The IBM XT 286, also known as the IBM 5162 or IBM model 5162, was a short-lived computer from IBM announced in 1986 and released in 1987. It was a difficult computer to understand at the time, and had a mixed reception when it was new, although it has a cult following today.
What the IBM 5162 was
The IBM 5162 looks like an XT, but internally, it is much more similar to the 5170 AT. It was a deliberately mediocre machine, and in some ways, represents everything that was wrong about IBM in the late 1980s.
In 1986, IBM had a problem. Their top of the line PC, the IBM 5170, was selling reasonably well, but there were good clones available from various sources, some of them big names, and some of them complete unknowns. But the problem for IBM was that you could get a knock off product that looked substantially like theirs, and worked exactly like theirs, and cost 20 or 30% less.
Some people bought IBM anyway, because it seemed like a safe choice. The conventional wisdom was there was no guarantee the company who made your knockoff would still be around to honor the warranty if anything went wrong, but IBM would be. But the longer those competitors failed to go out of business, the less wise that line of thinking seemed.
The IBM XT 286 was an attempt to get in front of that problem. It was exactly what it sounded like, a mashup of XT and 286 technology. It used the XT case, and it initially used the same hard drive and as the XT, but it contained a new motherboard that was much closer to the 5170 then it was to the 5160 motherboard it replaced in the case.
The 5170 motherboard was too big to fit in an XT case. But by 1987, it was possible to rearrange the board and make everything fit on an XT sized board. By doing this, IBM accidentally invented what came to be known as a baby AT board.
But the whole thing came with some compromises. IBM didn’t want to make it too good, after all. They wanted something that would be price competitive, but still leave some incentive to buy the more expensive, and more profitable 5170.
The top of the line 5170 ran at 8 MHz. The 5162 ran at 6 MHz. On paper it matched the original 5170 in speed. But even though it should have been 25% slower than the 5170, in practice, the difference was more like 10%. That’s why the 5162 has a cult following today. It is a better machine than it seems like it should be.
The reason for that is a design compromise from 1984. RAM chips that could keep up with the 286 processor were expensive and in short supply in 1984. So to get the margins IBM wanted, they had to slow the CPU down when it came time to access memory. This trick is called a wait state. And when the faster 5170 running at 8 MHz came along, it used the same trick.
The 5162 didn’t have that. So the result was a slower computer that was better than it sounded.
But in some instances, IBM took back what it gained by saddling the machine with the same slow hard drive that a regular XT had. If you were lucky, you got a 5162 with a fast AT class hard drive, but you didn’t know until you got the machine home what you were going to get.
The other problem with the 5162 was the height of the case. XT cases were not quite as tall as AT cases. So when the AT came out, some manufacturers took the liberty of using the full available height. It allowed them to fit more functionality on a single car than would have fit otherwise.
But then IBM came along and introduced a machine that had an electrically compatible bus, but the case wasn’t as tall. So the card would plug in, but you wouldn’t be able to put the cover back on.
Making matters worse, even though the badge said it was an XT, your XT cards wouldn’t necessarily fit. Sometimes the design of 8-bit cards kept them from fitting into the 16-bit slots. The 5170 and 5162 both contained some 8-bit slots to accommodate these types of cards, but two of the three 8-bit slots did not extend the full length of the case. So if you were upgrading from a regular XT, you might not be able to fit all of your existing cards into a 5162.
These problems weren’t unique to the 5162. Some clones had similar problems, including the widely popular Tandy 1000, so peripheral makers quickly adapted. But critics reviewing the machine at the time weren’t pleased.
How the 5162 could have been worse
IBM didn’t make the machine as bad as it could have. The Tandy 1000 TX, which also dated to 1986, is a good example of what might have been. The TX, which amusingly may or may not have been XT spelled backwards, used a 286 processor, but otherwise, the machine was still an XT. No 16 bit expansion bus, no extended memory, just the faster processing speed and extended instruction set.
Tandy got away with it because they priced it at $1,000. The XT 286 cost $3,395 to $3,995, depending on the configuration. When you charge 1/3 to 1/4 as much, you get more leeway.
The IBM 5162 today
The 5162 is much less common than a 5160 or 5170. It wasn’t on the market nearly as long, and didn’t sell as well. It had only been on the market for about 6 months when the PS/2 line hit the market. So it didn’t exactly come out at the best possible time, since everyone knew IBM was about to release something, and this machine probably wasn’t it.
With a modern storage solution, it outpaces a 6 MHz 5170 and nearly catches up with an 8 MHz 5170, thanks to the lack of wait states. But like reviewers said at the time, you could get an AT clone with zero wait states running at 8 MHz or faster and get all the benefit of both approaches. But it’s a historical curiosity, and hobbyists love historical curiosities.
If you go to buy a 5162, be careful. Some listings for IBM PC XT 286s aren’t really a 5162, but rather, a regular 5160 upgraded with a 286 motherboard along the way. There may be more of those floating around than there are real PC XT 286s. They aren’t worthless, but they aren’t worth what a genuine 5162 is.
The PC/XT 286’s legacy
Some say the 5162 was a failure because it didn’t sell. But sometimes you introduce a product just to keep people from buying something else. That happened a lot in the computer industry in the 1980s and 1990s. Essentially, the 5162 was a teaser, to keep people interested until IBM had its newer and better thing ready, and to entice you to wait a few months rather than buying a clone, or one of those newfangled Motorola 68k based machines.
Arguably the tactic worked. Everyone was talking about that new generation of computers, but if you are reading this on a PC, that PC has a lot more in common with the old generation than the supposed new generation. That’s because everyone talked about these cool new computers, but then they bought an IBM or a clone.
If you measure your success in units sold, the 5162 was a failure. But it kept IBM in the game.
It also served as a proof of concept for one of those future machines. One of the computers in the PS/2 line was the Model 50. It had a 286 processor, the same as a 5162 or 5170. But it ran at 10 megahertz. It had that same wait state trick the 5170 had. But soon after, IBM released the Model 50Z. The Z stood for 0 wait state. So it was like the 5162, but it ran at 10 megahertz. That machine was very popular, because it gave probably the best balance of price and performance in that product line.
That’s the charitable way to look at what IBM was doing. Now let’s look at the other point of view.
The turning point
IBM announced the 5162 on September 2, 1986. Two days later, Compaq announced the computer IBM didn’t want to build, a PC based on the newer Intel 386 CPU. Worse yet for IBM, the announcements of the two machines ended up sharing space on the front page of the September 8, 1986 issue of Infoworld magazine.
So in some ways, the 5162 is a nice machine. And it’s an underrated machine. But in other ways, it represents just about everything that was wrong with IBM in 1986 and 1987. In 1986, while Compaq was figuring out how to put a 386 CPU into a PC architecture intended for the 286, IBM was mixing and matching pieces from its 1984 product line to get a slightly new flavor of mediocre for 1987.
If you want to know where it all started to go wrong for the mighty IBM, you can do worse than argue it was that first week of September 1986.