Fifteen Classic PC Design Mistakes

By  |  Sunday, June 14, 2009 at 11:33 pm

PC Design MistakesThere’s no such thing as the perfect computer, and never has been. But in the personal computer’s long and varied history, some computers have been decidedly less perfect than others.  Many early PCs shipped with major design flaws that either sunk platforms outright or considerably slowed down their adoption by the public.  Decades later, we can still learn from these multi-million dollar mistakes.  By no means is the following list exhaustive; one could probably write about the flaws of every PC ever released.  But when considering past design mistakes, these examples spring to my mind.

Special thanks to Steven Stengel of the Obsolete Technology Homepage for providing many of the photos in this article.

Apple III (1980)

Apple III

The Apple III was Apple’s first computer not devised by Steve Wozniak, Apple’s co-founder.  Instead, a committee of engineers designed it to be the “perfect” business system.  With an absurdly high price (options ranged between $4,340 to $7,800–about $11,231 to $20,185 in 2009 dollars) and numerous bugs at launch, the Apple III was doomed to failure.

Problem #1: No Power Supply Fan

The Apple III’s lack of power supply fan caused system to heat up, warping the motherboard and unseating certain socketed chips.

Apple III ChassisWhat Were They Thinking?

According to Apple insiders, Steve Jobs’ zeal for a simple and silent computer design forced the Apple III team to exclude a cooling fan for the power supply.  Apple later suggested a simple fix for the heat-warping problem: raise the Apple III a few inches off a hard surface and drop it, hopefully re-seating the chips in the process. Fortunately, that advice wasn’t required for later Apple computers that lacked fans.

Problem #2: Limited Apple II Compatibility

To run an Apple III in Apple II mode, one had to first boot from a special floppy disk.  Once in Apple II emulator mode, the user could not use any of the Apple III’s enhanced hardware, including 80 column text mode or the real-time clock.  Compatibility with Apple II software was not perfect, as many software packages used direct memory writes in the form of PEEKs and POKEs that didn’t line up with the Apple III’s memory structure.

What Were They Thinking?

Like IBM and the PC/PCjr, Apple wanted a clear product delineation between their “home” machine (the Apple II) and their “professional” machine (the Apple III).  As a result, Apple II compatibility on the Apple III was intentionally crippled.

DEC Rainbow (1982)

DEC Rainbow 100

In the early 1980s, mainframe giant Digital Equipment Corporation tried its hand in the personal computer market with the DEC Rainbow.  Since industry-wide PC standards had not yet been decided, DEC decided to support the two major operating system platforms at the time: Z-80-based CP/M and 8088-based MS-DOS. The Rainbow contained both these processors and was capable of booting both OSes.  Interestingly, despite the machine’s use of the 8088 and MS-DOS, the computer remained largely incompatible with the IBM PC.

Problem #3: No Way to Format Disks

DEC provided no official method of formatting blank diskettes for use in the Rainbow.  As a result, Rainbow users were forced to buy pre-formatted floppy disks from DEC.

What Were They Thinking?

The Rainbow’s 5 1/4″ floppy disk drive used a proprietary sector format that was incompatible with that of either IBM PCs or CP/M machines.  That alone is not unusual for a machine released in 1982.  But preventing users from using non-DEC floppies seemed like a nickel-and-dime move on DEC’s part to make an extra buck on exclusive disk sales.

Later, third party options for formatting Rainbow disks sprung up, spoiling DEC’s monopoly on floppy sales.

Coleco ADAM (1983)

Coleco Adam

The Coleco Adam was a classic case of overreaching in the video game and home computer markets.  Coleco had recently found success with its ColecoVision home video game console, and it also wanted a piece of the home computer market as well.  Despite its faults, the Adam had one of the best home computer keyboards of its time, with an excellent layout and feel.  This small plus typically gets buried under the system’s numerous drawbacks.

Problem #4: EM Pulse Erases Tapes

Coleco Adam Data CassetteEvery time a user switched on or off the system, the Adam’s power supply emitted a strong electromagnetic pulse that could scramble or erase the data stored on magnetic data cassettes that the Adam uses to store its programs.

What Were They Thinking?

In this case, they weren’t.  Insufficient testing and a headlong rush to market likely led to this embarrassing flaw.  Users revolted by the thousands, and nearly 60% of the first batch of Adams were returned as defective.

Problem #5: Printer Required

The Adam could not be operated without the bulky, noisy, slow printer attached — both the system’s power supply and power switch are on the printer itself.  If the printer failed or needed repairs, this took the entire system out of commission.

What Were They Thinking?

Coleco engineers probably figured that reducing the number of cables going in and out of the Adam would make the system more user-friendly, but it turned out to be a half-brained design decision that failed to take the big picture into account.

1 2 3 4 NEXT PAGE»


184 Comments


Read more: , ,

162 Comments For This Post

  1. JV Says:

    The IBM PC Jr was my family's first PC. Interestingly, we didn't have the wireless/IR keyboard pictured here. Ours had a relatively normal non-chicklet, wired keyboard. Maybe IBM ditched that design at some point, or offered 2 versions?

  2. John Turner Says:

    I believe the wireless chiclet was abandoned for a non chiclet wired keyboard due to the complaints about the original keyboard. -but I could be remembering incorrectly…

  3. Crusty Says:

    The Apple III wasn't the only computer to suffer socket issues. The Amiga 500 shipped with defective sockets that caused the chips to work loose during shipping. We used the drop method also (although I would recommend finding a concrete floor with low knap carpet) to fix as many of the defectives as possible.

    Those were some of the best tech support calls I've ever had.

    Good times…

  4. Galvo Says:

    Also Radio Shack Color Computer 3 I was told.

  5. John Says:

    The TRS-80 or Tandy Color Computer 3 only had one socketed chip – all others were soldered directly to the board – but yes, it was sometimes necessary to reseat that one chip (GIME chip) if a machine ceased to function properly – we often pulled the chip and reseated it thus curing the problem – never heard of anyone dropping the CoCo 3 to reseat it – but who knows…

  6. mike Says:

    GREAT ARTICLE, Brings back a lot of memories. Here are a few more.

    1.) The original Tandy TRS-80 (model one) created so much radio interference that it completely destroyed TV reception on channels 2, 3, 4 and 5 for about 100 yards.

    2.) The original Macs booted from a 128K disk, load a few fonts and there was little room for anything else without endless disk swapping.

  7. Galvo Says:

    400K floppy, 128K RAM Macintosh (later 512K RAM models & upgrades by replacing 4164 with 41256 DRam chips)

  8. PapayaSF Says:

    Jay: Ah yes, the original NeXT cube. Steve Jobs said it had no floppy drive because floppies were "70s technology," but of course it was way too soon to discard them. One big complaint was that it forced software vendors to ship programs on those odd optical disks that cost (IIRC) about $50 each. Later NeXT machines did have floppy drives.

  9. Galvo Says:

    Yes 2.8M floppy drives, I've never seen one or a floppy for them.

  10. John Turner Says:

    2.8M Floppy Drives (well, 2.88M) were used in some of IBM's MCA Systems (PS/1 or PS/2). I have some of the drives that are new – I've been meaning to hook them up and play around with them.

  11. James Says:

    THE classic mistake: choosing the 8008 rather than the 68000. Yes, a couple of decades and billions of dollars have put plenty of bags on the side of the obscenity that is the x86 architecture, but how much better off would we be today had IBM started with the far cleaner 68000.

  12. Galvo Says:

    I think you mean 8088. The 8008 was in EARLY stuff like the Micral & the Radio-Electronics Mark 1.

    But yeah, not that IBM would have wanted to make too good a micro & risk their minicomputer income. They did use a 68000 for other products eg. http://www.columbia.edu/cu/computinghistory/cs900

    This seems to be a popular topic – http://www.anonymous-insider.net/unix/research/19

  13. Peet McKimmie Says:

    The biggest mistake of all time was the design of the original IBM PC power supply, with the fan blowing out the back. This became an "industry standard" that has led to nearly 30 years worth of computers sucking in filth and dust-bunnies from every orifice. (Including sucking grit into removable-media drives…) If they had pointed the fan *inwards* they could have put a filter over it to keep the dust out, and the machine would then have been positively pressurised and thus would have stayed clean.

  14. Galvo Says:

    I agree 100% about the sucking fan but I also think hot power supplies & dangerous voltages shouldn't be in the main case anyway. However reviewers always complained when a computer had an external power supply. But large plugpacks that were supposed to go straight on the outlet were a pain! At least now most of us have power boards.

  15. keptwench Says:

    If IBM had started with the 68000, we'd all be using Amigas today. I wish.

  16. Galvo Says:

    Amigas are the least bad desktop machine I've used, but possibly they wouldn't have existed if IBM PC had used 68000s. Who knows?

    (I saw an Amiga X1000 motherboard recently, should see it working next week)

  17. Lou Wilson Says:

    Another problem that probably killed the TI 99/4a was a marketing snafu. You couldn't buy them in stores, you had to access a direct marketing organization. They were sold like Tupperware or Amway. I wasn't about to shell out 3 or 4 hundred dollars to my neighbor who had no friggin' idea what he was selling or how to support it.

  18. John Says:

    Western Auto sold them here back in the early 80's until the machine was discontinued. J&R Music World also sold them at around the same time.

  19. Dave Zatz Says:

    “Fortunately, that advice wasn’t required for later Apple computers that lacked fans.”

    Ah but the cube also had problems… cracks. Possibly also caused by heat? http://www.google.com/search?q=g4+cube+cracks

  20. Chris Tucker Says:

    RE: C64 Disk Drives. Yes, they were slow. Compared to the datasette, they were blazingly fast.

    Originally, the 1540 drive was intended for the VIC=20. When the 64 users demanded a floppy drive, the 1540 had to be modified to work with (as I recall) the somewhat faster C64 serial line.

    If the 1541 hadn't been so slow, I doubt if the utility cartridge as we came to know it (Super Snapshot comes to mind) would ever have been developed. Fast loads and saves, ML monitors, copy protection free memory images that could be saved to disk, all that might never have been, if not for the slow I/O of the drive.

    Eventually, the power users installed JiffyDOS in their 64/128 and the drives, and that was that, re slow data I/O.

  21. C.T. Hunnefield Says:

    According to a recent C= book, (On the Edge by Brian Bagnall), the reason why the drives were so slow in the C64 was because there were trace lines for compatibility with the old 1540, and NEW serial lines for the high speed 1541.

    The tapeout to the Japanese board manufacturers had both sets of lines in, however, the Japanese firm decided that the extra serial lines were 'not necessary' and DELETED THEM! :O

    Thousands of the boards were manufactured like this, and it was too late to do anything but use a software patch to fix the problem by providing a compatibility mode between the drives. It made loading, even on the 1540 VERY SLOW! The funny thing is, that supposed cross compatibility between the 1540 and 1541 never worked 100% anyway. The 3rd party 'Fast Load' cartridges fixed the problem by throwing out the 1540 compatibility, making the 1541's drive speed more reasonable.

