Recently I came across a minor mystery—the model numbers of the original IBM PC. For such a pivotal product, there is remarkably little detailed original information from the early days.
When IBM started selling the PC, it used two methods to identify various items. The vast majority of peripherals and accessories used a seven-digit part number; for example the original CGA was part number 1504910.
The system unit itself was Type 5150 (the PC; 5160 was the XT), and the type designation was extended with a three-digit model number, such as 5150-013. The same scheme was used for “big ticket” peripherals that were not necessarily tied to the PC. For example the original monochrome monitor was designated as 5151-001 and the printer as 5152-001 (the color monitor was Type 5153 and so on).
For monitors and printers, the three-digit model number wasn’t all that meaningful as there often was only one model. But for the 5150 system unit, there were lots of different models.
Now, the problem is that there appears to be no official documentation whatsoever from the old PC days that would list the model numbers with clear specifications.
A major problem hindering research is that there are no mentions of the IBM PC 5150 in the IBM announcements archive prior to 1983. The announcement letters are usually a very reliable source of information as to what IBM released when, but for about the first year and a half of the IBM PC’s life, there’s just nothing.
The IBM PC was well publicized when it was released. But if one looks at the original press release from August 12, 1981, there is a reasonably detailed technical description of the IBM PC… yet the 5150 type is not mentioned at all, let alone any model numbers. Or the part numbers of any accessories.
The press release presents two presumed-to-be typical configurations (16K RAM with cassette support only, and 64K RAM with a floppy drive) but mentions no part numbers and notes that the PC “can be tailored to fit the user’s needs”, strongly implying that other configurations were available.
Sure, there’s a list of IBM PC models on Wikipedia… but the table doesn’t even always match the provided references, and the references are woefully incomplete anyway because, again, there’s apparently no official data from the 1981-1982 period.
Officially Announced IBM 5150 Models
Let’s start with an area where we can be on firm footing, which is IBM PC 5150 models that are listed in the IBM announcement archive.
March 1983
The oldest reference I could find is announcement letter 183-030 from March 8, 1983, which lists several new IBM PC models. The announcement coincided with the introduction of the IBM 3270 Personal Computer Attachment, which used IBM PC 5150 machines with a special keyboard. To that end, IBM introduced 5150 models X14, X64, and X74, which were “equivalent of Models 14, 64, and 74, respectively” minus a keyboard.
This establishes the existence of models 014, 064, and 074 (sometimes written with and sometimes without the leading zero) in addition to the newly introduced X14, X64, and X74. Sadly the 183-030 announcement letter presumes that the reader is already familiar with the technical specifications of models 014, 064, and 074, and provides no further detail.
Note that March 8, 1983 was also the date the IBM PC/XT (Type 5160 Model 087) was announced.
March/April 1983
The next oldest source of information is announcement letter 183-068 from April 28, 1983. This announcement establishes new models 114, 164, and 174 which “replace the IBM Personal Computer Models 14, 64, and 74, which are withdrawn from marketing”. In addition, a new model 104 “replaces the Model 1, which was withdrawn from marketing earlier”. Note that the 183-068 letter says that the new models were in fact announced on March 11, 1983.
Under the hood, the new PC models used a newer system board with 64K-256K memory capacity (and using 64Kbit memory modules). The older 5150 PCs used a 16K-64K board which used 16Kbit memory modules. It is apparent that when the PC/XT was introduced, IBM wanted both the 5150 and 5160 systems to use the same memory modules and obsoleted the original PC system board with limited memory capacity.
For the newly introduced models 104, 114, 164, and 174 we do have the technical data and also pricing. Not only that, but the 183-068 announcement letter is an excellent source of information, as it also details the previously introduced models X16, X64, and X74:
- Model 104: 64K RAM, no drives, $1,355
- Model 114: 64K RAM, one single-sided 180K drive, $1,864
- Model 164: 64K RAM, one double-sided 360K drive, $2,104
- Model 174: 64K RAM, two double-sided 360K drives, $2,633
- Model X14: 64K RAM, one single-sided 180K drive, $ 1,594
- Model X64: 64K RAM, one double-sided 360K drive, $1,834
- Model X74: 64K RAM, two double-sided 360K drives, $2,363
The obsoleted models 001, 014, 064, and 074 are not described in any detail, but combined with the information in the 183-030 announcement letter, we can deduce the specifications of the following models:
- Model 014: 64K RAM, one single-sided 180K drive
- Model 064: 64K RAM, one double-sided 360K drive
- Model 074: 64K RAM, two double-sided 360K drives
All in all, the 183-068 announcement letter defines the IBM PC 5150 lineup as of March 1983.
