38

In the mid-1990s while a student at a US university during a computer science lecture, my professor (not a TA or grad student) told us a story of "witnessing" a large, then old-fashioned metal hard disk platter somehow breaking free of its mounting while spinning and thus flying through the air to embed itself violently halfway through one of the walls of the room.

The unfortunate hard drive in question was a cabinet-sized device. But not a tape machine.

At the time, I don't recall any questioning of the veracity of this story, but in hindsight it sounds pretty fishy. IIRC the point of the story was something about information density and how then-modern (1990s) hard drives were so much smaller & more capable. So it might have just been a convenient way of making his point. Or at least an embellishment. I'll allow that it could have been passed down and not meant as an outright fabrication but was still presented as a real event.

Doing some web searches turn up nothing. On the other hand... can't prove a negative.

Are there any convincing or reliable accounts of an incident like this? I guess the alleged event would have taken place between ± 1975-1990 based on what I think my prof's rough age was at the time.

  • 3
    Something like this? youtu.be/zs7x1Hu29Wc – snips-n-snails Feb 16 at 18:39
  • 3
    Are you sure it was a disk and not a drum? I have heard of magnetic drums coming off their bearings and flying across the room, but I have never seen it happen. – Mick Feb 16 at 18:54
  • 4
    Since those disks came out in 1965 and I heard the same story in 1975/76, I'm guessing it was between 1965 and 1974 if it actually happened. – cup Feb 17 at 6:09
  • 3
    heh this reminds me at a small accident at my prior work place where a SW bug corrupted breaking sequence of a linear motor that has been tested which head then fly through brick wall and slice another one in next room ... and that was low weight (~0.5kg) and no spinning only about 2 m of acceleration and blunt shape (something like a cube not a slicer disc saw). I saw a lot of unbelievable stuff over the years (both electrical and mechanical) so the story is from my point of view plausible not at all far fetched – Spektre Feb 17 at 9:53
  • 8
    At my first job, I was developing PC software, but the office ran mainly on a somewhat dated ICL mainframe. It had cabinet-sized hard disks. The office manager had on his wall a picture frame holding a shattered hard disk platter, which (as he was very fond of telling everyone) had "escaped" from its drive and smashed itself against some other hardware in the computer room. I never saw any evidence of the incident other than the shattered disk, but I never had any reason to doubt the story either. – Spudley Feb 17 at 10:30
23

There is also this tale that I read recounted in Philip and Alex's Guide to Web Publishing, although the disk pack was intentionally hurled through the wall in a fit of temper:

Excerpt from Philip and Alex's Guide to Web Publishing article

That night, the Foonly crashed. A Foonly was a clone of the PDP-10, a mainframe computer designed by Digital in the 1960s. MIT and Stanford people loved the PDP-10 but couldn't afford DEC's million dollar price tags. So there were these guys in a basement in California smoking dope and wirewrapping clones that were one-third the speed and one-twentieth the cost. Nobody every figured out why they called the machines Foonlies.

Moon was a superb hardware engineer and nobody doubted that he would get the Foonly up and running. Still, people were a bit surprised when a huge steel cylinder came crashing through the machine room wall. The cause of the crash had been one of those washing machine-sized Control Data T-300 disk packs. The cylindrical missile had been the spindle holding together the bad 12-inch platters. Moon had hurled it through the wall after determining its guilt in the crime of the Foonly crash. I went back to my office and taped up a poster.

This story illustrates that great programmers are not necessarily patient. One of the things that drove them crazy about the object systems of the 1970s (Smalltalk, Lisp Machine Flavors) was that if you changed a class definition, the existing instances

  • 3
    Wow, that's the first actual reference to a similar-sounding event! Its plausible to believe that this original story got modified in retelling to become a hardware accident vs. a frustrated programmer. – UuDdLrLrSs Feb 17 at 17:56
  • Sounds plausible. – Raffzahn Feb 17 at 20:51
  • 31
    Oh dammit, what did existing instances do, if you changed the class definition??! – Gnudiff Feb 17 at 21:10
  • 1
    My hardcopy of the book is long buried under other volumes. I imagine they had to be reloaded or reinstantiated somehow from the new definition. – Samper Williams III Feb 17 at 21:14
  • 20
    @Gnudiff "...if you changed a class definition, the existing instances of that class did not get modified. You'd have to restart your program, maybe even reboot your computer, if you changed your mind about how to represent something. You could lose 20 minutes or even more." ref.: philip.greenspun.com/wtr/dead-trees/53011.htm – OnoSendai Feb 17 at 23:52
62

Some "informed speculation" based on my "day job" which involves worrying about containment of rotating objects if they break (specifically, rotating parts inside jet engines)

The rotating parts of early disk drives were substantial objects. For example the IBM 350 had 24 inch diameter platters rotating at 1500 RPM, with a rotating mass of about 10 pounds. That means the outer edge of the disk is traveling at about 100 mph.

