not what I am seeking.

I would like to add legacy ports to recent PC with limited expandability and few of the "old fashion" IO ports. What I am looking for is a PCI board with a Super IO or Multi IO chip to provide a complete set of ports such as serial, parallel, USB, floppy, IDE, PS/2 keyboard, PS/2 mouse, etc.

Essentially a Winbond or VIA SuperIO chip on a PCI board with ports. Is there such a thing? I haven't seen one if there is one available.

Thanks and have a nice day!

Andrew Lynch
 
I don't believe so. You can certainly find ones to add Parallel and Serial together. USB is not a legacy port. There is no PCI floppy controller - trust me, I've looked for years. IDE controllers can be had for PCI, but not combined with anything. PS/2 ports I am unsure of, but I haven't seen it. You can convert from USB though, so almost all of the functions you seek can be had with separate parts. If you find some epic card that does all of it shoot me a PM, 'cuz I'll want one too, lol.
 
I didn't include the catweasel because it's rather.. different.. not what you'd normally refer to when talking about a floppy controller.
 
Hi! Does anyone know of a PCI multi IO or PCI Super IO board? I've seen various serial, parallel, USB, Firewire, etc PCI boards but that's not what I am seeking.

I would like to add legacy ports to recent PC with limited expandability and few of the "old fashion" IO ports. What I am looking for is a PCI board with a Super IO or Multi IO chip to provide a complete set of ports such as serial, parallel, USB, floppy, IDE, PS/2 keyboard, PS/2 mouse, etc.

Essentially a Winbond or VIA SuperIO chip on a PCI board with ports. Is there such a thing? I haven't seen one if there is one available.
Hey, Andrew - did you ever have any luck with this?

I came across this old thread because I was in search of a PCI controller for a ps/2 keyboard. I know that there are USB -> PS/2 keyboard convertors out there, but these seem to be hit and miss, and from my limited reading on geekhack.org, the ones that do work well only pass through scan codes for the 104 keys, not the extended keys found on terminal or POS keyboards (like my Cherry MX-8000 I use at work). I was thinking that it should be relatively simple to design an PCI board that setup a ps/2 port using a standard keyboard encoder, essentially guaranteeing that the clicky terminal and POS boards that many of us use would be fully usable on current machines.

And then I saw this post. It made me start thinking about other legacy ports we've lost - Parallel, Serial (if it's not gone already, it soon will be - and USB to serial adaptors are hit and miss for me), to a lesser degree, floppy..

Does anyone else see a need for this stuff? PCI ports are already going the way of the ISA bus in the late 1990's... they're already being phased out for PCI-E. I've absolutely no idea of the design spec for this stuff, but how difficult would it be to toss a serial port, parallel port, floppy port, and ps/2 keyboard and mouse port on a card? The parallel, floppy, and serial ports could (would?) use headers, while the mouse and keyboard would get external access ports. Thinking out loud here, is there a single multi-i/o chip that could do this, or would such a card need multiple controller chips with data i/o and power being supplied from the edge connectors?

I don't really mean to introduce a community project here, as I'm certainly not able to design such a beast... but I'm hoping that someone else might see the need and jump in - I'm good for monetary support/built/testing :) What do you guys think? Is such a beast possible?
 
Track down a board with the legacy ports, heck Gigabyte makes a z68 chipset board with PS/2, parallel, and serial ports on the back panel, right next to the USB 3.0 ports!
 
We all have boatloads of old boards - how does that future-proof us to allow us to use some of our (semi)modern PC-based equipment on machines built in another few years?

Gigabyte makes a board now that has these headers... but how long will that last? Most mainstream boards already have lost the PS/2 connectors in favor of USB. Parallel died a long time ago, as did floppy diskettes. Serial is almost gone too. In another 2-3 years, it's all a memory...

I know that it's possible to mod many keyboards to USB using any number of methods, some of which require custom logic boards and custom firmware, some of which just use gutted usb->ps2 convertors. My point is that for many terminal/POS keyboards, that method doesn't work - only 104 keys of the say, 122 keys have scancodes that are passed.

Why? All the keys on these terminal/POS keyboards work perfectly with any ol' computer with a PS/2 port! The only solution I can come up with is that they all go back to legacy chipsets embedded in the multi-io chips of modern PCs - except that the ports that we plug them into are disappearing, and FAST. Most of the multi-IO chips I've seen use LBC or some such, which requires that they exist directly on the motherboard. So if the motherboard doesn't have headers for these legacy connections, then we'll never be able to use them even if the PC supports them.

I'm looking (hoping) for a way to ensure that when I get a new PC in a year or two, I'll be able to hook up my Cherry G80 or my Model M and go to town like I've been doing for years, using all available keys past the 101/104 mark, and without having to invest in a new keyboard. I'd rather continue with the ones that I've used for years, like, and don't want to let go of. And well, if its easy enough, having Serial, Parallel, and Floppy embedded on this fictitious card would be nice and fill up that extra PCB space - and if there's a multi-IO chip around that does all of this and can fit on a card and be wired to the PCI or PCI-E bus, then solved! (once a design gets worked out, I mean).

