Sales numbers for the IBM PC 5150? (or: What was behind the PCjr decision?)

[Trixter]

I'm doing some research for a video and part of that is figuring out why IBM made the decision in February 1982 to develop a home computer (what eventually became the PCjr). I think crazy sales numbers is the obvious answer, and there are plenty of sales numbers for the IBM PC over its lifetime, or the first three years, etc.... but what I am trying to find is the total unit sales from introduction up to that day in February 1982 when they decided to make the PCjr. Is there any way to find this information?

Alternately, is there any documented explanation of why IBM decided to make the PCjr?

 

[1ST1]

Low cost version? Enter the market where ATARI 400/800, Apple II, TRS80, VIC-20, CBM PET, Sinclair ZX80/81, ... started to be successful?

 

[Eudimorphodon]

Alternately, is there any documented explanation of why IBM decided to make the PCjr?

If you dig around the articles published the aftermath of the failure of the PCjr a thing you'll notice is a lot of commentators pinning the problems on IBM not actually having any coherent plan for the system, or at least radically misjudging who the likely customers were going to be. The TL;DR of it seems to be that nobody knew if they were trying to build a machine that would convince the newbies who might otherwise by a VIC-20 to pony out a few hundred more bucks in exchange for better features and some vague future-proofing promise of upward expansion, or if it was going to be a cheaper "you can take work home" option for experienced PC users which would dash the ambitions of MS-DOS clone makers trying to undercut the PC on price, which was subsequently undone by how crippled it was...

why IBM made the decision in February 1982 to develop a home computer (what eventually became the PCjr)

What's the source for this "February 1982" date? I was under the impression that the genesis of the "Peanut" project was closer to early (February?) 1983.

Here's an article about the PCjr that doesn't really cite sources but does have some numbers for early sales figures for the PC, like "150,000 over its first year". In its version of the Junior origin story it basically lays out the scenario that IBM was drawn like a moth to a flame to the insane sales churn and unit numbers that the "Home computer wars" kicked off by machines like the Commodore VIC-20 and C64 were generating. Machines in the bottom basement of that price war were pretty unprofitable unless you could sell a boatload of them (IE, Commodore), but IBM percieved that maybe there might be potential to create a product that would be able to upsell at least some of those customers into the "missing middle" between these cheap home machines and "real computers" like the IBM PC. (At this point Apple was starting to be the main player in this niche; the B&W TRS-80 line had once been a strong player here but its technical obsolencence was starting to rapidly spiral it into obscurity.)

It wasn't a crazy idea; the IBM name alone carried a lot of cachet, and with the right product they probably could have really put Apple in a vice. What killed the Junior was IBM saddling the engineers with the impossible task of building a machine that could somehow fill this "Apple Killer" niche without being too capable, lest it cannibalize even a single sale for the 5150/5160. The chicklet keyboard, limiting the floppy controller to a single drive, initially not supporting expanding the RAM beyond 128K... all intentional sabotage which left them with machine that wasn't really a good choice for *anybody*.

 

[krebizfan]

Here is an article that lists sales for 1982 and expected sales for 1983: https://www.nytimes.com/1983/03/27/business/big-ibm-has-done-it-again.html

The PCJr was afflicted by the efforts to make sure it didn't compete with the aging 5150 platform. It would have cost about $50 to add in a DMA chip and dedicated video memory which would have solved most of the problems with the Jr. Shaving those 5 to 10% off the cost forced massive cuts in price to keep the 500,000 Jrs moving out the door.

IBM production numbers indicate IBM was expecting Jr sales to be about half of combined 5150 and 5160 sales in 1984.

 

[Trixter]

What's the source for this "February 1982" date? I was under the impression that the genesis of the "Peanut" project was closer to early (February?) 1983.

I originally did this research in 2013 and unfortunately I've lost the source for it, but it was very clear that design started in February 1982. That said, the more I think about it, the more it makes sense it would have been 1983, so I'll try to find my source for that.

As for the rest of your comments, I agree, and similar conclusions are in my work from 2013. IBM tried to serve two markets simultaneously, and ended pricing themselves out of the home market, while creating a product ill-suited for the business market.

 

[Eudimorphodon]

The PCJr was afflicted by the efforts to make sure it didn't compete with the aging 5150 platform. It would have cost about $50 to add in a DMA chip and dedicated video memory which would have solved most of the problems with the Jr. Shaving those 5 to 10% off the cost forced massive cuts in price to keep the 500,000 Jrs moving out the door.