    The end result was a board that SHOULD have been the fastest floppy access on the planet to the slowest. Very sad story. MILLIONS of man hours wasted…

  22. Rocky Says:

    Back to those Commodore 64 1541 disk drives…. It's one thing to have a ridiculously slow disk drive (thus giving birth to the disk drive accelerator market) — but to REPEAT THE SAME MISTAKE WITH THE AMIGA is unforgiveable!!! How could Commodore make the same mistake twice!!!??? I realize they were developed by different design teams… but c'mon!!!

  23. C.T. Hunnefield Says:

    Actually they made the mistake THREE times if you include the VIC 20! Four if you include the C128 (although that was done for compatibility…)

  24. Brian Mueller Says:

    This is the type of post I really enjoy on Technologizer. For me this is what separates it from the spate of other technology blogs. Thanks!

  25. Bill Says:

    One of the worst pieces of hardware engineering ever was the connector mess created by IBM on their original PC. They used a male DB-25 thereby requiring a major repair for a simple bent pin. They used the DB-25 connector for the parallel printer port even though that connector had previously been used primarily for serial application and thereby caused enormous cable confusion and probably a lot of damaged interfaces.

  26. Galvo Says:

    Then the Amiga 1000 used DB-25s for both parallel & serial, but the opposite genders to a PC so if someone used an IBM style cable they wrecked something.

  27. Peter Says:

    I worked at Data General when they introduced the DG/One. Aside from the poor contrast on the LCD display, it was actually a pretty good machine, and 85% PC-compatible.

    Unfortunately, the 15% that was incompatible, was the serial port. The chip used in the PC serial port was an Intel 8250. This part wasn’t available in CMOS, which the DG/One badly needed to reduce its power requirements, so the engineers chose the 82C51, functionally, but not register compatible with the 8250.

    The rest is history…no PC programs that used the serial port would run on the DG/One. Yes, there were DG/One versions, but the serial port incompatibility, low contrast screen and DG’s misguided emphasis on selling to business, rather than individual users made the DG/One and also-ran.

  28. frijole Says:

    @Dave Zatz: “Ah but the cube also had problems… cracks. Possibly also caused by heat?”

    In a word: No.

    They’re seams from where the polycarbonate flowed together inside the form. Much of the press at the time was quite skeptical of this cause, but its true.

    The heat dissipation on the Cube was great. Most of the internal chassis is a large heatsink that moves the heat to the center, and if you put your hand over the vent of a running Cube, you can feel the hot air moving out if it at a fairly rapid clip.

    One interesting thing, though, is the presence of a fan mount on the bottom of the chassis. This was rumored to be there for a fan used in the dual-processor version that was never released.

  29. spaceman spiff Says:

    The bit about reseating the chips on the A-III motherboard when they popped out of their sockets due to thermal expansion is true. I worked at ComputerLand in the Silicon Valley at that time and I remember getting the technical advisory from Apple about this, and how the “approved” method of reseating the chips was to lift the system box about 6″ above a solid desk or table, and let it drop flat to the surface… We were ROFLAO with that one! Amazing, but it worked! Since many of our customers could not believe that this was the correct “fix” for the problem, we made copies of the official Apple notice and gave it to all our customers who brought the units in for service.

  30. Ryan Says:

    You forgot about the Commodore 64 disk drive decable.

    They bodged the serial chip + firmware up on the C64, meaning that the read performance from the disk drive was only a fraction of what it could have been. Like 20% at best.

    Plus their drive implementation (1541) was so convulted that it ended up containing what was almost an complete computer.

    Still, the C64+1541 combo was cheaper and performed better than the equivelent Apple combo. But you don’t hear that nowadays, what with the AppleMafia going around rewriting history 🙁

  31. Anachronda Says:

    “No way to format disks” barely scratches the surface of the terrible awesomeness that is the RX50 diskette drive.

  32. Nathan Says:

    One additional detail in the mistake of the TI/99/4A sidecar expansion: The tiny speech synthesizer always had to be first in the line! (You can see it set up like that in the picture) If you even bumped the machine when in a configuration like that the connectors would easily move and disconnect, crashing whatever you were doing. If the speech synthesizer could have gone last it would have been a bit more stable.

  33. Jay Says:

    How about the first NeXT cubes which came with no conventional hard drive, only a removable magneto-optical drive. Very cool vision to have a cartrige with EVERYTHING you want on it… documents, applications, operating system.

    Unfortunately, the magneto-optical drives were very slow and noisy. This was further agrivated by a Unix based operating system that liked to “swap” out to the disk drive often.

    Luckily it was a simple fix to install a hard drive on your own.

  34. Scott Skaife Says:

    It is absolutly untrue that you could not format disks on the DEC Rainbow. I had one of these as a kid. We formatted disks all of the time.

  35. Ken D'Ambrosio Says:

    Hey — the Rainbow was actually kinda cool. And those floppy drives were easily the coolest *looking* of any out on the market. And trying to do both CP/M and DOS was a neat idea, even if the execution didn’t pan out so hot.

  36. nc Says:

    yes you could format disks on the rainbow 100. it was a nice machine. dont know why the author thinks it couldnt format floppies.

  37. Mark Anacker Says:

    It’s true that DEC didn’t provide any way to format blank floppies on the Rainbow, and they didn’t intend to. But an enterprising company in CA figured out how to do it, and sold a formatter utility. I vaguely remember DEC trying to put them out of business – guess the sale of high-priced floppies brought in a lot of cash.

    Bad marketing and poor technical decisions doomed the beast (as well as the rest of the company).

  38. Zenith Says:

    Zenith z-100 also did CP/M and MS-DOS. I can’t recall if it ran a Z80 or an 8085 for the CP/M side.

  39. rickb Says:

    No one recalled the Osborne? What a POJ.

  40. Dave Says:

    The 1541 is a great floppy drive. I never had a problem with the second version that included the ALPS drive. It was even greater when adding the parallel transfer expansions like SpeedDOS (we used to built our own reading out the data from an original chip and wiring the cable ourselves). The design flaw was that although everything in the drive was ready for 40 track operation some moron decided to only have it use 35 tracks.

    I learned LOGO on the Apple ][e and while it was a quite good PC for its time I absolutely hated it. Since it didn't have the OS in ROM you had to boot it from floppy, which made using this box pure torture and expensive, because without the two drive unit you basically couldn't use it. You also had to mount the floppy drives what was extra annoying, because a quick floppy change was not possible. Swapping floppies without unmounting and remounting opened up a big can of worms.

    The Amiga was an awesome system, but Commodore cut the corners a bit too tight. Having it ship with even a small hard drive would have made all the difference…uh yea, and investing into a decent MS-DOS expansion board would have made more sense than spending effort and money on bringing out their own PC clones. By the time an 8086 (80286?) expansion board was available the train already left the station.

    Also banking on a not very popular version of CP/M for the C128 didn't pan out to be a smart move, although at the time it was difficult to tell where the trend would go.

    Above all, Commodore couldn't sell water in the desert. That they still managed to make the most sold computer of all times (C64) is a miracle, which by itself was and still is a miracle. In regards to power consumption and processing capability it is still the "greenest" computer ever built.

    Lastly, a very recent design flaw is my HP printer that doesn't print b/w when the color cartridge is empty!

  41. Galvo Says:

    Even a small hard drive was really expensive in 1985. Especially a SCSI drive, it was possibly cheaper but much slower to buy an Amiga sidecar & put a 20M ST-506 drive in that with an Amiga partition.
    I remember how glad I was to hear later that the 70M drive someone bought for their IBM clone was SCSI & any bigger hard-drives would all be SCSI.
    Then another cost cutting that cost us plenty of hassles & waste. Compaq dragged PC compatibles back to PC BIOS dependant IDE alias ATA drives. We ended up with layers of disinformation and a succession of new limited standards & workarounds to get old x86 code BIOSs to believe in the smallest harddrive you could buy.

  42. swb311 Says:

    What about the Telex 1186 workstations? The only computer I’ve ever seen that used an Intel 80186 processor, and the unit took up two cases – one with two floppies (or one floppy and one mfm/rll hdd) and a power supply, and one case with the motherboard (and one 8bit isa slot!). They connected the things together with some type of DB50 connector (two if you had a unit with a hard drive) and a proprietary power cable between units. Very strange beasts.

  43. Tom Says:

    Commodore’s external 1541 disk drive may have had its problems (it got hot enough to keep your coffee warm, for one), but its internal design had a few advantages. For example, because it *was* a computer in its own right (memory, CPU, operating system, and everything), you could run programs on it that didn’t require the C64 (except to actually load the program into the 1541’s memory). One popular copy program allowed you to duplicate floppies by attaching two 1541 drives together. They sensed when you inserted the disks, and worked by themselves.

    The real problem with the C64, in my opinion, wasn’t the floopy drive, but the absence of a true UART chip to handle RS232 operations. Instead, the computer simulated the chip in software, which was why it couldn’t handle modems faster than 1200 baud.

  44. Andrew Says:

    You’ve got the wrong date on the Apple III – should be 1990, rather than 1980…

  45. Andrew Says:

    Hang on – I’m an idiot :). 1980 is correct

  46. Bob Ackerman Says:

    The TI 99/4A seems to have been designed by the defense contractors inside Texas Instruments. The Peripheral Expansion Box, which sadly you don’t show, could have taken a direct nuclear hit and survived. The cable to the PEB had an enormous connector that plugged in where you show the sidecar modules plugging in. It had a foot, a little downward-pointing stub, where the connector joined the cable to the PEB. That was because the cable was not just a ribbon cable, but a ribbon cable inside a heavy-duty rubber sheath, and the combination of the weight of the connector and the weight of the cable would cause the connector to unseat without the foot to support it. Inside the PEB, expansion cards were contained within snap-on aluminum casings that made the whole assembly — card and casing — nearly an inch thick. There was discussion in the TI 99/4A community about the necessity of those aluminum housings, and one user took a trip into the backcountry in his Jeep without the housings, and reported that everything held up.

  47. Todd Says:

    The PCjr’s IR keyboard was lots of fun in school lab settings, where Timmy could point his keyboard at Sally’s computer and type naughty things.

  48. The Lancashireman Says:

    I used a Lisa around the time when it came out. We never had problems with the floppy disks but the hard disk failed twice during its short life.

    In my book the worst design decision in the PC world was the decision to use active high interrupt requests on the ISA bus, thus dooming users to searching for some possible combination of IRQ jumpering so that a new IO card could be installed without upsetting existing cards. If it had been active low, cards could have shared IRQs.