I mean, it WOULD be nice to be able to use some old serial-based or parallel-based equipment on our daily driver machines without having to pull out our tweener systems, wouldnt it?
 
Last edited:
It would be a straight forward project to stick a super I/O chip on a PCI card with either a hard PLX bridge or PLD acting as a bridge. I have an ISA card I built with one on it (2S/1P/1G/FDD/PS2 K/M/MPU-401). The problem I ran into was finding the darned Super I/O chips. Digikey carries only one and it's $16. There are mucho mucho grande quantity on China IC mart though. Trick is finding a surplus holder that will part with only a couple dozen tops and of course paying only with a pre-loaded cash card!

Edit: Ah yes PS/2 Model M's. When us humans kill each other off and destroy the world, roaches will inherit the Earth. In billions of years when the roaches evolve back into intelligent 10 fingered beings, at least they will have Model M's to still use.
 
Andy:

If you run into a dead on the multi card you may want to consider a Gigabyte GA-790FXTA-UD5 (rev. 1.0). It has everything you are looking for, including native PS/2, IDE, and floppy. It'll run any just about any AMD chip that you're familiar with. It was discontinued laste this summer but can still be had for less than $200 new. There's one on ebay now for under $100 http://www.ebay.com/sch/i.html?_nkw....0)&_osacat=58058&_trksid=p3286.c0.m270.l1313

What I like is the abilty to selectively boot into a IDE DOS setup and having access to a legacy floppy and CDROM. You can disable the onboard audio and stick your favorite PCI audio card in there if you wish. There is no onboard video so you can go PCI or PCIe. I have a great time running DOS 7 and old games on this box. Good luck with your project.
 
If I have a real need to run a particular ISA board and the goblins have stolen all of my ISA-equipped motherboards, I can always take a relatively modern microcontroller and fabricate my own ISA bus interface and emulate whatever I need to interface to it. That doesn't bother me much. What does bother me is the number of vanishing drivers for these things--and datasheets for old chips.

As far as the PS/2 or AT interface keyboards, there are several microntrollers with PS/2 and USB interfaces on them. All that you need is a little bit of code.

My PAL samples were run from a Neoware thin client booted to DOS from a USB floppy. Drive C: was a USB pen drive.
 
As far as the PS/2 or AT interface keyboards, there are several microntrollers with PS/2 and USB interfaces on them. All that you need is a little bit of code.
Curious then - from my admittedly limited understanding of PS/2 to USB interfaces, whether they're implemented with a PIC, ATMega, or Teeny processor, one must still supply the scancode conversions to pass through the USB interface. Often this is done using a 16x8 grid, covering the standard 104 keys. That works great for most keyboards... My question is this: how do the existing keyboard controllers handle the scenario of an Model M being connected, then pulling that and plugging in a Cherry POS with 124 keys, then pulling that and plugging in an IBM Model F terminal with 114 (or 122? forget) keys... etc. We're not having to flash the keyboard microcontroller with a new firmware with updated scan codes for each one, and not all of these keyboards have on-board controllers like many of the Cherry POS models... Why do we need to supply such information with the PS/2->USB convertors? Why can't one just use an existing controller (even going back to the Intel 8041/8042 if need be) to implement a true plug-and-play scenario like we already have by default with the on-board PS/2 keyboard controllers?

Hopefully I've explained myself well enough that you guys see where I'm coming from. I spent a few hours last night researching a few of the datasheets, trying to make heads or tails of some of this, but I very quickly saw that I was out of my element - I was coming up with FAR more questions than answers!
 
I think I follow you, but feel free to chime in if I've got the wrong idea.

The biggest reason is that the USB HID spec was designed for the ground up, not just for PC keyboards. But it was never simple. Even on a legacy PC, the PCs keyboard controller translates AT scan codes to XT ones by default, though that can be changed. The HID spec calls for 16 bit keycodes, so the possibilities are greatly enhanced.

So USB is a step forward, just as the AT keyboard protocol was a step forward from the XT one. It's a cross we all have to bear.
 
Thanks for the explanation, Chuck - much appreciated.

So this makes me wonder... C64, for instance, uses an 8x8 grid array to read the keyboard scan codes. Model M/104-key uses a 16x8 grid array to read the scan codes. When programming the controller for generic X-keys, so that any keyboard plugged with a compatible controller/logic just works regardless of the number of keys (i.e. 83-key, tenkeyless, 101-key, 104-key, POS/Terminal with > 104-key)... how does one setup the grid? Are the scan codes unique to each keyboard/logic board, or to the position of the keys themselves? - this is the concept that I'm having trouble understanding. Though if I can source a multi i/o from an old 16-bit card, I haven't given up on the idea of having a PCI/PCI-E expansion for this stuff too (I'd love to lose the tweener systems if I could/needed)
 
If you're talking about a physical grid, I don't think it was ever that simple. There's a lot of special-casing done and the keys themselves are rarely arranged in a logical way. Look at the wiring side of a PC keyboard--traces run all over the place.
 
Back
Top