FWIW, the Tandy 1000 did fine without a DMA chip and with shared system/video memory. It did *get* a DMA controller when you added a memory expansion board, but that was literally just there refresh the expansion memory... it did also improve floppy disk performance, but from using 1000s without DMA I'd wager the difference is pretty inconsequential for most applications. Ironically I don't think Tandy's design is even any more "expensive" than the PCjr's; I mean, it might have a couple latches in it the PCjr lacks? The main difference is that Tandy's design *will not work* with 64K, it relies on 128K at minimum. (It addresses and latches 16 bits at a time; my vague memory is the Jr.'s version is at least interleavened, but it's still much less good at syncing up the CPU and video accesses.) This would be a problem for selling the cassette-only 64K version... not that IBM actually sold any of those in the real world.

(Tandy's design also fixed the *huge* issue with the PCjr not playing nicely with expansion memory by making the RAM shared with the video system relocatable. IBM's design forces tacking expansion memory on *after* the video memory, which is fundamentally incompatible with DOS's memory map without a bunch of not-great workarounds. Again, it's a fix that only costs a 3 bit latch.)

Other than the horked memory map if you had to pick one absolutely unforgivable original sin in the PCjr's hardware it's got to be the keyboard interface. Not talking about the infrared connection, but how IBM designed it around firing an NMI just to deserialize the keyboard input. I mean, this is nuts; they completely ***tcanned system performance, especially for tasks like serial communication, to save the price of a shift register. (A lot of the performance issues with the PCjr that people think are related to DMA are actually because of this.) It's hard to even blame this on trying to dumb-down the Jr. to not cannibalize its older brother because... why? It's like an angry engineer intentionally buried a landmine in the system which nobody is going to notice until they've already bought the system to step on it.

 

[Trixter]

I am planning on mentioning all this -- but briefly, not in acute detail, as what I'm working on is for the masses. If I go too far into the weeds, I'll lose them

I agree lack of DMA was not a huge deal in terms of overall system operation, but floppy transfer rates are definitely slower on PCjr. I'm almost positive the PCjr drive needs more revolutions, as the CPU is handling it.

Serial was also hindered by other factors; there was no way to go faster than 9600 baud due to how it divided the color burst signal.

 

[Trixter]

What's the source for this "February 1982" date? I was under the impression that the genesis of the "Peanut" project was closer to early (February?) 1983.

I've asked someone on the PCjr engineering team to answer this, but there are some references online of 1982 being correct:

https://www.filfre.net/2013/07/the-unmaking-and-remaking-of-sierra-on-line/ mentions that Sierra was asked late 1982 to develop software for it.

https://www.homecomputermuseum.nl/en/collectie/ibm/ibm-pcjr-pc-junior/ mentions 1982, albeit with no references.

https://en.wikipedia.org/wiki/IBM_PCjr discusses how Peanut rumors started a year before PCjr was introduced in 1983, so that would have been 1982.

If I find my original source, I'll post again here.

 

[Eudimorphodon]

I would *love* to know what the story was behind that keyboard interface, if any of the old-timers responsible are still around to relate it. It's genuinely baffling in how evil and unnecessary it was.

https://www.filfre.net/2013/07/the-unmaking-and-remaking-of-sierra-on-line/ mentions that Sierra was asked late 1982 to develop software for it

I guess would probably take that with a grain of salt. (For instance, there's a comment in that article that suggests the date in the article relating to when Sierra On-Line's office was constructed was off by two years.) But of course it's always possible that IBM might have been kicking around at least an idea in late 1982? The way the article reads it actually kind of comes across like working on this IBM project was something they threw themselves into full-blast after things started going sour for them around mid-1983, though.

In any case, I guess the question of when the *idea* to make something for the home computer market actually crystalized into a hardware spec is perhaps the real question. IBM is the sort of company that might send out feelers when they don't have much more than a vague marketing plan.

https://en.wikipedia.org/wiki/IBM_PCjr discusses how Peanut rumors started a year before PCjr was introduced

The Jr. didn't hit stores until March 1984, so there's a lot of wiggle room in "a year before". But even if we go with the November 1983 announcement that's late 1982, not February.