  49. Archer Says:

    You thought the Osborne 1 was bad? What about the Osborne 2, announced months before it was ready, that sank existing sales, forcing the company into bankruptcy, and hence never shipping?

    This episode lead to the expression “Doing an Osborne”, meaning to sabotage your own business.

  50. Anonemaus Says:

    It’s a good thing Apple learned from their mistakes and never tried to introduce a non-upgradable computer again… oh wait… (article could stand a minor edit there)

  51. Jerry Chappell Says:

    Oh, another problem with the TI-99/4A was the reset key-sequence. Think of a standard keyboard with “=” where Shift= gives you the “+” character. Well beside the left shift key is a Fn key. And guess what Fn= gives you? That’s right, REBOOT.

    So many times I’d press that button while typing in a program from a computer magazine (a process that usually took an hour or more of typing, and you don’t want to save too often because it takes 5 minutes to save to tape), and press Fn= by mistake, forcing a reboot. Nasty.

  52. Marco Marrero Says:

    James: You’re right about the 68K but wrong about why.

    The PC was a cheap, experimental, “me too” project for the new microcomputer market. They selected the anemic 8088 (slow 8-bit bus variant of the 8086) because it was kinda CP/M compatible. Even the 2Mhz 8-bit 6809 in the CoCo3 was somewhat faster and could multitask.

    The generic hardware, documenting the ISA bus, and IBM licensing PC-DOS, made cheap clones possible.

    Ironically, IBM could have easily created their own micro and OS, they even had a 10mhz RISC CPU in 1981.

  53. blueguitarbob Says:

    How much of a geek am I, if I said that I owned most of these at one point? Plus multiple TRS-80s, an Osborne, and that Heathkit computer that you had to assemble yourself.

    I had actually forgotten about my DEC Rainbow. Good times.

  54. Robert Says:

    What a great article – thanks!

    The Amiga 500 had a side-mounted expansion port too – you could get hard drives and CPU upgrades in an expansion box that hung off the side of the computer. It worked, but it was really easy (for example) for someone’s girlfriend to accidentally separate the units with the power still on – frying the machine.

    I forgave her, though, because I went out and bought an A1200 to replace it. 😉

  55. Michael Tutty Says:

    In defense of the PCjr, only the IR thing was really a problem in any sense of that word, and as astutely noted in the article NO ONE had the resolution to work at any distance from their monitor in the 80’s.

    On the ports – if they had come out with proprietary ports in, say, 1995, I would agree it’s a bad choice. But the custom plugs made it impossible to set up incorrectly. Even now nearly 30 years later, my mom can’t set up her own PC, but she was the one who plugged everything into my PCjr. Maybe when everything is USB this will finally be licked, but it’s not there yet.

    On the sidecar – even the article notes that this was a design decision, in favor of not opening the case. This also passed the “mom” test, as nothing but USB has done since.

  56. farmkid Says:

    > Interestingly, we didn’t have the wireless/IR keyboard pictured here. Ours had a relatively normal non-chicklet, wired keyboard. Maybe IBM ditched that design at some point, or offered 2 versions?

    Yeah, when it became apparent that the chiclet keyboard was a disaster, IBM switched to a ‘normal’ replacement keyboard for new production, and offered a replacement free of charge to existing users.

    I worked in software development at IBM at the time, and attended a pre-release meeting at which the PC Jr. (and other unannounced products) were revealed in an attempt to get us on board. The Jr. was unanimously laughed at be the attendees, and the PC group couldn’t figure out why we wouldn’t get on board.

  57. nitsudima Says:

    My first PC was a PCjr, and it served me well. In addition to the side expansion ports, there was also a second 5 1/4″ floppy expansion kit (that also doubled the RAM to 256k, IIRC) that fit on top of the case. It was easy to install and only required you to remove the top of the case.

    As far as the keyboards go, IBM originally designed the chicklet keyboard to accept special overlays (look at all the space in between keys in the photo). The idea was to have software-specific templates fit on the keyboard to help users find shortcut keys.

    Backlash from the poor feel of the rubber keys soon caused IBM to abandon this keyboard entirely, and by the time we bought ours (late 1984, I think) it came with a pretty good standard feel (corded) keyboard right out of the box.

    In my opinion the PCjr’s biggest flaw was its lack of compatibility with the regular IBM PC. While you could get some business apps for it (Lotus 1-2-3 came on a cartridge!), there were many more programs that simply wouldn’t run on the jr.

  58. MadMorf Says:

    Another really annoying thing about the Mac 128k…

    No cursor movement keys on the keyboard!

    You were forced to position the cursor with the blasted mouse!

  59. Marc Kenig Says:

    The RX50 on the DEC Rainbow was a really poor mis-feature. I remember how, when removing a diskette the doors would flip open and catch your finger. Ouch. Even though you could format diskettes those were costly single-sided QUAD DENSITY disks – with very few third party vendors. The bottom one had to be inserted upside down. (The drive head mechanics were shared in the center of the drive).

    Reading and writing IBM-PC Double density disks was possible after a while, but only single sided disks formatted on the PC.

  60. mistermeta Says:

    And then there was the Hyperion, designed to be IBM PC compatible but some enlightened engineer decided that the video hardward should be “better” and made that incompatible, therby sinking the launch. Compaq ate their lunch.

  61. mistermeta Says:

    And then there was the Hyperion, a portable PC meant to be IBM PC compatible. Unfortunately it had a graphics processor that wasn’t – it was “better”. Compaq ate their lunch.

  62. Grishnakh Says:

    Lou,

    That’s incorrect. I had a TI-99/4a, and I remember us buying it at Toys R’ Us. We also bought later accessories from there, and from some mall store (JC Penney I believe). I certainly don’t remember buying it from some stupid multi-level marketing company.

  63. Mike Says:

    The best thing about the DEC Rainbow is that is had a VT220 emulator in ROM, and could be used as a VAX terminal as well as a DOS machine. That says something about its intended market.

  64. Dan Says:

    I guarantee that the Atari 400 membrane keyboard (also known as the PB&J keyboard, because you could wipe jelly droppings off of it with a damp sponge) was by far the WORST keyboard ever made. Each keypress took a good bit of pressure, making it impossible to build up any speed.

  65. George Says:

    How about the IBM PS/2? It had a great Microchannel Bus architecture, with advanced features for greater throughput. However, IBM threw the older, more open technology of the IBM PC ISA cards away for MCA which was so much better from a technological point of view. However, the higher costs of MCA and lack of backward compatibility were its biggest drawbacks. One could not plug an existing ISA card into an MCA slot. So if you bought a new PS/2, you also had to replace any existing adapter cards from your old PC. Since IBM held a much tighter reign on its MCA technology, third-party adapter cards were much less available and those that were on the market were significantly more expensive than the equivalent ISA card. But even despite the MCA problems in the market, the PS/2 models were priced much, much higher than PC clones of the day. Eventually, IBM brought back the ISA architecture in its PS/1 line of computers.

  66. Abbas Halai Says:

    My biggest issue with computers in general and I’m not sure how old this problem is or has been is the design or rather the placement of the Open and Close button on CD/DVD-ROM drives. Why is it below the tray? It should be above the tray to get to it easier. I’m surprised nobody mentioned this before. It’s my biggest issue with these trays, especially if you own a desktop which sits on the floor well below your desk. You have to reach out at a weird angle under the tray if it is open to close it rather and this usually ends up people harshly pushing the tray back in.

  67. Leo Batfish Says:

    @MadMorf: The original “skinny Mac” (128K) did have [left] and [right] directional keys — just not [up] and [down]. The reason given at that time was that Steve Jobs wanted to force developers to include mouse control (which they might otherwise have preferred to omit in order to make the porting job simpler). He may have been right . . . though as an owner, I wasn’t very happy about it. Just as early buyers of iMacs weren’t very happy about needing to buy an external floppy drive (because contrary to Steve’s statement at the time, many people still WERE using floppies back then). But at least it wasn’t as stupid as no-floppy on the Next. Or the round mouse.

  68. Drew Says:

    I wonder what a “The Best PCs That Didn’t Make It” would look like?

    I wonder what a “The Best OSes That Didn’t Make It” would look like?

  69. Frogg Says:

    Another note about the Adam. Back around (I think) 84 or 85, they had a good Xmas promo – If you bought an Adam, you would get a Cabbage Patch doll free. Guess which one was most in demand that year…

  70. Eric Durant Says:

    There actually was a large following of the TI-99/4A which used the built-in BASIC; numerous programs in it could be found in the hobbyist magazines for years. I’m sure the long and rigid expansion bus affected many users, but I don’t know anybody who approached using all possible peripherals. You certainly wouldn’t judge a modern computer by whether it could easily and simultaneously use all the peripherals available for it! The little speech synthesizer seemed to be the most popular, and hobbyists (well, all the users I knew) more often than not used audio cassette tapes to store programs via a simple cable. It was slow, but this was 1981. The Apple ][+ did all this better, but the TI was MUCH, MUCH less expensive. The TI-99/4A made an outstanding price/performance compromise for 1981, opening up home computing to many people who otherwise would have not had access computers for many years. I’m biased–the TI was my first computer and was an excellent introduction to what was possible.

  71. geekisrael Says:

    For some reason I didn’t see anything about Sinclair Z-80,z-81,spectrum computers etc(in the US they were called Timex) state of the art z-80 based computers, with terrible flat keyboards, good tape connection, their biggest mistake was trying to look like gadgets and not like computers with tiny 2” thermal printers (paper was just too expensive), instead of disk they had a close loop tape called wafer, and much more
    But man, how much programming I could do then in basic, having 1K RAM, and the expanded to 16K.

  72. Old Soldier Says:

    “No one general-purpose computer configuration can meet every user’s needs, though, so this idea ultimately failed in the marketplace.”

    What about today’s laptop? They are not exactly very configurable. Look how popular they are now. Any day now there won’t be desktop machine as we know it.

  73. Patrick Lajko Says:

    You forgot the Commodore 128. It was a mistake because:

    1. They stayed with very slow serial floppy drives. Although 1200 baud, faster than the Commodore 64, everyone else was going standard parallel drives.

    2. Adding the Z-80 was a great idea. Unfortunately they ran it at 1mhz while everyone else was 4mhz and some running it up to 10 mhz.

    These two things made what was a great idea a real slow turd.
    What were they thinking?

  74. Oldstone54 Says:

    The Lesson:
    Proprietary = Bad
    Open Standard = Good
    In the long run proprietary does not work. Hollywood should have to take a course in computer history. Of course every now and then someone has a bright idea… Those who do not understand history are doomed to repeat mistakes.

  75. realenterprises Says:

    no mention of the Atari 800?

  76. Joe Says:

    The TI-99/4A’s BASIC was slow for a multitude of reasons. The machine really was a study in bottlenecks. I’ve studied its architecture somewhat and have spoken to some of the engineers that were there at the time.

