Talk:IBM PS/2

Important things are missing
You should also mention the later models like Server 95, M77, P70, Laptops, and so on. So everybody thinks that the 386 is the upper end of the PS/2 models. And a picture of some computers is missing!
 * I'm working on it. :) - Thatdog 20:56, 7 August 2005 (UTC)

Datasheets for a 2 common ICs used in PS/2 mice, I'll link them shortly. Also the bits I've added on coms will be in much greater detail than 'computer-engineering.org'. At least it will be presented more clearly :D - Bsodmike 07:52, 25 November 2005 (UTC)

Also, there is no box at the bottom saying what model came before and what came after. -Anon —Preceding unsigned comment added by 128.84.134.19 (talk) 16:42, 3 September 2009 (UTC)

Contrast to USB
PS/2 mouse/keyboard interface is not powered, is it? What other distinctions are there with USB? ---Ransom --208.25.0.2 19:13, 12 December 2005 (UTC)


 * PS/2 keyboard/mouse most definately is powered, but thats probablly about the only similarity it has with usb. ps/2 is a very simple slow clock+data interface, USB is a high speed interface with differential transmission and complex signaling. Plugwash 19:16, 12 December 2005 (UTC)


 * PS/2 is also not safely hotpluggable. That is, in many instances you can hotplug PS/2 devices, however there is reportedly a risk of shorting components on some motherboards (depending on which pins of the connector happen to connect to the socket first). USB is, of course, designed for hotplugging. --Bk0 (Talk) 20:00, 12 December 2005 (UTC)


 * do you have any evidence of that? from my knowlage of how ps/2 works i can't see how hotplugging it could cause any damage. Plugwash 20:30, 12 December 2005 (UTC)


 * From http://www.computer-engineering.org/ps2protocol/ ("The PS/2 Mouse/Keyboard Protocol"):


 * The keyboard or mouse should not draw more than 275 mA from the host and care must be taken to avoid transient surges. Such surges can be caused by "hot-plugging" a keyboard/mouse (ie, connect/disconnect the device while the computer's power is on.)  Older motherboards had a surface-mounted fuse protecting the keyboard and mouse ports.  When this fuse blew, the motherboard was useless to the consumer, and non-fixable to the average technician.  Most newer motherboards use auto-reset "Poly" fuses that go a long way to remedy this problem.  However, this is not a standard and there's still plenty of older motherboards in use.  Therefore, I recommend against hot-plugging a PS/2 mouse or keyboard.


 * --Bk0 (Talk) 21:25, 12 December 2005 (UTC)
 * hmm i can't see how hotplugging would create any more of a surge than powering up the whole system at once. Plugwash 21:09, 13 December 2005 (UTC)
 * I've never had a problem with my modern machines, but I've heard anecdotally from people (repair techs and such) that at least in the past customers blowing fuses due to this was not an uncommon problem. --Bk0 (Talk) 00:07, 14 December 2005 (UTC)

Serial?
Is a PS/2 port a type of serial ports? 13:17, 7 February 2006 (UTC)

Simms vs Dimms
Someone noted in a comment "hmm i seem to remember it being most of the pentium era but our dimm article and pentium article have very little in the way of dates" (I don't feel like digging up the history to see who) ... and there it really depends on how you define the "Pentium era" - if it's only to the release of the Pentium Pro in 1995, then you're absolutely right.

There were plenty of Pentium systems selling through to sometime in 1998 when the Celeron finally was cheap enough to kill it (although arguably it really took the 300A - a bit later - and the various Super Socket 7 chips to really kill it), which was more what that might have been about.