 

[Eudimorphodon]

... I guess on the flip side it would also be very "IBM" in the "big bad corporate monolith" sense if they *had* put together some early prototype of the PCjr in mid-1982 only to have the whole thing bog down with uncertainty and infighting that resulted in the whole thing ending up in development hell for a year and a half, I suppose only an insider who'd decided his nondisclosure agreements were no longer binding could really let us know for sure. (Again, if there are any left.) I mean, there was that idea for the 5150 that a minimum configuration could run just using memory on the proto-CGA card that's described in the concept docs, it's certainly possible someone pulled that plan out of the filing cabinet after the Commodore 64 came out and said "hey, what if we build basically this, but with 64K chips instead of 16K, slap in a sound chip and a palette register, and leave off the slots? We could sell that for $595 too, right?"...

Again, lacking any internal documents it's impossible to know for sure.

 

[Eudimorphodon]

I agree lack of DMA was not a huge deal in terms of overall system operation, but floppy transfer rates are definitely slower on PCjr. I'm almost positive the PCjr drive needs more revolutions, as the CPU is handling it.

Still utterly blows the doors off a Commodore 1541.

I guess if you were really bored and had acccess to a PCjr, a DMA-less Tandy 1000, and a conventional 4.77mhz PC/XT it would be "interesting" to actually do some disk benchmarking and see if it's DMA, or just the PCjr being slower because of the video contention-related slowdowns. (Does disk I/O get faster in the Jr. if you use a large memory expansion and configure the driver to dedicate the built-in memory solely for video, punting DOS and the related buffers out of the contention zone?) I'd try doing a bake-off between my DMA-less Tandy 1000 HX and my Kaypro PC, but currently the latter has neither floppy drives installed or a compatible keyboard.

 

[krebizfan]

An other infra-red keyboard from the same time was part of the Apricot F1. That receiver is a lot more complicated than the PC Jr's. The Apricot does a similar processor loop to pick up keystrokes but uses a Z-80 just for IO so the 8086 doesn't have to. I think the PCJr keyboard design was the cheapest way to get an infra-red keyboard working. It seems no one at IBM had been using the prototype Jrs to see if they would meet requirements.

 

[rmay635703]

It’s worth noting that there were grumblings of IBM over producing 5150’s directly into the storage warehouses 1984+ due to being slow to react to a general electronic downturn that already was starting in 1983.
Their lack of awareness of how to profit from the Japanese ram dumping around that time was equally baffling.

It wasn’t long after that the “you couldn’t give away a 5 slot PC” commentary started.

IBM, basically took something that could have been successfully and shot themselves in both feet with utterly baffling decisions that saved no money to implement but POd their base.

I like to think of the PCJR as IBMs worse version of the unnecessarily non-expandable
Mac 128 but without any exciting software to go with it. (Kings Quest, meh)

 

[Eudimorphodon]

An other infra-red keyboard from the same time was part of the Apricot F1. That receiver is a lot more complicated than the PC Jr's. The Apricot does a similar processor loop to pick up keystrokes but uses a Z-80 just for IO so the 8086 doesn't have to. I think the PCJr keyboard design was the cheapest way to get an infra-red keyboard working. It seems no one at IBM had been using the prototype Jrs to see if they would meet requirements.

Gross. Looking at the PCjr technical manual it does look like when decoding what it gets from the IR receiver it's using a software sampling technique to "de-noise" the incoming signal. Which I guess explains the NMI, but what a waste. How anyone thought this was a good idea is just... something.

 

[Trixter]

I would *love* to know what the story was behind that keyboard interface, if any of the old-timers responsible are still around to relate it. It's genuinely baffling in how evil and unnecessary it was.

I don't have specifics on that, but talking with one of the engineers I was surprised to learn that the entire engineering team, minus the project leader, was aged 22-25 -- maybe it was just inexperience?

In any case, I guess the question of when the *idea* to make something for the home computer market actually crystalized into a hardware spec is perhaps the real question. IBM is the sort of company that might send out feelers when they don't have much more than a vague marketing plan.

This is true. I don't care one way or the other; if it's 1983, I have some material to rewrite, but I just want the truth. I sent some feelers out, will update if I learn something concrete.

The Jr. didn't hit stores until March 1984, so there's a lot of wiggle room in "a year before". But even if we go with the November 1983 announcement that's late 1982, not February.

I was indeed going with the November 1st announcement. As for stores, that's not exactly true; PCjrs starting shipping to stores January 13th, 1984. What's somewhat insidious, however, is that consumers with preorders didn't start getting theirs until March. IBM prioritized sending to retail outlets first to serve as display models, before fulfilling customer orders, which is slightly scummy.