    It’s built in GPL interpreted language was partly a consequence of the fact that all of the RAM in the machine was connected to the video controller and not directly CPU addressable. This was a cost saving technique with horrible repercussions. The Home Computer team also wanted a custom CPU that ran GPL directly, and were forced to use the 9900-family CPUs instead. (There’s a long story there, too, about how the 9900 wasn’t all the CPU it could be either…)

    In addition, the BASIC interpreter was *purposefully* slowed down. The BASIC had delay loops to slow down screen output so that it wouldn’t be too fast for children, apparently. If you turn up the volume on your TV set, you can hear the CPU drop into those delay loops after every PRINT statement.

    TI Extended BASIC thankfully eliminated a number of those delays, though it still was far from being a speed demon since most of it was still double-interpreted. I still don’t know why it did a “prescan” (I’m guessing it had to do with the highly flexible and extensible nature of its implementation, since add-ons could link their own subroutines into the language), but I do (un)fondly recall shoving a dozen lines at the start of my TI XB programs bracketed with !@P+ and !@P- to turn prescan on and off so it wouldn’t freeze for 3 minutes when I typed “RUN”.

    (That blob of code had to contain at least one instance of every subroutine and variable name used in the program. It didn’t have to be syntactically correct though. You could have line 1 be “GOTO 100”, lines 3 through 98 be filled with gobbledegook with the !@P+ and !@P- “comments” on lines 2 and 99. (Un)fun times.)

  77. James Says:

    Marco — Amen about the 6809. (I still have my CoCo 3, w/6309 and SCSI cartridge, booting OS-9, or rather NitrOS9, and running on a SCSI Zip drive.) The horrid corner-cutting Radio Shack committed on the CoCos was a shame and a sin; ask anyone who ran a GIMIX system with OS-9 Level 2 and the smart I/O cards available for it.

  78. RoverDaddy Says:

    @Bill: I too remember the problems with DB-25 connectors. Once while I was employed at IBM another young engineer managed to plug a serial device into a PC parallel port, resulting in a dead computer and smoke coming out of the vents. A quick look at the schematics (IBM used to provide schematics for their boards in the Technical Reference Manuals) showed that this particular cabling snafu delivered +12V to a 5V -output- port, which the parallel interface did not seem to enjoy in the least.

  79. Phishfarm Says:

    I remember attempting to use an Osborne, which had the incredible knack of powering down at exactly the worst possible moment. Adding insult to injury, they wouldn’t hire me, even with an inside connection. All for the better, I suppose.

  80. Vic20 Dave Says:

    This is not true

    ” Originally, the 1540 drive was intended for the VIC=20. When the 64 users demanded a floppy drive, the 1540 had to be modified to work with (as I recall) the somewhat faster C64 serial line. ”

    In fact the 1540 was faster than the 1540 and the VIC20 was faster than the C64.

    You had to turn off the Video output on the C64 to make it fast enough to cope with tape and disk loading. With the video on the CPU was interupted too often and the timing required to load a tape or a disk was not possible.

    For speed of operation give me a VIC20 over a C64 any day 🙂

  81. trrll Says:

    The Apple II also had a problem with oxidation building up on the socketed chips over time. I don’t think it really had much to do with heat. Part of my standard Apple II maintenance was once a year to take a screwdriver, pry each chip out an eighth of an inch, and push it back in. This cleaned any oxidation of the pins, and kept the computer running reliably. Socketed chips were nice in that it was easy to fix the Apple II if a chip went bad, but it did not provide the no-maintenance reliability required for a business system. This kind of problem vanished as the industry transitioned to chips soldered directly to the motherboard.

  82. Pat Says:

    blueguitarbob,
    You want geek? I still own two TRS-80’s and two TI-99/4A’s. One is still in the original box! We’ve also got a few random machine my dad built back in the 70’s that were all done with wire wrap. o/ I’ve got some PDP-11 parts lying around too. 😀

  83. jpamado Says:

    I agree with geekisrael about the Sinclair ZX-81. However, talk about a wrong design decision – the original 16 K expansion module connected directly to the PCB and was extremely sensitive to even the slightest touch. So, a long programming session could easily end in rage, or tears, just by touching either the computer or the memory module.

  84. Mike James Says:

    You should have bought the Memotech 16k expansion pack – better shaped and had velcro to stop it moving 🙂

  85. Moss Bliss Says:

    What they didn’t mention about the TI-99/4A expansion box (among other things — the device was barely mentioned) was that it was just that. A box with slots. And they charged 3 times as much for that as they did for the TI-99/4A itself (the price later fell to only twice as much).

  86. Dave Says:

    Drew said:
    I wonder what a “The Best PCs That Didn’t Make It” would look like?

    I wonder what a “The Best OSes That Didn’t Make It” would look like?


    Best PC probably would have been the C65 or the C128 if they had chosen to implement DOS instead of CP/M.

    Best OS that went nowhere hands down is BeOS. Too bad that it was just that, an awesome OS with hardly any applications. Second in line is OS/2 Warp, which had its trapdoors and successes, but if it would have been maintained by people more capable than big blue chair farters it would wipe the floor with both Windows and OS X.

  87. Nate Says:

    Re: Drew

    1 Answer to your 2 questions.

    Amiga.

  88. Gerry Butler Says:

    Hey! I had an Aquarius. I absolutely loved it. It was great to program in basic as the keyboard had an overlay that let you “type” all the BASIC keywords using CTRL-key combinations. For example, to type something like:

    for x = 1 to 100 step 2

    You had to only type: “Ctrl-F X = 1 Ctrl-T 100 Ctrl-S 2”. Quite fast and handy.

    I truly loved programming on this machine. I wrote neat graphics programs, games, played music. It was spiffy. You really don’t know what your are talking about if you think it was crap.

  89. JohnInMableton Says:

    C=64 rules! It has everything! And what it doesn’t have, you could add on and adapt. Still a wonderful game machine for those old games. You could load up a word processing program like PaperClip and produce documents on an HP Deskjet or laser printer that no one could tell the differnce from any professional type set service.

    My Dad gave me one, he was handing them out to his bank customers to use for payroll, loans, online banking etc back in 83. They hired some Nasa guys to do the encryption. They had people flying in from all over the world to see what that was all about.

    I suppose that’s why its not listed here, there were no design flaws at all, just marketing flaws. No one could accept that a machine that played games and could process encrypted bank transactions could be truly good at either. But it was! Now thats not a question.

  90. Tezza Says:

    Good article and I agree that these were all design flaws. The “are-we-there-yet?” slow 1540 Commodore disk drive and the “move-it-ever-so-slightly-and-you-die” ZX81 expansion pack should certainly be up there with these others.

    However, care needs to be taken when discussing design flaws vrs technology limitations at the year of release. The former are things which could have been avoided at the time with an alternative approach. However, some machines a year or two after release might APPEAR to have design flaws, but their design was probably the best that could be done at the time.

    The Osborne 1 for example, was a groundbreaking concept and did the best it could with the technology of the time. Although the screen was tiny, it had to be to accomodate those huge full height floppy drives. I wouldn’t say it had significant design flaws..it was just pushing the envelop at the time. Of course a year or two later, 1/2 height drives meant a much large screen could be dropped in those luggable cases. Enter Kaypro II.

  91. honeymak Says:

    but nowadays…….computers are thought of as appliance at home
    usually owning 2 @home……

    mistakes….sometimes just a problem of “timing”….not just time to market……XD

  92. Scott Marlowe Says:

    Rocky, the Amiga had slow / fast floppy drives depending on how you looked at it. They were noticeably faster than pc floppy drives when reading and writing large files, but random access was much slower. Why? They worked in a really interesting and strange way. Instead of having a start of track mark, a data block, then a sync mark, then another data block, then another sync mark and so on, the had no real sync block (they did, but it was mostly just empty space, and not used like regular sync blocks). On the amiga, when the OS was told to read a single block, the disk drive was spun up and the whole track, sync blocks and all was read in. Then the Agnus chip had a track decoder / encoder routine built in that would decode the whole track (I believe from MFM encoding but it’s been a while.) Writing would be accomplished by doing the same thing as reading, to get the blocks into memory. Then the block would be changed, and the whole track re-encoded and then wrote out at once.

    This design meant that that the tracks were not aligned with each other, or anything for that matter. So, preformatted DOS floppies came along, and someone figured out that if you had a fancy program on a pc, you could format around bad portions of the disk, and they’d work fine until some idiot reformatted the disks. The Amiga basically reformated each track every time it wrote to it and you’d see whole boxes of formatted disks thrown away because of bad sectors.

    OTOH, I could copy one floppy disk to another, while browsing the web, printing a document, etc and the disks would copy pretty much just as fast as if it was the only thing I was doing and would not interfere with the rest of the machine, something PCs had a hard time with for another decade or more to come.

  93. Eelco Says:

    The DEC Rainbow had another interesting feature: the twin floppy drives were mounted such that you had to insert the floppy disk with the label up in the top drive, and with the label down in the bottom drive. It did have one of the nicest keyboards I’ve ever used though and the displays were pretty good too (except for that one time when a new DEC service engineer did some preventive maintenance on a bunch of them; over the next couple of days several went up in smoke…).

  94. James Childress Says:

    The TI 99/4a was my first PC and I have to say I had a lot of fun with it as a kid. What eventually spelled the end of that system for me was the poorly design video cable. It took little force to break the tiny soldered connections inside the plug. My mom bought my system for me from JC Pennys catalog when they were being discontinued so that made it difficult to find replacements. I was lucky to find a second one but it too died after about few months from wire breakage. This could be a design mistake added to the list. A simple stronger solution would have been an RCA type plug and would have only needed an off the shelf RF adapter such as what the Atari 2600 used. I did enjoy the fact like the Atari systems you could use it as a cartridge based game system. Many of these cartridges actually added memory to the system. Saving was frustrating and time consuming with a tape and I had at least one occasion where I regretted not saving often. I had spent a couple hours typing in a long program and was nearly done when the power went out. My parents didn’t appreciate the scream I let out as a result.

    I had the opportunity to play with a PCjr in the early 90’s and after finding how useless the system was decided the make use of the monitor as a video display unit, but of course it was one of the propriety components. IBM couldn’t just use a normal connector so you could use any monitor or use that one with something else. The connector attached to the monitor had about 12 wires coming from it and no one I knew if it could be adapted. IBM most likely would not have shared such information anyway much less make an adapter to extend the useful life of the monitor.

  95. Chris Lees Says:

    It was actually possible to upgrade the RAM on a 128k Mac. Apple engineer Burrell Smith added a few extra lines on the PC board so customers could replace their existing RAM chips with the new generation to give a total of 512 kilobytes. It required a soldering iron, and nobody told Steve Jobs, but it was definitely designed this way.