Fast page mode DIMMs were really rare outside of the Macintosh world (which was late to adopt EDO). EDO memory was more readily available in SIMMs and mostly used in DIMMs in higher end systems -- I remember seeing a fair number of motherboards that had both sorts of sockets. It was probably the switch to SDRAM which finally pushed things over the edge to all DIMMs; there may well have been SDRAM in 72-pin SIMMs, but I don't recall ever seeing it or an ad for it. Nate 08:56, 6 March 2006 (UTC)


 * 30 Pin SIPP (Single Inline Pin Package) RAM modules were used for an extremely short time before the SIMM was developed. The later SIPPs were simply 30 pin SIMMs with pins soldered onto the contacts. As for Macintosh, they were also late to the SDRAM DIMM revolution. —Preceding unsigned comment added by Bizzybody (talk • contribs) 05:37, 16 April 2008 (UTC)


 * [[Image:SIPP.jpg|300px|thumb|SIPPs from the Commons]] SIPPs plugged in with a long row of pins in a straight line, and were a very poor mechanical design since the individual pins could get misaligned, bend, and then break off after repeated restraightening.


 * Structurally they were no different from SIMMs other than the fact that the SIMM card-edge contact design is much more durable and robust that a row of fragile pins. DMahalko (talk) 01:31, 17 August 2008 (UTC)


 * Additional personal contribution: The Socket7 era was where SIMMs gave way to DIMMs... and you could, technically speaking, plug a late-model (i.e. Socket 7 but still P54C) P5 chip into pretty much any S7 mobo, even the later ones would support a plain, non-MMX P166 or at least a P200 (at 66 bus x 2.5/3.0 multiplier) even if they were spiritually intended for use with much faster AMD K6/2s and 3s (at, say, 350mhz, a small tweak of bus speed and multi away). So you could say it was a "pentium era" change. However the P5 proper (as in, the P54C, non-MMX; otherwise you could argue that the Pentium never went away thanks to the PII thru P4, and the continuing use of the name for mobile and budget CPUs) launched into a world of 72-pin SIMMs and left it around the time the change was underway but not yet done.


 * I have at least one Pentium motherboard (possibly Sock5?) with 4x SIMM sockets which only accepts up to 128MB max, and a P5-compatible S7 which has just 2 (3?) DIMMs and a ceiling of 256 (384?) MB... plus a rather particular midrange model that has BOTH (2 SIMM and 1 DIMM, IIRC, but particular magic was needed to make both work at once so it wasn't exactly the most cheaply upgradeable board in the world). Would ya like some pics? :)


 * Also I am in dispute about the PS/2 system leaping right from SIPPs to 72-pin SIMMs. I'm baffled by how the staunchly 16-bit 8086 or 80286 models could be made compatible - or gain any advantage from using - expensive 32-bit memories, or why the 386SX models would use them unless IBM came up with some kind of bizarre hybrid model that had a 32-bit memory bus on one side and a 16-bit expansion bus on the other. Would they not have introduced the 30-pin, 8-bit SIMMs with the 16-bit models (as this was around the time that they entered the market anyway; e.g. certain 286-based clones (one of which I own), the Atari STe, laser printers, Amiga memory upgrade boards with room for further expansion etc), and then upgraded, quite possibly through their own innovation, to 72-pin SIMMs for the 386DX and 486 models? I have yet to see ANY pre-486 board from any manufacturer that uses 72-pin memory instead of 30-pin, and have a couple of 486 boards lurking around - one of which is a later model with PCI and 72-pin slots, the other an SX25 with ISA and 30-pin memory. Tweaking everything on the older board to the fastest possible timings revealed that even a fully populated bank of 8x 30-pin modules (therefore, with more capacitance and latency than an older 2- or 4-socket system) was still perfectly capable of falling into absolute, fully reliable lockstep with a "fast" CPU on a 25mhz, 32-bit bus (in fact, they held up much better than the video card or hard disk interface did - I ultimately had to back the bus timings off a little when non-RAM-related errors occurred with both), let alone a slower 16-bit one.


