That isn't exactly true, especially with the older platter coatings (especially of the barn-paint variety). The magnetic particles (dipoles) in the coating differed in their coercivity (resistance to flipping), and a weak write or erase might not flip them all, leaving a ghost that can be picked up with the right equipment.
This variation also causes "print-through": ..
Modern metallic disk coatings are far less susceptible.
The reason was mechanical tolerances in the drive.
The write heads leave a wider stripe than the read heads, so minor mechanical wobble or wander won't cause read errors. Erase is done with a write head, and there is most often a thin stripe of data left over from the prior write pass.
One can read such leftover data with the right equipment. Some kinds of drives can be commanded to do an offset read to recover the stripe; no special equipment needed. Offset writes are also available, and can be used for sanitizing.
Overwriting multiple times causes those edge stripes to be mangled.
More generally, I sanitize hard drives with a 2-pound drilling hammer and a big granite boulder used as an anvil.
Maybe an intelligence agency can recover some data from the mangled result, but that isn't the threat I would worry about. Not that any ordinary individual could keep an intelligence agency out anyway.
Joe Gwinn fired this volley in news:260120141836249044% snipped-for-privacy@comcast.net:
Where, exactly, do you find liquid-coated iron-oxide slurry disks _actually_in_service_ holding information that would be of interest to a forensic study?
In the mid 1970s I worked in a shop that had for a couple of its jobs machining castings for IBM disc drives. I would run for 3 months straight 10 hours a day cast iron motor housings. That equalled 280 per week because the cycle time, including loading, was just under 15 minutes, so I could run 40 a day. There would be a break for a few months and then it would be back to seven day weeks. In between the cast iron parts I ran die cast aluminum parts for the same drives. They were much faster to run and didn't require 7 day weeks. Only 50 hour weeks. The aluminum castings had holes die cast into them that we tapped 10-32 and 8-32 with form taps. I tapped over 10,000 holes without breaking any taps and it was the first time I tapped anything. The reason for the low tap breakage was the accurate hole size, good tapping fluid, and a good setup. None of which I can take credit for. I didn't see the drives they went into until sometime after 2000 when I saw one at RE-PC, a Seattle computer recycling store. They have a little museum in back. I recognized the castings in the drive. I don't remember the capacity of the drives but I do remember being shocked at the time that a disc about 2 feet in diameter held less data than even old hard drives. Eric
Actually -- IIRC, the name Winchester came from a double hard disk, each containing 30 MB, so it was called a 30-30 -- thus the "Winchester" nickname. :-) And that had fixed platters, not replaceable ones like most of the washing-machine drives. The washing machine drives were typically 14" platters. I *think* (but am not sure) that the 30-30 had platters around 8" or 10" diameter. I was once at a "strippers" breaking the drives down for various materials. I could not get close enough to actually measure the platters -- too many swinging sledgehammers. :-)
Well, the crashed platter on the wall is 14" in diameter, and came out of a diskpack containing at least four platters.
The drive was built by Control Data Corp (CDC), had a glass top, and lots of failure mechanisms.
My favorite: We had just gotten the drive, and my job was to write an I/O driver for it, for use on a SEL 32/85 midicomputer. I had it pretty much working, but there were these scattered read and write errors, and I was pulling my hair out trying to find the bug in my brand new I/O driver software. This went on for at least a week.
One day, for no rational reason that I can recall, I got a General Radio stroboscope from the test lab, and used it to inspect the spinning disk pack through the glass top. Every so often, the stopped image would jump about 5 degrees. Hmm. What the *hell* could cause that? That pack is spinning at 2400 rpm and has a lot of rotational inertia. It would take a factor more torque than the motor can provide to cause such a jump. The only thing that is even plausible is a problem in the main spindle bearing, and that's what it turned out to be.
Turns out that CDC had gotten a bad batch of bearings. The drive was new, so I doubt that the raceways were spalling from fatigue, so there had to be hard particles in the grease from the start. Bet CDC changed bearing suppliers after that.
A Field Tech from CDC arrived and replaced the main spindle bearing assembly, but was unable to align the carriage so it would move freely without scraping the insides of the magnet, using the factory instructions. A coworker and I, who both worked on cars, stepped in with the traditional auto repair approach using shims (a sheet of paper wrapped around the voice coil and removed after everything is tightened). This worked, and was faster than the factory procedure.
The first small hard drive I saw, other than a 14" one under a dome in
1971, was a 5MB Fujitsu that appeared on my bench with a note to see what I could do with it.
It had no controller, only step and direction inputs and a direct connection to the read/write heads. I found out how to set up an interleaved sector addressing format with a WD1771 floppy controller chip and was able to position the head and read it with an oscilloscope, but designing a fast enough write driver was beyond my experience. Anyway, better drives came out soon and we didn't need to use it.
The drive I saw in 1971 was part of an experimental field-mobile communications relay center built into six linked semi trailers with fold-out sides. The idea was good but it needed an impractically flat and level site and more Burroughs and Army techs to keep it running than TTY operators using it.
20 years later we could pack a more capable version into a satellite.
"Jim Wilkins" fired this volley in news:lc5vae$vch $ snipped-for-privacy@dont-email.me:
Even in the day of the first Fuji and Shugart 'bare' drives (my first was
5MB, also), there were bolt-on controller boards from OMTI and Shugart Associates to drive them. Huge and power-hungry, they were.