  96. Alan Says:

    My memory of the 128K Macintosh was that it did have Appletalk, and you could buy a laser printer to go with it (for $3,000) instead of a 144 dpi B/W printer. But, unfortunately, 128K of RAM wasn’t enough to use the printer. You needed to wait for the 512K to be released to use the laser printer.

  97. Bill S Says:

    IBM = mainframe, DEC = minicomputer, young’n.

  98. Andrew Says:

    I’d put the Osborne 1 as one of the best, not worst. Maybe rickb had a lemon. In 1981, when my workplace had PDP-11’s and fledgling rackmount 8080 systems with 8″ floppies, I had a computer I could pick up in one hand, take as carryon on a plane, and use anywhere. It could run on batteries and the serial interface gave access to mainframes, bulletin boards and email – basically it did the job of a laptop, but 28 years ago.

  99. georges Says:

    You’re forgetting one thing regarding Lisa:
    It was the development platform for The Macintosh, and as such extremely successful!

    The original macintosh with 128k of ram was just too small to be used to code modern software (Pascal, with Apple GUI).

  100. Simon Kenyon Says:

    the biggest mistake was the design of the plate on PCI/PCIe cards. the little flange on the top means that the back of the case is overly complex. this adds a lot to the cost and restricts innovation.

  101. John Says:

    Well nice article except one thing.
    They forgot the Biggest mistake ever made.
    The Name:Commodore Amiga.
    2 “little” mistakes just forced probably the whole planet to follow the PC-compatible evolution
    Mistake #1 The infamous “Guru Meditation Error”.How it is possible to hire someone to design the next generation OS and just before the product is released fire him . The result: the OS was half finished and numerous bugs the writer intended of fixing remained unfixed.The business market reacted with NO business apps.
    Mistake #2 Lack of what makes a “pro” system.Limited expansion capabilities just lead to the common believe that this was just a “game” machine.
    Few remember the Autocad Clone that was running 6 times faster or the amazing Sculpt 3D plus other numerous applications that surpassed after MANY years

  102. Robert Says:

    The Sinclair ZX81 (sold by Timex)
    This was a very good machine for it’s (very low) price. It only had four chips in it total (Z80, RAM, ROM, ULA) when it’s use of that ULA chip was very innovative. The way the screen was generated was a weird and evil hack, but it probably knocked 20% off the price. I personally never suffered from ‘rampack wobble’. As for the printer, I ended up with a 3rd party thermal printer rather than the ‘sparky’ one.

    I think the only thing that actually annoyed me about it was a tiny technical detail in that the internal static RAM chip was attached to the DRAM bus for the rampack rather than the ROM bus. If it had been attached to the ROM bus everything would have worked AND when you had a rampack you would have got user defined characters for free. Ho hum.

  103. Robert Says:

    C=64
    This was a very nice machine indeed, everything built into the machine worked very well in general. But. Problem One, rather than having specific static video ram like the previous PET machines the C=64 could put it’s screen anywhere in the DRAM. Unfortunately the DRAM wasn’t fast enough for this so the CPU had to be disabled for 64us about 1200 times a second when the screen was on. This interfered with the tape drive to the extent that the old PET routines for the tape unit would fail. This was made worse because the tape was actually a downgrade from the old PET one. So screen was turned off when the tape was on.

    The serial bus (for the floppy disk) didn’t have this problem. It’s problem was that it was supposed to be hardware assisted and supposed to run at anything upto about a megabit per second. There was a problem with the 6522 chip that had the serial shift registers, they couldn’t fix it so the serial bus became a completely software device. In effect it was crippled.

    Both of these issues could have been fixed given time; but there was no time. There were already machines around with faster processors and no prizes for second place.

  104. Mark Says:

    The Macintosh 8500 would top my list. For its time, it was a powerhouse, but like many of Apple’s products over the years, it came with too little RAM and too small of a hard drive. The problem was, upgrading the RAM in this machine was a terrible task sort of like replacing the fan belt in a modern car. Later, Apple got it right with the G3 and G4 cases which made memory and drive upgrades incredibly easy. Unfortunately, they have reverted to near 8500 difficulty with the Mac Mini.

  105. Frank O Long Says:

    Looks like Apple did me a favor by refusing to accept a Credit Card for purchase. I ended up “special ordering” a 32K Ram Color Computer (COCO-1) and its processor (6809) ran circles around the IBM at the time.
    Still have and use the Coco-II and Coco-III.
    Must have been great PC’s as NASA used them for the space program…

    Frank O

  106. Alan Wendt Says:

    Or the TI 790, with the 31-bit word (the sign bits of both 16-bit words were forced to be the same), and the horrid thermal printers that all the big-box stores now use to print receipts.

  107. Gordon Schumacher Says:

    The reason that Microchannel cards were so *@%*@!! expensive – and why basically nobody else but IBM ever adopted them – was not a technical issue. It was twofold: first, the *mechanical* design was a flying pain in the rear; and second, if you wanted to use the spec – either to make a card or a motherboard – you had to cough up a goodly wad to IBM for the privilege. Yup, it’s VHS vs. Betamax all over again… and indeed, same result.

  108. Raybo Says:

    The Original Amiga 1000 had the problematic “sidecar” expansion concept. Commodore called the bus “Zorro”. With the release of Zorro 2 in later models, the original became known as Zorro 1. Whether due to poor design, poor Commodore documentation, or just ‘orneriness’ of the vendors, there were always major problems with trying to get multiple sidecar modules to work together. Some needed to be left or right of others, some just wouldn’t work with certain others, etc. When the Amiga 500 was released, the bus was moved to the other side of the unit as well as set at a different height due to the case design, and the bus had changes as well. This made all A1000 modules incompatible with the A500.

  109. Dan Rocha Says:

    I used to service Apple III computers, and I think the ‘pick it up and drop it’ story is apocryphal. First of all, it’s unlikely to work as the affected chips are more likely to bounce out of a socket than in. And second, there were service bulletins suggesting ‘pick it up and drop it’ for the Apple Graphics tablet, which addressed wires sticking together and that makes me think someone has blurred the two stories.

  110. Roger Baker Says:

    I had a Commodore 64 + 1541 disk drive and a Atari 400. I think I upgraded the Atari memory to 32K. In the movie Airplane, the air traffic controller is using a Commodore computer. I used to buy a magazine with computer programs in them and I typed them in page by page, line by line. Eventually, they came up with a checksum system telling you if you typed a line in wrong. I learned some programming as I had to hunt my errors down and make changes.

  111. Frank O Long Says:

    Worked for GSA, ADTS, ELectronic Services Div, Region 3, Wash DC.
    Early 80’s we bought three SUPERBRAIN-64K PC’s. I installed one in the GSA Central Office for the clerks to use. Only took them about a half hour to scramble the CP/M I had loaded. Culprit was an electric pencil sharpener plugged into the same branch circuit. PC had no immunity to A/C line noise.

  112. Robert Says:

    As someone else pointed out, the non-standard use of DB connectors was a huge mistake on the IBM PC (perhaps this doesn’t qualify because the PC was an industry-spawning success). IBM chose a DB-25 for parallel port instead of the standard Centronics connector, and they then used a DB-9 for serial RS-232 instead of the standard DB-25. Argh.

    And no mention of Eagle Computer’s 1600? That’s a classic mistake! They had been a CP/M microcomputer company, but came out with the 1600 which was an IBM PC *compatible* (not clone) which was MS-DOS compatible but not a complete hardware clone. Most software for the IBM PC would not run, and I’m not sure if it accepted the same plug-in cards (probably not). It did not sell well.

    Another Eagle classic mistake was to have it’s president get so pumped up on the day of the company’s initial public offering that he crashes his Ferrari and kills himself.

  113. Artless Says:

    I have long been puzzled that none of the commentary on the Lisa that I have ever read has even mentioned what I considered the greatest defect of the Lisa: that no Lisa would deign to read or write a floppy that any other Lisa had formatted. This made it nearly impossible for a group of programmers to develop a software system on a group of Lisa’s unless they paid Apple several thousands of dollars for Apple’s networking “solution”; sneakernets were next to impossible. Apple claimed this was intended to protect their software copyrights, but that was never credible; it was clearly intended to force purchase of Apple networks, which worked every bit as badly as most early attempts at networking.

    The manager of the project on which I discovered this “feature” simply returned all three Lisa’s and bought PCs, one of the most realistic decisions I ever saw a manager make.

    While reading this article the reason for this omission finally occurred to me: most likely no reviewer has ever had more than one Lisa in hand at the same time.

  114. EricF Says:

    This article was cool and nostalgic..
    but it forgot a few others..
    the Commodore +4 and Commodore 16.
    Both of those were introduced after the C-64. what a mistake that was.

  115. Tux Says:

    Fantastic article, and the comments are an interesting cross-section of computer history.

    My first machine was an Apple II+ which had been used by a family of mice as a den. It would smell faintly of animal waste when it got warm. Ahhh… memories.

    Probably the worst machine we had in the house over the years was the Apple Macintosh Performa 450. This was the era when Apple was putting out all sorts of different, oddly named and numbered configurations, this being one of the least expandable. It worked well for a time, but left you no room for growth.

  116. mike Says:

    Some of the things that have been identified as mistakes were really intentional (spelled “thought out”) compromises.

    That being said, I somewhat miss the era when there was so much diversity in computing hardware and software. In that environment, there was much more innovation–at least when compared to today’s standardized commodity computer market. Today there is too much need to conform.

  117. LinuxNut Says:

    My 1st pc was the TI 99/4a – loved it. For every idiosyncrasy, it had some upsides. The article doesn’t delve on this, but the CPU was the same 16 bit part you’d find in their mini-mainframes (TMS 9900?). The unit had decent video as well – complete with hardware sprites. I spent way too much time on it gaming, but did learn to program basic and even some assembler. Yes, it would support a hard drive. Hands down one the better home pc ‘attempts’ given the timeframe… they just did the modular hardware thing like everybody else was doing in ’80.
    As somebody commented, the PE (peripheral expansion) box was built like a tank. IIRC, there was a followup machine in development that would have really surprised techies, but alas the home pc market was getting crowded so it was dropped before production.
    Did anyone mention the Tandy systems with the 80186 processor?

  118. Clark Bunch Says:

    @Mike;
    that conformity you hate has definate advantages. If you hand me a flashdrive with a file on it, I know with 100% I can plug it into my USB port and be able to read it. It (almost) doesn’t matter what brand of computer I have, who sold it, nor when it was made, I can open the fild and read it. If I’m missing codex, drivers, etc. my machine will identify and download them immediately. Universal compatibility (computers, projectors, printers, Internet) is the benefit, and it makes the “conformity” worth it.

    Apple, can you hear me?