 * IBM might have wanted to save some space with the 386DX models by making the early upgrade, but what was in it for them to use them on the SX (typically running at 16mhz, sooooometimes as high as 20 or 25), presumably with an extra shifter chip or three interposed to de/multiplex 16 bit signals on one side to 32 bit on the other, tens of millions of times per second when the existing, cheaper to manufacture standard would work just as well? 87.115.98.16 (talk) 12:28, 4 August 2012 (UTC)


 * (Oh yeah ... and there's pictures of PS/2 Mobos about... some of which, including a Model 70, quite clearly seem to have 30-pin slots. The caption written by that machine's owner says it has 1MB onboard - and there are 4 chips. I know that 256kb 30-pins existed, and so did 1MB 72-pins (briefly)... but, 256KB 72-pin? Never seen/heard of them before...) 87.115.98.16 (talk) 13:46, 4 August 2012 (UTC)


 * I believe later 32-bit PS/2 models used 72-bit SIMMs, possibly starting with the Model 70 (see here - 3x 72-pin slots, 6 MB max). Other models used 30-pin SIMMs (e.g. Model 25-286) or proprietary modules (e.g Model 80). This website has a lot of PS/2-related information. Regards, Letdorf (talk) 20:36, 5 August 2012 (UTC).

Models
Is anyone good with making wikicharts/tables? The section with all the model numbers would be much easier to comprehend if it was in table form instead of a bunch of disjointed sentences. /Blaxthos 23:30, 11 December 2006 (UTC)

PS/2 85xx series vs the 95xx series?
There was basically two generations of PS/2 models. The first generation 85xx models and then the 95xx models that were known as "Premium" models. I believe the IBM logo changed colour, and the 95 series used IML partitions as compared to the 85 series which used floppy-based reference disks. Also, the premium models used SCSI and IDE, were more modern looking, and were supported at least through 1994. I'll have to dig through my specs, but I think the two generations are significant enough to mention. There was also a IBM PC 700 series desktop (6885 or 6886) that had both PCI and MCA slots as well as a few IBM PC servers (720 or 701?) that had MCA slots, but were not listed as part of the PS/2 family.

NTD 28dec06 03:02EST


 * Well...Yes. [| History of the PS/2]
 * IBM 8641-MZV "Server 520" was a dual Pentium-133 with MCA and PCI.
 * Thanks! 71.193.2.115 (talk) 12:34, 15 May 2008 (UTC)

Plug and Play BIOS?
Ummm, no. Not when to make any hardware change you had to first boot the computer with a special floppy (hardware Refrence disk_ just to access the BIOS setup, but then had to use a special reference disk for any hardware you added. You couldn't just pull out hardware either. The system had to be booted with the setup disk- I don;t recall if that had to be done before or after removing hardware. Even moving a card to a different slot required booting with the setup disk.

That is NOT Plug and Play, it's substituting floppy disks for jumpers and switches. It's also why the MCA Mafia (The MCA mafia are alive and real, they have jour disks ) has the large archive of reference disk downloads.\ 71.193.2.115 (talk) 12:37, 15 May 2008 (UTC)
 * Plug n Play as we know it today did not really exist before the Microchannel systems were developed, so while it wouldn't pass for the modern form it did fulfill its function of eliminating hardware DMA/IRQ/Address jumpers so that the end-user would not have to deal with that arcanery. The PS/2 was meant as a leap forward over the awful AT ISA slot system of fiddling with jumpers on the cards.


 * It's not really fair to harp on the Reference Disk system configuration, because most early PCs did not have a built-in BIOS menu system. ROM memory was expensive and limited in capacity so the early PCs could not provide such capability. I remember having to run configuration programs on disk for other non-IBM 386 and 486 motherboards.


 * Actually that is where the hidden "Configuration partition" started out. Without enough ROM space for configuration tools, instead the hard drive got a dedicated weird partition to store all the system configuration data.


 * DMahalko (talk) 01:25, 17 August 2008 (UTC)

Interface on the bizarre PS/2 hard drive?
This is part of a very long and difficult debate. The PS/2 models 30 (286), model 50 (386), and model 70 (486) had an option for a special IBM hard drive that just plugged straight into the system without a separate power cable or any configuration jumpers.

This drive was not SCSI, and apparently not an IDE, MFM, RLL hard drive, because it has nothing like any of those connectors. It's just one huge card-edge connector spanning the entire width of the back of the 3.5 inch drive. I believe the model 30 used a huge ribbon cable with four ferrite shielding cores to connect it to the main board, but the model 70 had a "planar card" that the drive just slotted right into when the drive was installed into the system.