Note that looking at the model numbers, it is quite apparent that the middle digit of the model number probably has something to do with the floppy drives.
June 1984
The next big change came in announcement letter 184-077 on June 19, 1984. 5150 models 114, 164, and 174 were withdrawn from marketing, and with them also the single-sided 180K floppy drive (used in model 114).
The new models were:
- Model 166: 256K RAM, one double-sided 360K drive, $1,995
- Model 176: 256K RAM, two double-sided 360K drives, $2,420
It is worth mentioning that the new model 166 with 256K RAM cost $1,955 (June 1984), while the predecessor model 164 with only 64K RAM had originally cost $2,104 (April 1983).
Also note that the last digit of the model number probably had something to do with the installed RAM size.
Closely following was announcement letter 184-082 on June 22, 1984. This updated the keyboardless ‘X’ models analogous to the standard system units. Models X14, X64, and X74 were dropped, and replaced with the following:
- Model X66: 256K RAM, one double-sided 360K drive, $1,725
- Model X76: 256K RAM, two double-sided 360K drives, $2,150
December 1985
In announcement letter 385-154, models X66 and X76 were withdrawn from marketing. This was announced on December 3, 1985, with an effective date of February 28, 1986.
April 1987
Models 166 and 176 together with the most basic model 104 remained on offer until April 1987. They were withdrawn from marketing on the same date that the new PS/2 line of computers was announced, ranging from 8086 to 80386 based systems, a world away from the original 5150. The withdrawal was published in announcement letter 187-072 on April 2, 1987. The 5160 XT was retired on the same day.
Summary
Official IBM announcements clearly establish and document models 104, 114, 164, 174, 166, and 176. Additionally, keyboardless models X14, X64, X74, X66, and X76 are well documented. For these models, we have the specifications, introduction dates, and pricing.
Also mentioned are models 001, 014, 064, and 074, but only as withdrawn. For the latter three we can deduce the specifications as they were the same as models 114, 164, and 174 respectively, but using the older system board.
Vaguely Documented Models
Now let’s go back to the pre-1983 models. There were lots more early models, some mentioned in IBM announcements, some not, or they were mentioned in announcements that did not survive. But they were mentioned in IBM literature.
For example the IBM PS/2 and PC Product Reference from April 1987 additionally mentions PC 5150 models 013, 813, and 823 without providing any specifications whatsoever. One can only guess that the IBMers putting together the list in 1987 already had a hard time figuring out just what those models were.
There is one copy of an official announcement by the IBM Data Processing Division (not Entry Systems Division which made the PC!) from August 12, 1981, i.e. the initial PC announcement date. This announcement details the following two models:
- Model 813: 48K RAM, one single-sided 180K drive, serial adapter, $2,385
- Model 824: 64K RAM, two single-sided 180K drives, serial adapter, $3,045
Now this is quite fascinating as model 824 is not even mentioned in the later IBM reference materials, although 813 is. But before getting further, let’s clarify two possible points of confusion. The first pertains to floppy drives.
The single-sided floppy drive was designated as either 160K or 180K. It is the exact same drive, using the exact same floppies. While PC DOS 1.x formatted the disks as 160K, PC DOS 2.x could format them as 180K. Analogously double-sided drives can be designated as either 320K or 360K, but it is the same drive.
The other point is the PC system unit part numbers.
In IBM announcements and newer literature, the units were designated either as (for example) “Type 5150 Model 166” or as “5150-166”. In older materials from the pre-1983 era, the type and model may be concatenated; the type and model may be written as “5150013” rather than “5150-013”.
An IBM internal newsletter from April 1982 mentions that 48K system unit 5150013 (i.e. Type 5150 Model 013) will no longer be manufactured and the replacement is a 64K system unit 5150014 (i.e. model 014).