They were costly, too, and honored the non-arbitrating SASI protocol, but they worked quite reliably. One of those was the basis for our first Lt.Kernal hard drive we engineered for Commodore computers. It wasn't long before embedded-controllers took over in small drives, and SCSI arbitrated busses became the standard.
The Write pulse was quite a zap, comparable to an H bridge motor drive and with very fast rise and fall times.
That company built automated semiconductor test equipment which had the specialized requirement to deliver very long test patterns to RAM chips being tested, continuously and as fast as possible. There could be three test vectors in the coax cable from the rack to the test head and back. They hoped the hard drive would reduce their need for expensive static RAM. I remember figuring out how to make the WD1771 format a track as one big sector to use as a buffer. jsw
Umm, Don? The very first hard drives were measured in KB, not MB. By the time they were shrunk down and reliable enough to put them in the first PC's it was a 10-KB drive.
Want to see one? I kept the first one when it failed, and 40-KB was the smallest I could find. And we could only address 30-KB of it, no partitioning back then.
To get into the MB ranges that Mainframes needed they were using "washing machine" style drives with a whole stack of platters on a common axle. And about the same size, too.
For "Removable Storage" you parked the heads and popped the lid on the Washing Machine and the whole platter stack unlatched and pulled out, and you put a huge cake cover over to protect it for shipping and storage.
That TI test station from the 1980's used a card reader from their warehouse to save money. It was built into an old steel office desk, with the wire-wrapped processor in a drawer. jsw
The first programming job I ever had was to do statistical analyses on a Wang Programmable Calculator. It used punch cards, but we had no card punch, so they bought 'pre-perforated' cards, and I punched them out with a stylus on a rubber mat (standard equipment with the sandwich-style readers of the Wang).
Depends on what you are calling a "PC". The IBM PC originally came with an interface to audio cassette tapes, and optional single-sided single-density 5-1/4"j floppy drives. (Which were on the order of 70-90 KB, depending on format.) At the time, the older 8" floppy drives were 250 KB single density single sided, and grew to 1 MB Double-sided Double Density. (I used these in a SWTP 6800, and later
6809 system, along with some 5 MB hard drives and later some 25 MB hard drives. And IBM used the 8" floppy drives (single sided single density) in a keypunch replacement machine. One CRT looking up through the table, with two 45 degree mirrors, so someone sitting on each side of the table could read their part of the CRT. The only other things there were a keyboard and an 8" floppy drive on each side. While the floppy was formatted using 128 byte soft sectors, it actually *used* only the first 80 bytes -- because it was pretending to be punched cards, after all. :-) IIRC, it was 26 sectors per track, and 77 tracks per disk. That was 160K of data stored on a floppy which could handle 256K when used by anyone else.
The IBM PC *XT* came with a 5 MB hard drive, also 5-1/4" format.
Now -- if you want to go back far enough -- but then you are
*not* talking about an IBM PC, nor any of the later clones thereof -- you can get hard drives with fixed heads instead of moving heads. You had one head per track, and accessed another track by switching heads. This seriously limited the number of tracks you could get. There were some with platters on the order of 36" or so (I never got a chance to measure one, just see one on the wall of a computer hobby shop in the early days). I've seen the drives -- standing on edge, with horizontal shafts. And some drives used hydraulics to move the heads.
There was also another kind of floppy disk. I've seen them offered at hamfests, but never bothered acquiring one. Vertical spindle, sealed housing (which looked sort of like a mushroom), with the permanent floppy disk (no jacket) drooping inside -- until it spun up and came into contact with the heads above the floppy. I don't know what the density/capacity of that was, but not very spectacular. I think that it was fixed heads, too.
And what system was this installed in? I don't think that it was an IBM PC -- even the floppy drives used in those were larger than that.
How big (physically) was the drive?
Now -- IBM made some desktop computers prior to the PC -- your choice of BASIC or APL (I think) -- or one model which had both, and you could switch between the two with a front panel switch. I think that it was the 5100. O.K. Here it is:
No hard disk with that -- just a cartridge tape drive. I was tempted by that one -- but there was no way that I could afford one. :-)
The drives which I was talking about above were hard drives, multiple platters on a spindle, two spindles in a single housing.
The drives I was talking about were not removable -- all the platters were sealed in the housing. And they had higher density than the "cake cover" platter assemblies used in the washing machine drives. (BTW -- people have been known to competitively "race" those washing machines. Move the heads slowly (a few tracks at a time to one extreme, then a rapid seek to the other extreme, and it was enough mass accelerated hard enough so the drive would move a little on the floor. Two people, writing programs on one mainframe to each control his own drive (operators, late at night) could cause the drives to slowly move across the raised flooring. Yes, those washing machine drives were typically 14" platters. I've still got one somewhere around. And there were smaller drives using one fixed 14" platter and one removable 14" platter in some of the minicomputers -- such as those by DEC and Data General.
But a 40 KB hard drive -- less storage than the early removable floppies, and less than the address space of an 8080 or a 6800 as used in early home computers from before the IBM PC. I can't see the need.
Now -- go far enough back and you get to *drum* memories -- again head-per-track. A length of sewer pipe turned to a high surface finish and coated with oxide, and spinning at speeds which made them rather impressive gyroscopes. Now *those* I could see as being as little as 40K.
PolyTech Forum website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.