Some people have claimed it was an ESDI hard drive, but this does not appear to be correct either, since ESDI used similar cabling to the early MFM and RLL drives.

(I recently found some of my photos of my old PS/2 hardware I threw out years ago, and I'll see about posting one of the drive pictures.)

Any ideas? DMahalko (talk) 03:37, 17 August 2008 (UTC)

Aha, I finally found my old picture collection! (Click on the images for more info.)

In the PS/2 Model 55 SX:



In the PS/2 Model 70:



Overall it is a totally weird hard drive interface, and I've never seen anything like it. The closest modern equivalent is the all-in-one SCSI bus known as SCA, and this was not SCSI.

People claim it to be ESDI but it seems to me that this is just a random guess someone put down to attempt to identify it. I have never seen any hard evidence with IBM's official stamp on it declaring this to be ESDI. DMahalko (talk) 02:36, 21 September 2008 (UTC)


 * This is what IBM called the "Integrated Fixed Disk and Controller" (MCA ID DF9F) - an RLL hard disk with an on-board controller - analogous to IDE/ATA but using a subset of the 16-bit MCA bus on a 72-pin edge connector. According to this article, the BIOS identified this as an ESDI disk, even though, as you point out, the actual ESDI bus isn't used. However, the on-board controller was very similar to the "real" MCA ESDI controller (ID DDFF) used on other PS/2 models. Letdorf (talk) 23:43, 21 September 2008 (UTC).

Windows on PS/2
What is the newest version of Windows that one of these can run? My cousin gave hers to me. ShadowVulpix (talk) 20:55, 20 August 2010 (UTC)


 * You can install Windows 95 on a PS/2 model 70, but the major limitation is going to be disk space. Early versions of Win 95 without Internet Explorer can fit within about 40 megs of space. Later versions that include Internet Explorer will need a bigger drive to fit everything.


 * My above images of the Model 70 shows a SCSI card ... I bought that card specifically to use an external drive to install Win 95, because the internal IBM drive did not have enough room. DMahalko (talk) 22:14, 7 January 2011 (UTC)

PS/2 72-pin SIMMs... vs everyone else
Does anyone know what the technical difference was between PS/2 72-pin SIMMs and the 72-pin SIMMs used by everyone else? For some reason IBM SIMMs only worked in PS/2s and nothing else. Meanwhile clone SIMMs were not usable in the PS/2.

Since the card edge pinout was identical I would have to assume IBM assigned the pins differently. (Or was it the clones that assigned the pins differently?)

DMahalko (talk) 19:52, 13 January 2011 (UTC)

PS/2 modules and common 72-pin modules are pin compatible. PS/2 machines just are very picky about memory modules.

The modules must be:
 * 70ns module (80ns for older models)
 * 36-bit parity module
 * Non-EDO (Non-BEDO) module
 * equipped with Presence Detection
 * of limited capacity (depends on machine model, 2MB for Model 70... see "IBM Personal System/2 Model XX Technical Reference")

Or you can modify more generic modules [mcamafia.de].

Tomáš Slavotínek (talk) 21:12, 2 March 2011 (UTC)

I had an IBM PS/1 386sx with 72 pin simms. The 386sx bus is 16 bit wide! — Preceding unsigned comment added by 82.155.192.143 (talk) 14:46, 29 January 2016 (UTC)

Micro Chanel Architecture
I think that the assertion the Micro Channel Architecture was "closed" needs challenging.

No only was the architecture fully documented (in fact better than the previous PC/PC AT architecture) but to prove the point, other manufacturers produced "clones" based on it.

On that basis how can it be closed?