  119. karl Says:

    The original IBM PC was an offshoot of an I-B-M small business computer program in the 1970s /80s….. the original computer was named a 5100, soon to be followed by the 5110 & 5120 models. These were powered by INTEL 8080, 8085 CPUs, running at a blistering 2 MHz clock. In the original models, there was only a max of 2 8″ Floppy diskette drives. These “machines” were built in a desktop format w/ the keyboard an integral part of the machines. The only options were a few wide-carriage dot-matrix printers. The “final” model in this series of “small office” machines was the DATAMASTER, aks the System 23. It too featured an integrated screen, keyboard, although there was a variant model, the 5322, which had the CPU, diskette drives in a separate floor model cabinet. Only the screen & the keyboard were on the desktop. This model was also powered by the INTEL 8085 chip, and featured a cut-down version of I-B-M’s BUSINESS BASIC, similar to that found on the SYSTEM 3x mini-computers of that era. The interesting feature of the SYSTEM 23 was that there was a central 15 MB or 30MB hard disk accessory, which was also a floor-standing cabinet (about the size of a two-drawer file cabinet. Thes hard drives could be shared w/ a max. of 4 “workstations” connected to the unit. There was also 3 different printers available, one dedicated to letter writing since it featured a fine-print head, and had feeder options. The other two printers were wide carriage units but also had a fine-print mode. What makes this system interesting is that the screen guts was on a separate chassis that came undone from the main system chassis. And, it was virtually identical to the original green screen of the I-B-M PC introduced in August 1981. The BASIC language implemented in the SYSTEM 23 was fairly robust in that it fetured a very powerful SORT function. It also had a very powerful INDEX function which was a VERY GOOD feature, since w/ a CPU speed of 2 MHz, reading through hundreds (or thousands) of records to retrieve the desired one could take forever. The file system was also borrowed from the SYSTEM 3x platform in that one could access a file for SEQUENTIAL, RELATIVE, or INDEXED access, and records were composed of character data, numerical data, and the PD format (packed decimal) which saved a ton of precious HARD Disk space. The SYSTEM 23 also had room for two 8″ Floppy drives built into the main chassis. The similarities between these office machines and the original PC are numerous.

  120. ZonS Says:

    Too bad for the big prices

  121. nealoren Says:

    Nice post, I am not aware of this thing.we discuss on 15 classic pc design mistake.but at that time his is very popular and its need of customer as new feature require it modified. no electronic device is perfect after some year we have to add or create new feature in that device.but according to time its important but according to money its loss.so while designing and manufacturing company note out this thing..

  122. Tanith Says:

    More recently there was the Apple Titanium 17¨ whose hinges suffered metal fatigue and snapped off! There were only helín by epoxy and one screw anyway!!!!

  123. mike Says:

    OK, here’s a modern example.

    Why can’t I use a Kindle to:
    1.) Surf the web
    2.) Send an email.
    3.) Write a book review
    4.) Make a phone call.

    –mike,

  124. OlsonBW Says:

    “swb311 Says:
    June 15th, 2009 at 9:38 am
    What about the Telex 1186 workstations? The only computer I’ve ever seen that used an Intel 80186 processor, and the unit took up two cases – one with two floppies (or one floppy and one mfm/rll hdd) and a power supply, and one case with the motherboard (and one 8bit isa slot!). They connected the things together with some type of DB50 connector (two if you had a unit with a hard drive) and a proprietary power cable between units. Very strange beasts.”

    As a desktop computer maybe. I’ve got a 3Com LAN Manager server that has a 80186 CPU chip in it. It might still run. It hasn’t been booted up since the late 80s.

  125. OlsonBW Says:

    “George Says:
    June 15th, 2009 at 12:30 pm
    How about the IBM PS/2? It had a great Microchannel Bus architecture, with advanced features for greater throughput. However, IBM threw the older, more open technology of the IBM PC ISA cards away for MCA which was so much better from a technological point of view. However, the higher costs of MCA and lack of backward compatibility were its biggest drawbacks. One could not plug an existing ISA card into an MCA slot. So if you bought a new PS/2, you also had to replace any existing adapter cards from your old PC. Since IBM held a much tighter reign on its MCA technology, third-party adapter cards were much less available and those that were on the market were significantly more expensive than the equivalent ISA card. But even despite the MCA problems in the market, the PS/2 models were priced much, much higher than PC clones of the day. Eventually, IBM brought back the ISA architecture in its PS/1 line of computers.”

    I worked for a bank and LOVED supporting the PS/2s. The main downside I saw was that they were 10mhz at a time when everything else was at least 12 mhz.

    They were much easier to support and support is what costs most in computers, not the initial purchase. I’d say it probably cost about 10% of the support costs that ISA PCs cost because pretty much everything just worked.

    Remember jumpers on accessory cards? No
    Remember having to put accessory cards in a special order to get them to work? Not on a PS/2.
    Remember all the fun with DMA settings? Not on a PS/2.

    I could go on and on. I agree that it cost more in the beginning and the CPUs weren’t as fast. But after that, they were a dream to support.

  126. OlsonBW Says:

    “Frogg Says:
    June 15th, 2009 at 12:59 pm
    Another note about the Adam. Back around (I think) 84 or 85, they had a good Xmas promo – If you bought an Adam, you would get a Cabbage Patch doll free. Guess which one was most in demand that year…”

    They could have called it the Eve doll to go along with the Adam computer.

  127. Scott Marlowe Says:

    Speaking of 80186 machines, I also worked on one, the Burroughs B-28 cluster system. They were popular in the USAF in the 1980s. I even wrote a program for ours to keep track of student grades called classbook… Been way too many years. I think our workstations had 128k of ram, and we had a shared 3x35Meg Atasi hard drive storage unit. Atasi hard drives, the bane of my life back then.

  128. Pops Says:

    A bit of folklore: The floppy drive on the DEC Rainbow was the RX-50, purchased from a tiny startup called “T&E.” After the acquisition of T&E closed, DEC found out that the name stood for Trial & Error.

  129. 80sKid Says:

    re: PC Jr – While the PC Jr was a flop, it did have better graphics (than cga) and sound (3 voices) which was picked up by Tandy in their decently priced and popular Tandy 1000 machines. Many games supported it. EGA was expensive back then.

    re: PCI slot plates – They are the same plates as was used on ISA cards and date back to the original IBM PC. There just hasn’t been a point in time where it was wise to change this.

    re: 80186 – Tandy 2000 used it too, pretty sure it had some major compatibility issues. The 186 was widespread for embedded use tho

    I know the Tandy line too well from drooling over Radio Shack catalogs as a kid. Had a 1000

    Also had the very first Mac, hated it, but fondly remember the unique sounds of the disk drive hum…almost musical. The ImageWriter printer had a unique sound as well.

  130. shatunovaddict Says:

    Is it bad that I want to own all of these computers, just because? ^_^

  131. James Parry Says:

    Number 10 sounds like the xbox 360 too me…

  132. Don Says:

    What about the Altair, the Amssai, the Comodore PET, 64 and 128. I had the TRS-80 model 1 and upgraded the 4k model to 16k, added the 32k expansion interface and two disk drives, the unauthorized lower case mod and had one heck of an exensive computer. That SSSD disk drive cost $800. I remember having to drive to San Francisco to buy my first box of 10 diskettes for $50. We used to cut a notch in the other side so we could use both sides and save $$. Still have the TRS-80 Model 1, a TRS-80 Model III, and a Commmodore 128. (and many others) Remember the LDOS to replace TRSDOS. Those were the days. How about the North Star Horizon w/ CPM?

  133. Ben Says:

    Joe said of the TI-99/4A:

    “It’s built in GPL interpreted language was partly a consequence of the fact that all of the RAM in the machine was connected to the video controller and not directly CPU addressable. This was a cost saving technique with horrible repercussions. The Home Computer team also wanted a custom CPU that ran GPL directly, and were forced to use the 9900-family CPUs instead. (There’s a long story there, too, about how the 9900 wasn’t all the CPU it could be either…)”

    The TI had 16k Video RAM (for the graphics chip) and 256 bytes CPU RAM internal. There was a 32k expansion card for it as well. The 256 bytes and 32k were directly addressable by the CPU. I’ve never heard that they wanted a chip to execute “GPL” – always thought it was an “abstraction layer” to make it easier for programmers coming from other processors.

    “In addition, the BASIC interpreter was *purposefully* slowed down. The BASIC had delay loops to slow down screen output so that it wouldn’t be too fast for children, apparently. If you turn up the volume on your TV set, you can hear the CPU drop into those delay loops after every PRINT statement.”

    Shame to say, but it didn’t have delay loops. I’ve the GPL source code for TI BASIC, and if it had delay loops, I would’ve removed them.

    As for the Extended Basic prescan? The TI did a check of the program and built variable tables to speed up execution. Even TI Basic does this. Imagine how slow it would be without it?

    TI Basic was a factor of 3-5 times slower than other Basics on average.

  134. Joe Says:

    Ben, if you following this thread, feel free to email me off-line. i n t v n u t AT g m a i l DOT c o m.

    As for the CPU-that-executes-GPL detail, I actually got that from the lead designer on the TMS9995. He’s also the one that intimated to me about the delay loops. It could be that he misremembered that part.

    I am curious to see the GPL source for TI BASIC and anything else you might have. The Home Computer was the first computer I owned, and so it is special to me in some way. I have some interesting documentation from that era also, more focused on the VDP and TMS9995.

  135. Gregg E. Says:

    The PCjr used the same audio chip as the TI-99/4A, but didn’t implement it as well so it was more difficult to program than on the TI. With a couple of fairly simple hardware hacks to the video and audio circuits, a PCjr could be made 100% Tandy 1000 compatible while remaining able to run any PCjr program that used its ‘advanced’ video and audio capabilities the way IBM intended. Tandy used TI’s audio chip and its 16 color video modes were graphically identical but accessed differently.

    The PCjr is 100% responsible for the creation of “King’s Quest”. IBM contracted Sierra On-Line to write the game as a “killer app” just for the PCjr. The company went and did icky CGA graphics and PC speaker noises for it too, so the game’s market wouldn’t be restricted to only the PCjr.

    Floppy drives (excluding a few weird SCSI ones) are all ‘serial’ devices. They have one line for writing data and one for reading. The other wires are control and signal ground lines, and in some uses (apple ][ and Macintosh) power. Half of the 34 lines on a PC floppy cable are grounds! The TI-99/4 and 4A used standard Tandon or Shugart drives, the same as the IBM PC. Thus the TI had one of the fastest floppy disk systems among the home/micro computers. Unfortunately TI never produced a double density controller, leaving it to 3rd parties to make them.