If the spindle of such a disk broke, it's possible that the complete platter would "drift "sideways, still spinning, until it hit the inside of the drive casing. At that point it would attempt to orbit around the inside of the drive, until something gave way - more likely the plastic lid of the drive than the disks themselves.

So it's conceivable that a 10-pound rapidly spinning object could have launched itself across the machine room at say 50 mph. I guess that could make quite a dent in flimsy office drywall.

Whether this ever actually happened is another question, of course.

As an example of how much damage this sort of thing could cause, there is a nice piece of evidence preserved in one of our works test facilities. On one test failure, a similar sized piece of metal came loose, though it was spinning at nearer 10000 RPM than 1500 and the axis was horizontal not vertical. It smashed through a quarter-inch steel plate that was supposed to contain it, hit the brick wall of the building, and because of its rotation climbed up the wall, ran across the ceiling, down the other wall, and across the floor.

It made three complete orbits of the room before gravity finally took control, it lost contact with the ceiling, and came to a stop. The marks on the walls and ceiling are still there after about 40 years, as a reminder of what can go wrong!

  • 1
    Mind to add what material the disk was? And Mass? As well as thickness/size? As all of this does have quite an influence here. – Raffzahn Feb 17 at 2:39
  • 2
    The disks were made of aluminium. Possibly an IBM 2310 or IBM 2315. They came out in 1965. – cup Feb 17 at 5:40
  • 9
    Saw a lightweight flywheel make its way through the top of a transmission, all the components above it, through the steel hood, and still have enough energy go as as high as the scoreboard at the end of the track (1/4 miler)... Driver says it departed the vehicle at around 10,000 RPMs. The amount of energy that spinning objects have can be surprising. – Brian Knoblauch Feb 17 at 17:05
  • 3
    Until we moved, we had a room with similar scars. Ours were also caused by HDDs, but in our case, we used a catapult used for high g electronics testing to squash a stack of them flat, with shrapnel coming out the sides in a kind of planar explosion. – Baldrickk Feb 18 at 14:21
  • 5
    <physics> If a solid disc of uniform density is rotating such that its outer edge is travelling at speed v, then it has the same amount of kinetic energy as if it were moving (without rotating) at speed v /√2 (or about 70% of v). So the IBM 350 platters probably wouldn't travel any faster than 70 mph, but there's plenty of room below that to do some damage.</physics> – Michael Seifert Feb 18 at 21:22
15

I've heard of disk heads being propelled at great velocity, though that may be just a scary story told to young people, too.

This is the disc store from a 1965 KDF9:

KDF9 disc, from 1965

If a read/write head contacts the spinning disk, it could (I suppose) get ripped off the actuator arm, and thus be accelerated to the extent that it penetrates the glass on the door and causes mayhem in the machine room.

Physically-large removable disks would not have this problem - the disk was lowered into a well in the drive, in the manner of top-loading washing machines. Thus there's quite a lot of plastic and metal between the disk and the unlucky bystander.

This Bryant device looks much more likely to cause mayhem if the disks ever got loose, due to the vertical mounting of the huge platters.

Bryant large disk drive

It occurs to me that when disk drives were first being developed, it's quite likely that the mechanisms were exposed (since engineers would presumably be tinkering and adjusting the mechanical aspects of the device) and then your chances of witnessing unconstrained projectile ejection are somewhat greater. So maybe this is a tale from pre-commercialization of disks.