.if they *had* put together some early prototype of the PCjr in mid-1982

From an oral history from one of the engineers, the first prototype was 100% ready Spring 1983, but then repeatedly kept failing FCC certification due to excessive RFI, and it took them until September 1983 to figure out a solution (their patent for "spraying liquid metal" to the inside of the plastic case).

Had PCjr gotten into consumer's hands BEFORE the Christmas '83 season, I wonder if it would have changed things. Then again, maybe not, since pricing + positioning is what truly killed PCjr.

"hey, what if we build basically this, but with 64K chips instead of 16K, slap in a sound chip and a palette register, and leave off the slots? We could sell that for $595 too, right?"...

LOL. Not sure, although from same oral history, the cartridge slots were a direct response to marketing requesting a way to target the "videogame and educational markets".

All this is going to be a youtube video series starting in November, if all goes well. Like Columbo, there are just a few things that are bothering me, some loose ends to tie up.

 

[vwestlife]

It's paradoxical that the IBM PC was so remarkably simple in its design, and then IBM somehow decided it would be a good idea to make a cheaper low-budget computer by adding a whole bunch of new features to it. And then Commodore did the same thing, by adding more and more features to a $49 ZX81 competitor until it ended up more expensive than the C64!

 

[Eudimorphodon]

it took them until September 1983 to figure out a solution (their patent for "spraying liquid metal" to the inside of the plastic case).

Gotta say, I like the mental image of a very small T-1000 oozing out of your PCjr in the middle of the night to cause mischief in the den. Yet another reason to go home with an Apple IIe instead!

Had PCjr gotten into consumer's hands BEFORE the Christmas '83 season, I wonder if it would have changed things. Then again, maybe not, since pricing + positioning is what truly killed PCjr.

I think the PCjr as it existed on debut would have had a rough time whenever it came out (the reaction to the chicklet keyboard alone when it was first unveiled was nothing short of horror), but missing Christmas 1983 certainly didn't help.

A thing you've got to remember about late 1983 is this was the peak of the "Video Game Crash", IE, the period where retailers and consumers alike completely lost confidence with the oversaturated console video market. The crash was the result of a number of converging factors, but in addition to rampant quality and oversupply problems in the video game space itself by 1983 home computer makers had gone all-in on marketing that pushed the idea that instead of buying lousy and limited "toys" like game consoles for their kids they should get them an "educational" computer instead. This, combined with a price war between the various PC manufacturers (kicked off by a fight between Commodore and Texas Instruments that nearly ended up bankrupting TI) meant the video game console was essentially dead in the US; if you were going to get the kids an expensive "tech toy" for Christmas in 1983 it was probably going to be a computer, so both computer makers and retailers were pretty much in full panic mode trying to stuff the channel in time for it.

TL;DR, Commodore was pretty much the only company that really "won" in the end, at least in the US. Commodore had already slashed the list price of the C64 from $595 to $300 in June 1983 to strangle TI to death, and Coleco's attempts to get the ADAM out in time for Christmas (it technically launched in October, but problems ramping up production resulted in insuffient sales before Christmas to make a profit, and severe quality control issues that resulted in as many of 80% of the roughly 90,000 units that got out before the end of the year being defective completely ruined the machine's reputation going forward) just inspired Commodore to make further price cuts/offer package deals that in the end resulted in more than half a million sales. And the resulting fallout from all this bloodthirsty competition ended up haunting the home computer industry through 1984, resulting in something of a backlash/downturn that didn't really turn around until mid-1985. Companies like Atari also bankrupted themselves trying to keep up with Commodore, there was a lot of consumer backlash and dissatisfaction from people who bought machines that turned out to not be particularly useful, at least without a lot of expensive upgrades... etc.

It's actually right in the middle of the post-Christmas hangover that the PCjr with its dinky chicklet keyboard and "competitive for early 1983, maybe?" pricing went on sale which, yeah, was not a smart play. The $600 diskless version of the PCjr might have made a *little* sense when the Commodore 64's list price was $595, but in early 1984 it was a laughable prospect. (*Nobody* was going to pay $600 to play game cartridges in March 1984; they shouldn't have even put that on the table.) Meanwhile, the $1269 list price of the floppy disk version wasn't quite so out of line if you compared it to the *list* price of an Apple IIe, but it was easy to get Apple IIs for a lot less than list, and that delay until March gave Apple time to get the IIc slapped together... and of course it's *not* Christmas, so really, the pressure for anyone to buy anything right away just isn't there. Maybe ask again when it's back to school season? But, man, I've heard schools just *love* Apple IIs...