    There were several companies that produced “2nd story” upgrades for the PCjr which added a box to the top of the computer then the original lid set on top. Some of them were just a RAM expansion to kick it up to 512K or 640K and a bay for a second floppy. Some included a normal parallel port in the connector sidecar which provided the bridge to the upper chassis. There were several companies which made many different upgrades for the PCjr. PC Enterprises was the biggest, and if they didn’t make it themselves they bought it from another company and sold it. A fully tricked out PCjr could have VGA video, Soundblaster, 33.6 modem, EMS RAM, a hard drive and running Windows 3.0. Add an NEC V20 CPU and not only did DOS programs run a little faster it could natively run CP/M! I used 22-Nice to enable it, then booted CP/M then ran WordStar. IIRC I could use the same CP/M and WordStar disks I used on my Xerox 820-II.

    Here’s a big, fat computer design mistake. The original Amstrad PC. It was somewhat PC compatible, but like the PCjr had some non-standard video modes and its power supply was in its monitor, which connected to the CPU box via a thick cable and a CPC (Circular Plastic Connector) with a threaded collar. Absolutely no way to upgrade the monitor or the built in video. Might have been possible to install an 8bit ISA videocard, but it only had 2 or 3 slots and IIRC they were short slots so the full length cards of the day, built from TTL chips, wouldn’t have fit.

    And another one! The TI Professional. A completely generic PC clone, except for its videocard, which TI called “3 Planes”. It was much better than IBM’s CGA (EGA wasn’t around yet) but there was very little software written to use it. Fortunately swapping the TI card for a standard card made it 100% PC bog-standard, but made it 100% incompatible with any of the expensive software you’d bought that needed TI’s videocard.

  136. Patrick Lajko Says:

    I had forgotten about the mistake in the CGA colors of the PC: cyan, magenta and yellow — not red, blue and green. They used printing colors rather than video colors. So you could never get real color images or graphics. They improved with the alternate color set: red, yellow and blue. Still not right as these are the colors for mixing paint (remember grade school and mixing colors with watercolors?). How could all the IBM brains not know the difference between CMY and RGB?

  137. user Says:

    For a modern one, a graphic card: EVGA Geforce 8600 GTS. It adjusted the fan speed in accordance with the temperature, but for some reason reason it would slow down when it would heat up, causing it to run at 85 C and feel blistering hot to the touch. Using software to control the fan would fix this, or installing an aftermarket cooler.

    Blue & White Power Mac G3, early G4 – Had a cooling system that made no sense. Big fan blows in through the side of the case hitting the expansion cards, CPU got indirect air because it really didn’t get that hot. Only problem was the hot air had no where to go (the power supply fan wasn’t enough), so this bad design caused turbulence, resulting in had been an exhaust fan on top of the unit.

    What else.. I have a Compaq Deskpro that won’t boot unless it has a Compaq CDROM drive installed, but other than that it’s a reliable solid computer.

  138. samuel Says:

    dVmvK5 odfBxZakGhH3v96M2Qsk

  139. Enrique Says:

    For those who want to know why the 1541 drive was sooo slow with the C-64, here’s an old post from the guru Jim Brain.

    http://www.geocities.com/rmelick/6522f.txt

    > We early PET/CBM freaks knew, from playing music, that there was
    > something wrong with the 6522’s shift register…

    So it was a bug in the VIA 6522 shift register which caused the data transfer to be handled in software rather than hardware, slowing down all the system.

  140. Joe Cassara Says:

    #16: Commodore PET 2001 chicklet keyboard: Even a genius like Chuck Peddle is allowed to goof-up now and then.

  141. Barc Says:

    When I started here at my current employer 20+ years ago, they had just got started with personal computers. Just not PCs. Since they were already using a Burroughs mainframe, they used the Burroughs B26, then B28, machines, mentioned by Scott Marlowe above. Burroughs supplied (though I don’t know if they charged for) mainframe emulator software. This architecture used what the article calls “sidecar expansion”. Burroughs called the extensions “slices”. A brother-in-law in the CG Reserve reported seeing a room full of metal shelving at his base whereon each machine was augmented with slices to the limit of the shelf length (file servers? Never found out). (We never had any that large here.) They ran on CTOS (Convergent Technology OS) until Burroughs acquired Convergent. After that, they just changed the name to BTOS.

    The text-based system, lacking a mouse, had an interesting feature in the word processor. (To be fair, I never really used any other text-based word processor; by the time we moved to PCs, they were GUI. I imagine other WPs worked similarly.) To copy a block of text, you moved the cursor with the arrow keys to the beginning of the text, pressed a key (I think it was a soft key) to “mark” the starting point, then moved the cursor to the end and pressed another key to “bound” your selection. Then you could copy the highlighted selection.

  142. carfan Says:

    Steve Job’s insistence on fanless computers sounds similar to Soichiro Honda’s insistence on air-cooled engines in the 60s. The Honda engineers had to build an air-cooled formula-one racing engine! Without water, cooling was never enough, so they bored a hole in the clunk case to let air in. But, then the engine spilled oil from the hole, so they installed a centrifugal separator at the whole to separate oil from air. A formula-one car with such an engine actually ran in a race (or the qualifying session of it)!

  143. David V. Says:

    The TI99/4A was also my first machine. I poured all my preteen savings in it (I was 11) and couldn’t even afford the price of the cassette player cable after that (which was actually a cable that could read one player and read/write another). So I went with my dad’s multimeter to measure the pin-in/pin-out configuration on a demo unit and built a clone cable from scrap parts. I think the computer side of the cable was a DB-9 connector, but I only had a D-25 or so: I ended up hacksawing the latter 😛 Quite the week-end project for a kid.

    I quickly ran into the limitations of TI Basic and learned to program with those limitations. After over a year, I finally had saved up enough for the “Mini Memory” cartridge. It added 4KB of CPU-addressable memory, a patch to TI Basic that allowed to call into routines placed in that memory, plus “CALL POKE” and “CALL PEEK” extensions. It also included a line-by-line assembler, but it took more than 3 of the 4 KBs of memory. I ended up writing an assembler in TI Basic to be able to use most of the 4KBs available.

    The TMS9900-based chip in that machine was “interesting”. Unlike most of the home computer chips at the time (6502/6510, Z-80, and even 6509) it was a full 16-bit CPU. However, most of its “registers” were “software”. That too contributed to slowness. If I remember right the registers were R0 .. R15, and R15 was called the “workspace pointer”, which pointer to a sort of call frame against which the other registers were indexed. I remember the BLWP instruction (“branch & link workspace pointer”).

    The “video processor” was pretty good for that time (TMS 9928, I think). I was later selected as the part for the Japanese “MSX” standard of home computers. As mentioned, it supported “sprites” (up to 15 or 16 16x16pixel blocks that could be overlayed anywhere on the 192×256 bit map). It was also mentioned that the 16KB RAM was all attached to the video RAM, and that’s correct. However, TI Basic didn’t allow one to switch the video to a bit-map-like mode, leaving about 14KB for storage of basic programs. With the “Mini memory” cartridge, I did eventually find out how to switch modes, and the “high res” mode (192×256 pixels) allowed you to choose 2 colors for each 8 pixels (in an odd layout that was clearly geared toward rendering 8×8 pixel characters). I kind of remember I had two “pages” of this that I could use, but I might misremember.

    So many good memories…

  144. Bruce A. Brown Says:

    I worked for IBM in the 1980’s and since the PCjr was a failure they were practically giving them away to employees. I remember the price being around $500, which included some software and add-ons. They had redesigned the IR wireless keyboard so it had better keys with the letters on them, kind of like a current laptop keyboard (still pretty chintzy though). I had two of these for some reason.

    Anyway, I purchased some non–IBM add-ons and performed a hack that uncrippled the internal floppy controller so it could run 2 floppies. This hack (available on IBM forums an IBM internal only Newsgroup kind of thing) involved a coworker making wire-wrapping changes to the floppy controller and me installing software changes to the DOS Format command using the Debug utility.

    These add-ons and modifications gave me a 640K, 2 floppy, 104 key wired keyboard system for about $750. Sure it wasn’t very elegant (a couple sidecars for instance) but it worked fine especially for the money. Around the same time – 1983 or 4, I used my IBM employee discount to save my dad some money on the purchase of a real IBM PC. This 256K, 2 floppy system still cost over $2000 and that was with a 50% employee discount. PCs cost a LOT MORE back then, adjusting for inflation you could at least double the dollar amounts to get current equivalents. So taking in to account inflation and my discount today’s equivalent would be $8000, with that you could buy a super high-end giant monitor and hard drives (including a SSD), dual graphic card, ultra fast gaming system.

    My son and I spent many hours playing the original King’s Quest and I ran Managing Your Money (forerunner to Quicken). Yea, I know my situation was special being an IBM employee and having the knowledge and friends to be able to do the hack, but I have fond memories of the PCjr despite its shortcomings.

  145. Mystgreen Says:

    Concerning the DEC Rainbow. Other problems it had with it’s design was like 3, stair-stepped rows of male ribbon cable connectors (two parallel rows up upright, gold-plated pins) combined with cables that were so close to being too short that they literally had to be stretched a fraction of an inch to make the connections (a cost cutting effort I am sure) and placed next to other obstructing devices that had to installed first (There were no card-slots [another design mistake?] so they used piggy-back boards atop the motherboard for other options) so that my fingers slipped off the connector while pressing them down and impaled themselves on the adjacent rows of pins! I only worked on one of these machines. The rest of the hardware crew ran into the exact same issues had most of our hardware staff running around with band-aids on thier fingers for the next few days. They referred to the DEC Rainbows as the Iron Maidens of computerdom.

  146. Stuart N. Says:

    I am surprised that there was not one mention of the miserable IBM 5155 PC portable computer.

    I nearly went blind having to squint at this tiny monochrome display for 2 years. Portable? They must have weighed at least 25 pounds.

  147. Yay Says:

    Looks like Commodore did everything right with their C64/128 and Amigas 😀

  148. Charlie V Says:

    The Coleco Adam’s keyboard was indeed excellent – i had one for my Apple II.

    You read that right: someone made a kit that would let you connect the Adam’s keyboard cable to the keyboard socket on the Apple II’s logic board.

    I ordered it from an ad in one of the Apple II magazines (probably InCider). It came as two circuit boards (a tiny triangular one, and one about 3×5 inches), a bunch of loose components, and some photocopied diagrams.

    You had to solder all the components to both boards. Then you had to open the Adam keyboard’s case, snip a couple of wires, solder in some jumpers to the tiny circuit board you’d just made, stuff it all in, then close the case back up.

    Inside the Apple II, it was simpler: unplug the cable from the built-in keyboard, plug in the cable coming off the other circuit board you’d made. The coiled cable from the Adam’s keyboard was long enough to reach around the back of the Apple when the keyboard was sitting in my lap.