BTW, whether or not one owed IBM royalties for producing clones is neither there nor there. That was the case with the previous PC/PC AT based boxes and I can assure readers (I worked for IBM OEM at one stage) that royalties on the previous generations WERE paid! — Preceding unsigned comment added by 86.134.40.250 (talk) 13:03, 27 September 2011 (UTC)


 * If IBM required third parties to pay royalties to make MCA hardware, then it was certainly not "open," whether documented fully or not. Jeh (talk) 01:34, 30 May 2018 (UTC)


 * I feel kinda silly answering this more than a decade later, but it was during the introduction of these PS/2s that "Open Standards" was even defined in the computer industry. For example "ISA", was derived from the IBM PC AT-bus, was not an official "standard", certainly not one external to the company, until Compaq, with a Gang of Nine named the original AT bus, ISA (Industry Standard Architecture) and simultaneously even proposed its successor, EISA —in opposition to the licensing of IBM's MCA. "Prior to developing EISA Compaq had invested significant resources into reverse engineering MCA, but its executives correctly calculated that the $80 billion already spent by corporations on IBM-compatible technology would make it difficult for even IBM to force manufacturers to adopt the new MCA design. Instead of cloning MCA, Compaq formed an alliance with Hewlett Packard and seven other major manufacturers, known collectively as the "Gang of Nine", to develop EISA." -from Compaq article.

It should be very obvious in hindsight that MCA was closed, the only reason someone should be confused about this, is that Open Architecture was not yet defined precisely as these MCA machines were coming out. One could not be more wrong about "whether or not one owed IBM royalties for producing clones is neither there nor there". Interestingly, as a side-note, IBM never sued any clonemakers for BIOS copying, they settled with such companies. It was Apple v. Franklin (1984) which established that software/firmware, or indeed any "non-human readable" code was copyright-able. This is the first legal establishment of US copyright law in the industry, perhaps a further reason why "openness", especially on IBM's part pre-'84, might be confused, also, a few years later IBM jumped on the bandwagon and started touting "Open Standards" in all sorts of marketing and advertisements, which again might also have been confusing for the OP. It was a complete turnaround after much FUD to the contrary from IBM. 75.71.166.197 (talk) 21:09, 20 February 2022 (UTC)

Gross Bias
The PS/2's controversial hardware design was tied to a marketing strategy that was similarly unsuccessful. During the 1980s, IBM's advertising of the original PC and its other product lines had frequently used the likeness of Charlie Chaplin. For the PS/2, however, IBM augmented this character with a notorious jingle that seemed more suitable for a low-end consumer product than a business-class computing platform

Gross, someone make this not sound like a personal essay, please. 71.254.122.149 (talk) 00:15, 16 January 2012 (UTC)


 * And why "controversial"? 31.52.254.10 (talk) 19:24, 29 May 2018 (UTC)

Pricing information
Their price is very important, but misses from the article. 37.192.250.101 (talk) 05:24, 11 October 2014 (UTC)

Third generation?
So, we have "The Personal System/2 or PS/2 was IBM's third generation of personal computers..." - yet nowhere in this article, or IBM Personal Computer, do we see any discussion of any concept of First and Second generation IBM personal computer. If the AT can be argued to be the second generation, then let's make that argument explicit - or lose this statement. Snori (talk)
 * Agreed. I have changed it with sources from the time describing it as IBM's second-generation PC. DigitalIceAge (talk) 03:06, 21 February 2022 (UTC)

BIOS source code
The article says that IBM did not publish the BIOS source code. I can't be sure about PS/2 BIOS source code but I have a copy of the IBM PC XT/286 manual that has the BIOS source code. Perhaps the article means the Advanced BIOS source code was not published; that makes sense. Sam Tomato (talk) 19:58, 8 October 2017 (UTC)

Also, saying "publish BIOS entry points" is likely not accurate since users of the BIOS use software interrupts to use it, not entry points. Sam Tomato (talk) 19:58, 8 October 2017 (UTC)

Definitely an 8086 in the earlier models, not 8088?
Is that true? I know there were some oddball less-than-100% clones that also used the 8086, and even 80186, but previous IBMs (and most full compatibles) jumped straight from the 8088 to the 80286. Does that mean that even the lowest end PS/2s had a full 16-bit architecture? What speed did they actually run at (it seems missing from the article)? And what did that mean for the performance and compatibility vs an otherwise similarly clocked 286 model (and indeed, over a typical XT), other matters like the lower end video card (because of smaller processor address space?) aside? 80.189.129.155 (talk) 01:34, 14 April 2019 (UTC)

