THE DISKETTE EXPERIMENTS, PHASE ONE AND TWO (out of three) By: Paul Panks (lumberjacks76@lycos.com) (Originally posted on the USENET group comp.sys.cbm on October 17, 2004). "The Diskette Experiments: A Hypothetical Longevity of the Floppy Disk" By: Paul Allen Panks (dunric@yahoo.com) See website for details. Written summary below: http://members.tripod.com/~panks/diskfun.html In three separate experiments (conducted at regular intervals), I have determined the hypothetical longevity and general integrity of data (both read and recorded) on the standard 5.25" floppy diskette. Such diskettes are common to the Atari 400/800, Commodore 64/128, Apple II/IIe and other 8-bit computers. 3.5" disks were not tested. Here are the results of my experiements (conducted on 5.25" floppy diskettes for the Commodore 64/128 line of microcomputers): ************************************************************* Experiment One: Freezing diskette (Conducted in January 2003) ************************************************************* Hypothesis: Frozen diskettes lose data, even if frozen overnight. I froze a diskette overnight in a freezer, without the diskette holder protecting the exposed mylar surface. I left the diskette intact inside the disk jacket, though. The disk received condensation on the exposed portion of the mylar surface (of the diskette), and was extremely cold to the touch. I noticed no physical alterations to the diskette, however (e.g. no warping or physical damages; only coldness and condensation). The disk drive booted the floppy disk directory without effort, and the contents read without problem. Programs all loaded without error. Conclusion: Freezing a diskette overnight probably won't damage the diskette enough to ruin diskette data integrity, but probably isn't a good idea, anyway. ************************************************************* Experiment Two: Overheating diskette (Conducted in June 2004) ************************************************************* Hypothesis: Heating a diskette to even slightly above room temperature is asking for data loss. I overheated a diskette next to a space heater, without the diskette holder protecting the exposed mylar surface. The diskette melted a little bit near the center, at which point I removed the diskette from the space heater's range of heat. The diskette was heated for approximately 10 seconds at a distance of 2-3" (inches) from the space heater. ***DISCLAIMER***: *Do Not* attempt this experiment *at all* unless you have _already_ safely practiced doing so!!! Heating any object near a space heater is *VERY* dangerous and should *NOT* be attempted by unexperienced person(s)!! The reason(s) for testing the heat durability of a 5.25" diskette were two-fold: One, I wanted to disprove the notion that a minimal amount of excess heat can ruin a diskette's data integrity. Secondly, I wanted to test the theory proposed that 5.25" HD (96 TPI) diskettes could still be formatted with a 1541/1571 disk drive despite the small Tracks Per Inch (normal is 48 TPI on the Commodore 64/128 computers for a 5.25" diskette). Conclusion: Heating a 96 TPI (HD) 5.25" diskette (or 48 TPI diskette) to 95-115 degrees fahrenheit does not appear to affect data integrity in any significant fashion, although outright *burning* of the diskette *will* affect data integrity! The 96 TPI diskette, once heated, formatted without error on both the 1541 and 1571 disk drives, but *could not* be reformatted once the diskette cooled to room temperature (approximately 5 minutes time). Additionally, the diskette had to be reheated to a high temperature once again in order to reformat over the previous formatted contents. ******************************************************************** Experiment Three: Diskette burial for 554 days (March 2003-Oct 2004) ******************************************************************** Hypothesis: Burying a diskette for over a period of a year, approximately 1.5 feet underground, will expose the diskette to dirt, rain and other unsavory elements, even when contained within a ZipLock(TM) plastic bag. This experiment took exactly 554 days to complete...1 year, 6 months and 7 days, to be exact. I buried a floppy diskette with (data on both sides) in my parent's backyard at a depth of approximately 18 inches. I made sure that I protected the diskette as reasonably as possible, including placing the diskette inside a paper disk jacket, while also sealing the entire contents inside an enclosed ZipLock plastic bag. Conclusion: Even when sealed in a Ziplock plastic bag, and despite being buried 18 inches underground, both the weight of the dirt atop the diskette and the elements of water and wind severely warped the diskette into a bowl-shaped wedge (think: bowl of soup wedged). The poor diskette was unreadable as is; that is, without removing the mylar internal diskette from the disk jacket itself and placing it inside a new disk jacket (scrapped from a throw away blank diskette I had no general use for). I noted the following damage to the internal mylar diskette as I inserted it into the second disk jacket: A) Physical damage present near the center of the diskette, where Track 18 was likely to reside. B) Minor physical damage to the outer diskette, but mostly on the left side where it had warped upward. Nearly