    Included were a couple of stickers (hand drawn with a Sharpie!) of the Open-Apple and Closed-Apple keys, meant to be stuck on the keys to the right and left of the space bar.

    It worked perfectly, and typing feel was very good. It was also quieter than the relatively ‘clacky’-sounding keys of the Apple IIe.

    I used it with both my grandfather’s II+, and then a late-model IIe I picked up at Goodwill. I also had four floppy drives, to run AppleWorks without having to switch disks. Typed a lot of college papers with that keyboard.

  149. Harley Says:

    What about the NABU computer – the oddest computer I’ve ever seen. Okay, I know it doesn’t belong in this list of epic well-marketed flawed PCs. The Nabu PC was a 1982 CP/M machine with the housing designed by a furniture maker. It measured about 2 x 2 x 2.5 feet high, was constructed out of melamine particle board and was very heavy. It rolled on casters, had a large sliding drawer on the bottom and you could use it as a small table. You could put the monitor on top, but it was a bit too high and no good place to put the keyboard. Heat build up was a serious problem. Although it did have a fan blowing in, there was no air flow because the only place the air could get out was through the 8″ floppy drives. The fan created more heat than it dissipated.

    I worked as a software developer for NABU in Ottawa in an open space 5th floor office with floor to ceiling windows. The company didn’t have $$$ for curtains or blinds and the room hit 80 degrees around 11 AM. I got about 2 hours of work done before I had to shut down the computer for a while to cool it off. Then it was an hour at a time.

    In 1982, NABU was the first company to deliver software and content through the cable network. A modem link was developed for interactivity with the network, because the older cable system repeaters didn’t support customer signals getting to the server.

  150. Justin Says:

    The ColecoVision Adam, my very first computer. I still love it.

  151. Robert Says:

    Ive got a classic Model M Ibm Clicky keyboard a true blast from the past ITs for Sale
    check it out on ebay here is the link to the auction i have many available
    http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=200456293270

  152. Abigail Clark Says:

    i have tried selling different products from Multi Level Marketing and i can earn a decent amount of money from them.~-:

  153. Michael Rudas Says:

    When IBM sold the PCjr to classrooms, the IR keyboard was a liability because every keyboard used the same interface code stream and you didn’t always know which PC you were controlling with a given keyboard. One little-known fact about the PCjr is that it was produced for IBM under contract from Teledyne–when sales tanked, it didn’t matter to Teledyne; it was a fixed-end contract and they produced EVERY ONE of the 250,000 units they were contracted to produce, the majority of which were ground up and sent to landfills.

  154. Carl Says:

    The Coleco Adam was a great computer at the time. In contrast to the Commodore 64, it had 80 KB RAM, included a tape drive and printer, a separate keyboard like “real” computers, and lots of arcade-style games.

    Unfortunately, the power supply died on my printer and I brought it to a Coleco service center. I never saw my Adam again. They said it would take 6 months to fix, by which time the service center disappeared and I had a new computer.

  155. Fairportfan Says:

    Not mentioned in the section on the PCjr is that if you bought a dedicated monitor, and set it up the way that IBM promoted it (with the monitor sitting on top of the system unit), there was a very good chance that the electromagnetic noise from the monitor would interfere with the machine’s operation.

    This was not a problem with the PCjr alone though – i first read about it a week after i exchanged my first Apple IIc for unreliable disk writes (the IIc was designed to sit on the desk with its monitor directly above it on a u-shaped stand). Had i known about that sort of problem, i might have had an easier time convincing the shop i had bought it at that there was a problem, as when we set it up plugged into a monitor that *wasn’t* sitting directly above it to check it out, it worked perfectly…)

  156. flv to wmv Says:

    how about the Design from Apple

  157. security guard tour Says:

    I like to think my security officers can work with guard systems. <a href="http://www.stotsindia.org/computers-and-technology/successful-security-guard-tour-systems-needed-four-our-large-company
    " target="_blank">security guard tour systems is what its all about. you need to closley monitor security with this kind of software.

  158. dreambox dm600 pvr Says:

    If you come across, even though the configuration is easy in this movement to connect and disconnect, crashing whatever you are doing the machine. If the voice synthesizer can go to that will be the last point is more stable.

  159. Kennesawga Garage Door installation Kennesawga Garage Door Opener Repair Says:

    Found your web sites on AskJeeves, great articles, but the site looks awkward into my browser setup, but performs fine in IE. get figure.

  160. Nuseweisump Says:

    Studies have shown that some pain killers can result in gastrointestinal problems. Be careful! That’s what i want to say here.

  161. Jae Schueler Says:

    I would say additionally writing a small “bait” application may possibly improve.

  162. Michael Rudas Says:

    I know I’m late to this party, but I have a few comments that might shed further light on the contents of this article and some of the earlier comments.

    * The IBM PCjr (I’m told) was made under a fixed-end contract by Teledyne—they were contracted to make 250,000 units; they manufactured every one of them. IBM sold only a fraction of those units, so the rest were destroyed and placed in a landfill.
    * TI only got into the home computer business because the TI-99/4 was originally designed as a game console for Milton Bradley; the game company backed out at the last minute, so TI sold it themselves. The original keyboard was originally intended to be split into two separate controllers and was mapped that way.
    * The Macintosh was never designed to be a stand-alone computer—it was supposed to be a terminal on a network, with the LISA as the server. When the LISA crashed and burned, Apple was forced to sell the Mac as a PC. That’s why the first-generation Macs were so woefully inadequate.
    * The slow disc drive interface on the Commodore 64 was caused by a rush job on the C64’s design. The newly-designed chip used to interface the serial IEEE-488 port had a bug, so the software guys had to work around the bug to get it to work, thus slowing the port down. The hardware bug was fixed a few days later, but it was too late to alter the “kernal” routines because of the rush to finish the design before the CES introduction.
    * The first-generation Commodore 1540 and 1541 disc drives were made by ALPS. They had the same problem that earlier ALPS 8-track players had had—the belt-driven flywheels were attached to the drive post by means of a friction fit and tended to fall off.

22 Trackbacks For This Post

  1. 15 cagadas de diseño de viejos ordenadores Says:

    […] 15 cagadas de diseño de viejos ordenadorestechnologizer.com/2009/06/14/fifteen-classic-pc-design-mista… por dThrak hace pocos segundos […]

  2. transparantezaken's status on Monday, 15-Jun-09 17:29:09 UTC - Identi.ca Says:

    […] Vijftien klassieke fouten bij het ontwerpen van computers: Apple, IBM, NEC, gebrek aan open standaarden http://ur1.ca/5pz3 […]

  3. Fifteen Classic PC Design Mistakes | Technologizer « Netcrema - creme de la social news via digg + delicious + stumpleupon + reddit Says:

    […] Fifteen Classic PC Design Mistakes | Technologizertechnologizer.com […]

  4. 15 errores clásicos en el diseño de ordenadores [ENG] Says:

    […] 15 errores clásicos en el diseño de ordenadores [ENG]technologizer.com/2009/06/14/fifteen-classic-pc-design-mista… por phal hace pocos segundos […]

  5. Fifteen Classic PC Design Mistakes | Technologizer | Spam by mulaz! Says:

    […] via Fifteen Classic PC Design Mistakes | Technologizer. […]

  6. Sun Country’s Technology Weblog » Post Topic » Fifteen Classic PC Design Mistakes Says:

    […] ever released.  But when considering past design mistakes, these examples spring to my mind. https://www.technologizer.com/2009/06/14/fifteen-classic-pc-design-mistakes/ Published in […]

  7. You sure don’t see a lot of sidecars nowadays. | dv8-designs Says:

    […] 15 Classic PC Design Mistakes, along with explanations as to what exactly they were thinking at the time. […]

  8. Fifteen Classic PC Design Mistakes « Tech Flaws Says:

    […] PC Design Mistakes I’ve just read these interesting examples of flawed tech over at Technologizer.com who ask the very same question I keep asking myself whenever I come across unexpected and pointless […]

  9. Tommy k Johanssons blogg om datorer & Internet Says:

    15 klassiska misstag inom datordesign…

    Genom åren har det lanserats en väldig massa olika datorer, och alla har inte varit så bra… På Technologizer har man sammanställt en lista över de 15 klassiska misstag inom design av datorer. Det är en välmatad artikel där man går igenom olika da…

  10. Josh’s Blog » Blog Archive » Fifteen Classic PC Design Mistakes Says:

    […] Imagine that: Steve Jobs insisted that Apple III remain silent and not have a cooling fan. – Link […]

  11. Tent City Springs Up In Bushwick Backyard - City Room Blog - NYTimes.com Says:

    […] with your computer? Take heart — at least you are not stuck with one of the 15 worst PC designs of all time. […]

  12. Top Posts « WordPress.com Says:

    […] Fifteen Classic PC Design Mistakes There’s no such thing as the perfect computer, and never has been. But in the personal computer’s long and […] […]

  13. Fifteen Classic PC Design Mistakes | Technologizer « Bikkiz Says:

    […] Fifteen Classic PC Design Mistakes | Technologizer Easy AdSenser by UnrealFifteen Classic PC Design Mistakes | Technologizer. […]

  14. 15 errores de diseño en ordenadores personales Says:

    […] perfección tampoco existe hoy en día, pero si comparamos nuestros equipos con alguno de estos recopilados por Technologizer nos damos cuenta de lo que hemos evolucionado. A continuación una selección personal de esos 15 […]

  15. Chad’s News - 15 Early Computer Mistakes Says:

    […] https://www.technologizer.com/…(via […]

  16. 15 micro-ordinateurs anciens au design défectueux ! - Préservation, Sauvegarde, Rétrogaming, Exposition... Says:

    […] https://www.technologizer.com/2009/06/14/fifteen-classic-pc-design-mistakes/ Actualités […]

  17. dholm.com » Blog Archive » Tumblelog 090706 Says:

    […] Fifteen Classic PC Design Mistakes, Technologizer has made a great list of the ten worst PC design mistakes. […]

  18. Algunos Errores de diseño en ordenadores clásicos | CyberHades Says:

    […] Technologizer […]

  19. Fallos en el diseño de computadoras | IdentidadGeek Says:

    […] Tecnologizer, 15 classic PC Design Mistakes AKPC_IDS += […]

  20. Thomas Boelskifte's Mac & Tech-Blog Says:

    […] Læs dem selv her 🙂 Ingen kommentarer | Trackback […]

  21. The Top 15 Technologizer Stories of 2009 | Technologizer Says:

    […] 15 Classic PC Design Mistakes: Benj critiques vintage PCs, from the era’s masterpieces (the Apple II) to its forgotten […]

  22. Elenin Says:

    Nice post….

    […] Quite nicely explained, despite this document is far from brand spanking new. We are in agreement with the majority of your main points, the only thing we would like to provide is to look at a couple more pictures. Your post was appreciated. [……