A May 1982 issue of the newsletter brings another valuable piece of information. On May 17, 1982 IBM announced system unit 5150064 (i.e. model 064) with a double-sided floppy drive.
Using the above information, we can add technical specifications of another model:
- Model 013: 48K RAM, one single-sided 180K drive
That’s in addition to the previously detailed models 014 and 064.
Model Numbering Scheme
It is near certain that IBM used a somewhat logical numbering scheme for the 5150 PC models. This is not my idea—I found it in a forum post. Putting together all available information, it is highly likely that the scheme is correct.
- First digit/letter has the following meaning
- 0: Original 16K-64K system board
- 1: Newer 64K-256K system board
- 8: Original 16K-64K system board, serial adapter added
- X: Newer 64K-256K system board, no keyboard
- The second digit designates the installed floppy drive(s)
- 0: No floppy drive
- 1: One single-sided 180K drive
- 2: Two single-sided 180K drives
- 6: One double-sided 360K drive
- 7: Two double-sided 360K drives
- The third digit indicates the installed RAM size
- 1: 16K
- 2: 32K
- 3: 48K
- 4: 64K
- 6: 256K
From the above it should be apparent that model 001 would have been the very minimal IBM PC, with no floppy drives and 16K RAM.
The 0xx and 8xx models are old models using a 16K-64K system board, available before March 1983. After March 1983, only models 1xx and Xxx were available, and only in certain combinations.
It is unclear how many of the plausible 0xx and 8xx models actually existed. In theory, the initial models could have been 00y, 01y, 02y, 80y, 81y, and 82y where y was 1, 2, 3, or 4. However, it is known that IBM did not build any 16K models with a floppy drive, hence model 001 did exist but 011 or 021 did not. The reason was that to boot from a floppy, the PC needed at least 32K RAM (since the boot sector was read to a memory location just below 32K).
It is likewise unclear if 0x2 or 8×2 models (with 32K RAM) actually existed. Similarly model 023 or 024 could have plausibly existed, as could 063 or 073, but did it?
On the other hand, the 8xx models seem somewhat pointless because they were just 0xx models with a serial adapter installed at the factory. It is possible that the 8xx models only existed to simplify ordering for certain customers.
Looking at surviving 5150 PCs may be less than helpful. For example, a customer might order a model 013 PC with 48K RAM and have the dealer install additional 16K RAM. This effectively turned the system into model 014. Likewise floppy drives could be added or replaced later, and several decades later it tends to be impossible to reliably tell what was installed when.
With this in mind, let’s put together a list of reasonably well documented pre-1983 PC models:
- Model 001: 16K RAM, no floppy drives (Sep ’81)
- Model 013: 48K RAM, one single-sided 180K drive (Sep ’81)
- Model 014: 64K RAM, one single-sided 180K drive (Sep ’81)
- Model 064: 64K RAM, one double-sided 360K drive (May ’82)
- Model 074: 64K RAM, two double-sided 360K drives (May ’82)
- Model 813: 48K RAM, one single-sided 180K drive, serial adapter (Sep ’81)
- Model 823: 48K RAM, two single-sided 180K drives, serial adapter (Sep ’81)
- Model 824: 64K RAM, two single-sided 180K drives, serial adapter (Sep ’81)
The list may well be incomplete. It almost certainly is, given the dearth of accurate and complete information. However, the technical specifications of individual models are very likely to be correct.
Interesting!
I’m surprised that the 5150 was in the catalogue all the way to 1987!
This might be a noob question that I ought to already know, but:
Was it possible to buy an 5160 XT without a hard disk, just to get a PC with more slots?
Also, was it possible to buy either a PC or XT with a hard disk but without any disk drives?
Interesting that they seem to never had offered a 128k option. Not that it would had made that much sense, but still.
Also, did anyone make third party motherboards (and/or cases) with the 5150 5 slot slot spacing?