  • 1
    I imagine you also have a higher chance of flinging disks around if engineers are tinkering with the mechanics. – user253751 Feb 17 at 11:20
  • 7
    A gouged disk platter was a popular wall decoration in any self-respecting software development office back in the 1970s :-) but I never bothered to ask the details of what happened to the drive, beyond knowing that surgery was typically needed. – another-dave Feb 17 at 12:08
  • 3
    Goldfinger: "And this is the disc store from a 1965 KDF9." Bond: "Do you expect me to talk?" Goldfinger: "No, Mr Bond, I expect you to die! There's nothing you can talk to me about that I don't already know!" – David Tonhofer Feb 18 at 14:23
  • 7
    The lady in the second photo seems to be more aware of the risks, is standing to the side and wearing safety goggles. – Transistor Feb 18 at 21:42
  • 2
    @JohnDvorak: She's standing out of the disks' plane of rotation, so all she needs to worry about is bits of detached, deflected shrapnel (which are much easier to shield against). – Sean Feb 19 at 23:04
9

Yeah, that sounds quite fishy. Usually I'd say it's impossible. I've seen classic stacks (and drives *1) break in many ways, but never having a platter go astray.

  • The disks are aluminium and the stored energy is rather small (after all, heavy disks mean high energy use).

  • A stack is fixed with about a dozen bolts roughly M4 size. An M4 bolt needs a break force of more than 7 kN (for standard 8.8), so it's pretty hard to break a dozen of them.

  • Next the drive is a massive steel shaft, needing even higher force to let go.

  • So even if they break free, there is no chance that aluminium disk could cut thru either.

  • Further the disk is covered by a steel frame, which any 'escaping' disk would have to penetrate.

  • And finally still have enough energy to fly across a room and penetrate a wall?

Not really likely.

Of course, there are many different constructions of disks, and some may be more prone to 'ejecting' a platter. But never in a way that dramatic. So my assumption it's a story based on misinterpretation of wording paired with retelling over many years and sources.


*1 - The most impressive I ever had to handle as service engineer was a disk motor coming loose. It was a drive sized like a half height refrigerator, or a top loading washing machine - well, the stack was loaded from above anyway :) The heaviest single part was the motor. About 50 kg any too large to fit into a alpha spider tunk :) It was a direct drive synchronous motor - mind you, these were the late 70s, way before today's tiny implementations - storing the most rotation energy within. So that baby got loose but even this huge moving mass wasn't able to even make a dent in the outer sheet metal of the drive.

  • 5
    just a note but bolts and screws need much less to break lose than one would estimate or compute from static case... especially in 24/7 vibration rich environment due to fatigue of material. Also screws tend to unscrew themself with vibrations if joint not designed properly for the task (like wrong screwing direction or upside down joints). I for example saw few failed hard drives that unscrew them selfs from their housing and drop down in the PC case during operation... The same goes for shafts but in such case it would be unlikely that only one plate would eject and not more ... – Spektre Feb 17 at 12:05
  • 7kN is the shear load for an M4 bolt accodring to DIN. – Raffzahn Feb 17 at 20:54
  • 5
    shear load is a static force parameter, dynamic loads are computed differently with very different equations that are dependent on load, cycles and sometimes even frequency and amplitude... and usually Force that can damage material in dynamic load is just a fraction of the static Force leading to damage – Spektre Feb 17 at 23:59
  • 1
    Speaking as someone who's been involved in breaking a great many bolts in a former job, it's a lot easier than the numbers make it look, and once the first one goes, the rest usually follow in short order. – Mark Feb 19 at 3:56
  • 2
    All true, just has anyone of you ever seed a disk work properly when there is a vibration like assumed to make the bolts break? Having worked on this disks for years (an with them for many more), I never seen such - the drive would have been shut down way before that due unreliable reads. This is not about some hobbyists using not tightening the mounting of a HD. – Raffzahn Feb 19 at 8:05
3

Are there any convincing or reliable accounts of an incident like this?

I couldn't resist posting this image taken from "The Commodore Deathbed Vigil" (at the approximate 9:43 mark; 11:00ish in YouTube linked below).

For those interested, I pulled this image from my copy of Cloanto's Amiga Forever DVD, but Dave Haynie also has the full video posted on his YouTube channel here.

The Hedley Davis Memorial Disk Drive

We actually had a similar incident at a company I worked for. Although in our case, it was a telephone embedded in the drywall, thrown by a co-worker after our lead sales manager punched his buttons for a week or two straight. - Those were the days!

  • Do you have a link to "The Commodore Deathbed Vigil" somewhere? – UuDdLrLrSs Feb 19 at 21:06
  • @UuDdLrLrSs, I added the source information to my answer. – Geo... Feb 19 at 21:40

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.