So yeah, I dunno. In some alternate universe where IBM could have gotten a couple hundred thousand of the disk-equipped PCjr into the channel by, say, October 1983, would it have done any better? (The diskless one, again, pointless. There's no world that made sense in. Maybe if they'd put together some kind of network system for it they could have sold it to schools, but that's the only thing I can even remotely fathom.) $1200 compared to the $700-with-printer-and-everything price point that Coleco had thrown down with the Adam certainly would have made it a tougher sell, but with a real disk drive and the IBM name on it... I'm willing to bet they would have suckered in a lot more people than they would have in March 1984, sure. I don't know if it would have really saved the system long term, I'm sure a lot of the customers who ponied up for it would have experienced significant buyers remorse, but there's something to be said about getting things *into people's hands* where you can leverage the power of the sunk cost fallacy. Pre-announcing it the way they did, four months before anyone could actually buy one, *really* looks stupid in retrospect. If your product is a little iffy and unlikely to live up to unrealistic expectations the *last* thing you want to do is give people months to think about it before you try to actually sell it to them.

 

[krebizfan]

Comparing with Commodore seems unreasonable. Commodore's business model was built around selling the main unit at breakeven prices and then turning a profit on the proprietary addons. Under the Consent Decree, IBM couldn't do that. IBM managed to sell 500,000 Jrs at a modest profit even with all the problems. I think that had anyone in IBM used the Jr before release and fixed the issues, IBM might well have sold the same 500,000 at close to the planned prices.

 

[Eudimorphodon]

Comparing with Commodore seems unreasonable. Commodore's business model was built around selling the main unit at breakeven prices and then turning a profit on the proprietary addons.

Whether it's "fair" or not, those were the realities of the market they were looking to enter. Even if you say that they were going after Apple instead of Commodore (which is a case that gets a little muddled by the confused mixup of a product they churned out: some of Commodore's competition had chicklet keyboards, not Apple's) the fact is that even Apple was making concessions at this point that made the "real price" of an Apple II a lot lower than their list prices in order to keep the low-end cannibalizing of their product to a reasonable level.

(Luckily for Apple their list prices were comfortably completely out of whack with what the product actually cost, so they could easily afford steep discounts... but not too steep, because they were all about maintaining the myth of being the "premium" choice.)

IBM managed to sell 500,000 Jrs at a modest profit even with all the problems.

Do you have a source for the "modest profit"?

The standard sources indicate that PCjr sales were a complete black hole for the first five or so months it was for sale, with quotes like the number of Juniors sold by the end of May being "under 10,000", and the Apple IIc selling "more units on the first day" (around 50,000) than the PCjr sold until at least August. On July 31st IBM held an event where they introduced the new keyboard and a new line of peripherals including, critically, RAM expansions, and discounting the price to $999; and yes, this did revive PCjr sales, letting them move about a quarter of a million of them by the end of the year. During this time some dealers were even giving away color monitors for free with the $999 package, which, yeah, is a pretty good deal even compared to a Commodore 64... but was IBM actually making a profit there? IBM stopped the discounts in January and the sales went back in the toilet until the discontinuation of the machine was officially announced in March 1985, at which point they went back on clearance rack until they finally ran out by the end of the year or so.

Anyway, the lesson seems to be that the Junior certainly would have been a huge sales success if it'd come out at $999 with a color monitor and a decent keyboard, but those are in fact pretty darn close to Commodore 64 prices. (In late 1983 a C64 + 1541 drive + 1702 monitor would have been in the ballpark of $999, give or take a couple hundred bucks.) If they hadn't screwed up the keyboard and had gotten it out the door for Christmas 1983 I would actually guess it would have sold pretty well for $999 without the monitor, or even *possibly* at full price (that keyboard really murdered them), but as it was they completely blew the opportunity to make a good initial impression.

 

[Trixter]

There was no modest profit My research indicates that IBM wrote off at least 45 million dollars in unsold PCjr inventory, which was between 100K and 350K units depending on which source you think is more accurate.

In September 1985 they introduced an employee discount of an expanded PCjr (128k+disk drive), monitor, and three software packages for $450 -- that's less than the manufacturing cost, so not even trying to break even at this point. At the same time, schools received one free PCjr setup for every five IBM PCs they purchased.