identical damage was evident on the right side, where the same warping was circularly present. C) Non-consequential smearing from dirt, visible just above the (writable) track/sector area of the disk's mylar surface (left of the main diskette ring hub). The disk was inserted into the new diskette jacket and the jacket itself was taped back shut via normal Scotch(TM) brand tape. Conclusion: Because the diskette contained data on BOTH sides of the diskette, I expected the back side to have less data loss than the front side (due to warping/other damage). I was right. The results of the data loss on the front side were as follows: The disk directory read, but failed after only 7 programs had been read (at least 40 programs were on the front side of the diskette in late-March of 2003). This suggested serious damage to the middle of the diskette, namely, Track 18. Not a good start. I began loading the remaining programs one by one. Program 1, of 38 blocks, failed after only 0.5 blocks had been read. Since it was the very first program on the diskette, this suggested it was written very, very close to where the directory itself was written. Program 2, a 3 block file, loaded 2 blocks out of 3 before it, too, failed. Program 3, a 7 block file, loaded fine, with 100% accuracy. However, Line 23 of the program (read in Commodore 128 mode) was complete gibberish, with weird commands like "CIRCLENEW TAN &STEPFOR" where normal, regular BASIC statements would be. Line 24 was normal, as was the remainder of the program. This suggested physical damage to the mylar surface of the diskette. Program 4, a 1 block file, loaded fine, with 100% accuracy, and listed fine as well. Program 5, a 63 block file, failed after only 1.5 blocks had been read. This was very similar to how the first program faired. Program 6, a 9 block file, loaded fine, with 100% accuracy, and listed fine as well (with no errors). This surprised me greatly, as a much smaller program did not. Program 7, a 3 block file, loaded fine, with 100% accuracy, and listed fine as well (with no errors). I have not yet checked the remainder of the 1st side with Disk Doctor 4.0, and have not read through each and every Track/Sector to see how it looks in a diskette editor. However, Side 1 lost approximately 500 blocks of data (roughly an 85-90% data loss). This was the most severely warped (bowl) portion of the diskette, and could only be read by the disk drive when I inserted the internal mylar diskette into a new diskette jacket altogether. The flip side, which was not damaged much at all, only contained minor errors. The programs occupied a maxium of 61 *total* blocks on the flip side of the diskette, leaving 603 blocks free. The directory read without error, although Track 18, Sector 14 was not readable. Because directory data was not on Track 18, Sector 14, data loss on that particular T/S was not evident. The programs (7 in all, out of 7 total) read as follows: Program 1, a 3 block file, loaded 2 out of 3 blocks, but failed. Listing the program only displayed most of it, with the last few lines just pure gibberish. Program 2, a backup of Program 1 (done intentionally), loaded all 3 blocks, and listed and ran fine. Program 3, a sequential file of 3 blocks in length, was read through the following "quickie" program: 5 PRINT"ENTER FILENAME";:INPUT A$ 10 OPEN 2,8,2,"0:"+F$+",S,R" 15 GET#2,A$:PRINT A$;:IF ST<>64 THEN GOTO 15 20 CLOSE 2:END The sequential file read most of the way through, but failed to read anymore data halfway through the last 1/6th of the file itself. Checking a backup of the same file (from another diskette), data loss was between 2-3% of the file, with only the last few sentences lost. Program 4, a BASIC program of 18 blocks in length, loaded 12 of the 18 blocks before failure. Listing the program revealed gibberish about halfway through the program. Program 5, a BASIC program of 11 blocks in length, wouldn't even load (gave a "?FILE NOT FOUND ERROR"; wouldn't even grace me with a "LOADING PROGRAM"..."?READ ERROR" message). Program 6, a BASIC program of 19 blocks in length, loaded with 100% accuracy and without error. The program LISTed and was RUN without any errors. Conclusion: Burying a diskette even 18 inches below the ground for a period exceeding 1 calendar year, even when protected in a ZipLock(TM) sealed plastic bag, is generally *not* a good idea. Water (from rain) and other elements warped the physical nature of the diskette, as did the 18 inches of dirt laying atop the diskette (inside a sealed ZipLock(TM) plastic bag). Even rescuing the internal mylar floppy from the disk jacket and transfering it into a fresh disk jacket failed to solve data loss. Physical damages to the diskette, although minimal to the eye, were nevertheless responsible for a total of at least 38 programs (out of 45 total) either loading only 65-70% of the way, failing altogether (0%-2% of the way) or loading without error (only 5 loaded without error, and the largest program of that variety was a mere 9 blocks in length). Final Conclusion: Diskettes are surprisingly resistant to temperature, but are very fragile to dirt, water and other environmental factors. Even minor physical damages to the diskette surfaces caused substantial diskette errors/loss. Errors noted (on either side) were: #20, #21, #22, #23 