If there never were any third party motherboards at least it’s likely that almost all 5150 have their original motherboards. For a 5160 it’s most likely that the motherboard and other cards were scrapped and replaced with a 286/386/486 or even a Pentium 1 motherboard at some point in time when they were considered relatively useless, vintage computing hadn’t taken off as a hobby and cases and power supplies were still relatively expensive. Thinking about this, this probably saved a bunch of Tandon full height drives from being scrapped as you could fit a half height 5.25″ hard disk or even a 3.5″ hard disk together with a 3.5″ floppy in one of the bays (at least using some minor metal working) and keep the Tandon drive, and use that with newer cards. That would allow someone upgrading to read/write existing 360k formatted disks. Anyone doing that upgrade would most likely not had been interested in 1.2M 5.25″ disks anyways.
I think your section header ‘March/April 1984’ should be ‘March/April 1983’ or just ‘April 1983’? I don’t see anything there referencing 1984.
For those who were never part of that world, it may also interest people to know that the IBM PC division kept up that numbering scheme (4-character “type” code, followed by 3-character “model” code) all the way up to when they were sold off to Lenovo in the mid-‘aughts. (It’s also possible that Lenovo has continued it to this day; I just haven’t kept up with Lenovo in recent years, so I’m not sure.)
These product identifiers were alphanumeric…or, at least, the 3-character model part of it was (as demonstrated by the “X##” models you documented); I’m not sure about the type code. I personally don’t remember ever seeing a type ID with a Roman letter in it.
I suspect you are correct that there is an internal logic to the way the model IDs are formed, but I am also sure that though it remained 3 characters throughout the lifetime of the IBM PC division, the kinds of details being encoded in the model# as well as the encoding method itself likely changed and morphed over the years to keep up with the rapid changes in the industry.
The Type-Model of my ThinkPad 770 from back in the day is still permanently etched into my brain, as deeply as my family’s landline number was from my childhood. 🙂 I had a 9549-1AU. I had to look these up, but my T42p was a 2373-KXU, and my T60p a 2623-DDU. Man, those were all fine machines…
It should also be noted that the IBM PC system Type-Model codes are wholly different than, and separate from, IBM part numbers and FRU#s (Field-Replaceable Units), which also were 7 characters in length and which could also be alphanumeric. You see these codes plastered over every IBM product, hardware and software, as well as even every scrap of paper that ships with an IBM product (hardware or software). For example, the OS/2 retail box for a given version, edition, and media combo would have its own unique 7-character P/N printed on the box, and then each separate piece of installation media would have a separate 7-character P/N, as would the manual, the registration card, etc. Different hardware components within each complete complete assembled machine would also bear unique IBM P/Ns and FRUs, so that parts could be ordered and referenced within maintenance manuals, as well as (I assume) also be used internally to piece together a build manifest for a particular system model. So I think the IBM PC Type-Model system was unique to the PC division, but the IBM P/N + FRU system was pervasive across the entire company.
Anyway, back to PCs, IBM would periodically publish a fat Personal Systems Reference book that outlined all current models available as of the time of publication. For ThinkPads, this was known as the “tabook”. There was also a separate publication that would list all historical models that have been withdrawn from manufacturing and the market, the “twbook”. I would include URLs to some PDFs, but I think I remember in the past that this feedback form would hold back my reply for moderation if I did so, so you’re just going to have to rely on Google. 🙂
It does appear that Lenovo has kept up at least some vestiges of the IBM PC system, as they have a “PSREF” section of their web site where they continue to publish similar-looking documents. They appear, though, to have retconned “PSREF” to now mean “Product Specifications Reference” instead of “Personal Systems Reference”.
Yes, it should be 1983, thanks. It’s marked March/April since the announcement letter itself is from April but references models “announced on March 11”.
I briefly looked at the 5160 XT model numbers and found that although there is some correspondence between the model number and hardware configuration, in the PC/XT it was already complex enough that there’s no simple mapping. Though I found a handy table: http://bitsavers.trailing-edge.com/pdf/ibm/pc/SA38-0037-00_Personal_Computer_Family_Service_Information_Manual_Jul89.pdf#page=76&zoom=auto,-139,641
Later on I’m pretty sure some of the letters/digits of the model number had a “human readable” meaning, but most did not. And yes IBM kept the Type/Model scheme going until the end. It was not new with the PC, either.
Yes, there were several 5160 XT models without a hard disk. There were no 5150 PC models sold with a hard disk, although it was possible to buy a 5161 Expansion Unit with a hard disk (or even two).