Yes, the PS/2 Model 25 and the original Model 30* had an 8086 processor, running at 8mhz. http://www.bitsavers.org/pdf/ibm/pc/ps2/PS2_Model_25_Technical_Reference_Jun87.pdf and per the same document they had a full 16 bit data bus. (* There was a later PS 2 Model 30 (286)) They were NOT, however, a full "16 bit architecture" in the sense most PC users would have used at the time -- the card slots were still 8-bit PC/XT-style slots and not the full 16 bit ISA/AT-style slots. I'm not sure from that document whether the MCGA's video RAM was a full 16 bit but given the overall performance of the machine, I suspect it was only the XT-style slots that were 8-bit. From memory, for non-graphics-intensive applications, the performance was roughly comparable to contemporary 8mhz and 10mhz "turbo XT" systems and noticeably better than original-generation 4.77mhz XT systems, while being noticeably slower than the 10-12mhz contemporary 286 systems (including the PS/2 model 50.) I never really used the 6/8mhz AT-class first generation 286 systems. For graphics-intensive applications, it was very unusual to have VGA cards in XT slots, so it's very hard to compare. I suspect that with the number of contemporary computer magazines archived and freely available on the web, someone could probably find some competitive benchmarks. Nate (talk) 23:35, 7 October 2019 (UTC)

hardware interrupt handling
The original IBM PC loses (fails to respond to) some interrupt requests from hardware. I think because interrupts were edge-sensitive instead of level-sensitive. Whatever, the ps/2 fixed that problem. This was one innovation that I think has been lost. I had discovered the problem for myself in practice, and had worked around it by using a backup timeout to trigger polling for the lost interrupt. When Microsoft introduced NT they provided access to their driver authors and I had a talk with the engineer writing the serial port driver, who told me that his driver used the same approach to recover lost interrupts. Thus I imagine the lost interrupt problem has never been fixed, which, if true, is one of the sad things in pc history. I am not sure though, because I do not build that kind of stuff anymore. --AJim (talk) 15:11, 3 November 2021 (UTC)

more keyboard information on "layout" ? other quibbles with the introduction
In the introduction, as it is, it reads, "Many of the PS/2's innovations, such as the 16550 UART (serial port), 1440 KB 3.5-inch floppy disk format, Model M keyboard layout,..." This leaves me confused about some things, I believe the writer's intention is to say the layout of the Model M was more important than the keyboard itself, but... The Modem M was introduced before and separately from the PS/2 lineup, so it's inclusion here is iffy. The keyboard layout really has little to do with PS/2 Computers, themselves, although feel free to find a published source saying the contrary.

I think the PS/2 keyboard PORT is much more notable, in specific reference to the PS/2 computers. And also, for instance, how many of the PS/2s actually included an Model M keyboard? (The PS/2 port article also lacks this information) So, I'm not sure how a reference to the "layout" came about. lol I've just talked myself into removing the reference to "Model-M layout". If someone wants to reference the "layout" in saying that the PS/2 lineup in particular contributed to the popularization of keyboard layout somehow, please reference a third-party source. I'm not going to search for such an external reference for the PS/2 port, because the article link serves as a reference, the references in the PS/2 port article should suffice. I'm noticing more and more decade+ old "citation needed" in Wikipedia articles where a citation isn't actually needed (For example, if the sentence itself mentions a third party source in sufficient detail, a footnote reference is simply redundant). Also a reference at the end of a paragraph may serve for several sentences, not just a single sentence). 75.71.166.197 (talk) 20:37, 20 February 2022 (UTC)

keyboards available at launch?
Would be nice to have some info on keyboards, layouts, and picture of what was bundled. 2A02:168:F609:1:284F:5687:CC7:7DF7 (talk) 16:18, 27 January 2024 (UTC)