and #27; no other diskette errors were noted or found). Sincerely, Paul Allen Panks dunric@yahoo.com ----------------------------------------------------------------------------------- THE DISKETTE EXPERIMENTS, PHASE THREE (out of three) By: Paul Panks (lumberjacks76@lycos.com) (reformatted from original e-mail posting for additional clarity) (Originally posted on September 27, 2006 on the Commodore listservs: homestead@videocam.net.au [and] acug0447@yahoogroups.com). The third phase of the now infamous '5.25" Diskette Experiments' have been completed. These experiments began back in 2003 and are in their third (and thankfully final) installment. I did the tests this evening and came to some (not surprising) conclusions. Here were the three tests: The Diskette Experiments, Phase III ---------------------------------- The Test Disks -------------- I used three separate test disks, one for each of the three tests. On Track 17-0 through 15-5 was placed a 61 block file (15,419 bytes), Cleve Blake- more's Dark Fortress (typed in from the January 1987 issue of Ahoy! magazine). On Track 19-0 through 22-1 was placed a 61 block file, the same aforementioned Cleve Blakemore game. The desire was to have two reasonably sized programs as the first two programs on disk, occupying an aggregate total of 122 disk blocks. The tests would determine three things about each disk: 1) Could a disk still be read by the disk drive? 2) If so, could the original data (both programs) be read and verified without a single bit error? 3) Thirdly, could the disk then be formatted and a simple read/write performed to it? The results of the tests are as follows, with some surprises: Drunk Disk ---------- The original test was going to be beer (1/2 cup) and water (1/2 cup) for thirty minutes. This test was later modified from using beer to oatmeal (1 package), non-fat dry milk (1/2 cup) and water (three 8 oz. cups), mixed in a bowl, which I then let stand for thirty minutes. The disk was then removed and allowed to quickly dry via a paper towel, wiped gently against both sides. The result was a disk that did not read, as the internal mylar floppy could very barely be moved by force from side to side by this person (more than any of the experiments, the oatmeal, water and dry milk mixture really stuck the disk to the jacket quite solidly). Conclusion: The disk is unreadable due to a physical (non-bit) failure of the disk jacket and internal disk mylar heavily sticking to the jacket itself. The Hot and Cold Affair ----------------------- This test called for putting ice cubes on the front side of the disk, while simultan- eously holding the back side of the disk over a stovetop range at Medium heat (held approximately 3/4th of a foot from the surface of the stovetop due to overwhelming heat and potential hand burn considerations). The disk was held over the surface for a period of ten (10) minutes, while carefully juggling the ice cubes on the 1st surface simultaneously. The disk was allowed to cool for a period of 20 minutes, then read. The result was a disk that did not read, as the internal mylar floppy could barely be moved by force from side to side by this person. Conclusion: The disk was unreadable due to a physical (non-bit) failure of the disk jacket and internal disk mylar semi-sticking to the jacket itself. However, after waiting overnight and trying the disk again, it read without error. Performing a simple read/ write on the disk -- writing (then reading back) a 2 block sequential file -- worked flawlessly. Although heating the disk and placing icecubes atop it is not recommended, the disk nonetheless did recover once it was allowed to cool overnight. The Walk About -------------- This round of testing required walking on a disk using three separate shoes, using moderate pressure, for a period of 1 minute each. The shoes were: two business dress- style shoes, with small platform-like heels, and golf shoes (non-metal spikes used). The disk was walked on using moderate pressure and substituting each shoe for 20 seconds each, for a total of 1 minute aggregate time. The disk was then immediately read by the disk drive. The result was a disk that did not read, as the internal mylar floppy could barely be moved (though easier than the first experiment) by force from side to side by this person. Conclusion: The disk is unreadable due to a physical (non-bit) failure of the disk jacket and internal disk mylar semi-sticking to the jacket itself. Experiment Phase III Conclusions -------------------------------- Two of the disks were a total loss, as the experiment never progressed beyond the first question asked ("Could a disk still be read by the disk drive?"). The disk in the "Hot/Cold Affair" experiment failed at first, but then worked upon waiting 24 hours for the internal disk mylar to cool overnight. Disks are not impervious to permanent and irreversible physical damage from oatmeal, milk, water, heat ( >= 250 degrees F ), ice cubes, shoes and golf spikes. Heating a disk and placing ice cubes atop it will cause it to temporarily become unreadable, although the disk should return to working order within 24 hours. It is nonetheless strongly recommended by this person that end users strictly avoid such implements (as described above) when at, near or around a floppy disk or drive. Paul