There were no floppy-less XT models, for an obvious reason… you’d have quite some trouble booting them up and getting anything onto the hard disk (not technically impossible, just something there weren’t readily available tools for).
The 5150 PC just kind of skipped over 128K. The initial XT came with 128K, then was upgraded to 256K and later 512K. It may be that IBM wanted the new flagship to have 128K and stick with 64K in the 5150.
This all happened rather quickly (relatively speaking); in 1982 it appears that PC models with less than 64K were all but gone, in March 1983 IBM rolled out new boards where 64K was the minimum, and starting June 1984 all 5150 models came fully populated with 256K.
Thanks for the link to the “XT document”.
Yeah, it would be hard to get software onto an XT without a floppy drive. But at least in 1987, possibly earlier, I think that (third party) network card with boot ROMs were already a thing. Especially for things like schools and whatnot it would probably had been of interest to not have floppy drives, to avoid students bringing in their own software that potentially could had been virus infected and whatnot.
Not just third party networks. The IBM Cluster supported RIPL. Troubleshooting the Cluster Adapter is part of the April 1984 5160 Guide to Operations. IBM wasn’t selling 5160s without hard drives while the Cluster was an active offering. The expected configuration was a 5160 with both hard drive and floppy being shared by diskless 5150s and PC Jrs.
As a side note, the Jr Cluster Adapter also required the installation of a power sidecar which wiped out the savings from not having a floppy drive.
@MiaM:
No, a 128Kb RAM 5150 wouldn’t make any sense. Due to a BIOS bug, a 5150 with a newer motherboard (of the 64Kb-256Kb type) can work normally only with the full 256Kb of RAM (and its motherboard type should’ve been called “256Kb-256Kb” due to that). It’s true that IBM was selling some ot these 5150s with the minimum possible RAM of 64Kb (which were anyway soldered to the mainboard), but the intention was apparently the customer to add the missing 192Kb.
“However, it is known that IBM did not build any 16K models with a floppy drive, hence model 001 did exist but 011 or o21 did not. The reason was that to boot from a floppy, the PC needed at least 32K RAM (since the boot sector was read to a memory location just below 32K.”
This has a letter O in “o21”. And the closing paren is missing.
https://xkcd.com/859/
Fixed, thanks!
@vbdadc
I think 128K (2*64) version costs more, than 256K (1*256) as it needs 2x more ICs and price difference between 64k and 256k ICs was not so large.
@LightElf:
IBM never offered a 5150 with a single 256KB bank. The 128K system would either be a 256K motherboard with only 2 banks filled or a 64K motherboard with a 64K expansion card.
“will no longer be manufactured and the replacement is a 64K system unit 5150014 (i.e. model 013)” – it should be model 014 here.
When the formation of the PC division took place, one of the conditions was, that they should not be tied by any IBM “rules”. That was partly to ensure speedy creativity but also so they could just shut it down if it failed.
This is probably why they where not announced by the regular announcementletters to begin with.
As the PC division took of it was eventually “merged” into big-blue including announcing new models and periphials by the regular weekly announcementletters.
FWIW, my 5150 had a model number of 167G (one-six-seven, not one-seven-six) and had 256K of ram and dual drives. I no longer have it, but I do have photos of the silver plate attached to the case with the model number info.
I must admit that I never knew that IBM made ram expansion cards. I thought that that was 100% a third party business…
Fixed, thanks.
So… what is that 167G? Was it some non-US model?
For one thing, the expansion cards were available right when the PC came out, before the whole 3rd party thing really started.
IBM continued making RAM expansion boards well into the PS/2 era.
Sorry, 167G has nothing to do with 5150 model types.
https://www.worthpoint.com/worthopedia/vintage-ibm-monitor-model-8512001-1876964457
“Listed IBM EDP Equip. 167G”
Or here, page 14
https://bitsavers.trailing-edge.com/pdf/ibm/pc/3270/3270_AT/SY33-0125-1_IBM_3270_Personal_Computer_AT_GX_Maintenance_Information_Jun1986.pdf
This is approve code.
Thank you! This makes a lot more sense than a completely unknown and strange model number.