Forgot your password?
Cyhist Apr. 10 1998 D
Document Actions
========================================================================= Date: Fri, 10 Apr 1998 14:28:08 -0700
Reply-To: "CYHIST Community Memory: Discussion list on the History of
Cyberspace" <CYHIST@MAELSTROM.STJOHNS.EDU> Sender: "CYHIST Community Memory: Discussion list on the History of
Cyberspace" <CYHIST@MAELSTROM.STJOHNS.EDU> From: Fred Cisin <fcisin@MERRITT.EDU>
Subject: Floppy disk formats
In-Reply-To: <199804091456.KAA01814@cyber1.servtech.com> MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
______________________________________________________________________
Community Memory: Discussion List on the History of Cyberspace ______________________________________________________________________
On Thu, 9 Apr 1998, christopher f. chiesa wrote:
>therefore, that the lower the level at which we can transfer physical information to digital, the less drastic our labors need be in the area of preserving, refurbishing, and/or reinventing those 2500 different-and- incompatible custom, PHYSICAL mechanisms. I say, virtualize as much of the job as possible -- and here's how I'd approach it:
Actually, and very fortunately, over half of those formats were HARDWARE compatible! The differences were in different file systems, directory layouts, and sometimes as "minor" as different parameter choices within the same file system.
>Judging from the content of this mailinglist, it appears that the specific technical details of many, if not most, if not ALL, of those 2500 mechanisms are fairly well known: the algorithms by which bits are encoded into magnetic-polarity transitions, the width, spacing, and overall "shape" of the magnetic "tracks" of interest, the identification of "blocks" or "sectors" within each track, the management of a diskful of blocks as a "filesystem" perhaps, and so on and so forth. These things are, if not standard, certainly "fundamental" to some degree, to the point where I expect that the algorithms by which they are performed are probably fairly "well known."
RELIABLE information about disk formats is almost non-existent. In some cases inaccuracies arise from simple misinterpretations (many people, including "tech support" have told me that THEIR disk format has 128 byte sectors, when that was the CP/M "virtual sector" size, NOT that of the physical sector) In many other cases, the person left holding the legacy system has NO information. ("What brand of computer is it?" "ADM3-A") Therefore, regardless of how much technical imnformation is purported to be available, I have yet to add a new format without a sample disk in hand.
>Assuming this to be the case, why not build a generic magnetic-media surface-mapper capable of delivering a complete map of the magnetic polarities across the entire surface of an arbitrary magnetic disk?
I do not have the resources to build such a device. Instead, virtually all floppy disk formats use a very limited variety of drives. Almost all 8" disks are 48TPI. 5.25" disks are 48TPI, 96TPI, or 100TPI (Micropolis and the Tandon TM100-4*M*), 3.5" disks are 67.5TPI or 135TPI. I've considered the possibility of building a system around the Amlyn drive (an early crude jukebox floppy drive with analog track seeking).
Because of the fortunately limited hardware choices, it has been fairly straightforward to build or modify equipment to look at what is on tracks on disks in those drives.
Because the NEC765 (and all subsequent PC FDCs) has a multiple sector read instead of an actual track read, some of my equipment is built around WD 179x FDCs. Other systems are based on intercepting the raw data between the drive and the FDC. The Central Point "Option Board" does a crude job of that. Once the physical parameters of the format are determined, then the rest of the work can be done using sector reading software on the PC.
>Initially, pass a "SQUID" across the surface of the media and record the magnetic field variations it detects. Use image-processing tech- niques on the resulting "map" dataset to isolate the expected/known positions of data tracks for the type of media you're working with (whereupon you can probably THROW AWAY the rest of the map data). Trace the magnetic variations "along" each track, and run the appro- priate (FM, MFM, etc.) decoding algorithm over them to extract a bit- stream. Use similar methods, along with either known specifications or inspection-based reverse-engineering, to reconstruct the byte- stream comprising the track data, extract the "raw disk blocks," in- terpret these as a filesystem, etc. etc., up to whatever level of abstraction you desire.
A very large number of systems use FM or MFM, and use the old "IBM" sector header system, as implemented in both the WD and NEC type FDCs. Therefore, at least the MFM ones of those are possible to read with PC hardware. I've concentrated on those that I can implement with the existing hardware. Mostly I just try to interpret file system issues.
Hard-sectored and GCR disks can not be read with the PC hardware, Even ordinary "single density" FM requires modification to most PC FDCs.
>As I said, in THEORY, at least, it seems almost EASY.
With adequate samples to study, it's merely a matter of trying to decipher the file system data structures. With what we USUALLY get, there are some challenges.
--
Fred Cisin (510) 436-2663 Computer Information Systems Department
Merritt College 12500 Campus Drive Oakland, CA 94619
Xenosoft http://www,xenosoft.com
______________________________________________________________________
Reply-To: "CYHIST Community Memory: Discussion list on the History of
Cyberspace" <CYHIST@MAELSTROM.STJOHNS.EDU> Sender: "CYHIST Community Memory: Discussion list on the History of
Cyberspace" <CYHIST@MAELSTROM.STJOHNS.EDU> From: Fred Cisin <fcisin@MERRITT.EDU>
Subject: Floppy disk formats
In-Reply-To: <199804091456.KAA01814@cyber1.servtech.com> MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
______________________________________________________________________
Community Memory: Discussion List on the History of Cyberspace ______________________________________________________________________
On Thu, 9 Apr 1998, christopher f. chiesa wrote:
>therefore, that the lower the level at which we can transfer physical information to digital, the less drastic our labors need be in the area of preserving, refurbishing, and/or reinventing those 2500 different-and- incompatible custom, PHYSICAL mechanisms. I say, virtualize as much of the job as possible -- and here's how I'd approach it:
Actually, and very fortunately, over half of those formats were HARDWARE compatible! The differences were in different file systems, directory layouts, and sometimes as "minor" as different parameter choices within the same file system.
>Judging from the content of this mailinglist, it appears that the specific technical details of many, if not most, if not ALL, of those 2500 mechanisms are fairly well known: the algorithms by which bits are encoded into magnetic-polarity transitions, the width, spacing, and overall "shape" of the magnetic "tracks" of interest, the identification of "blocks" or "sectors" within each track, the management of a diskful of blocks as a "filesystem" perhaps, and so on and so forth. These things are, if not standard, certainly "fundamental" to some degree, to the point where I expect that the algorithms by which they are performed are probably fairly "well known."
RELIABLE information about disk formats is almost non-existent. In some cases inaccuracies arise from simple misinterpretations (many people, including "tech support" have told me that THEIR disk format has 128 byte sectors, when that was the CP/M "virtual sector" size, NOT that of the physical sector) In many other cases, the person left holding the legacy system has NO information. ("What brand of computer is it?" "ADM3-A") Therefore, regardless of how much technical imnformation is purported to be available, I have yet to add a new format without a sample disk in hand.
>Assuming this to be the case, why not build a generic magnetic-media surface-mapper capable of delivering a complete map of the magnetic polarities across the entire surface of an arbitrary magnetic disk?
I do not have the resources to build such a device. Instead, virtually all floppy disk formats use a very limited variety of drives. Almost all 8" disks are 48TPI. 5.25" disks are 48TPI, 96TPI, or 100TPI (Micropolis and the Tandon TM100-4*M*), 3.5" disks are 67.5TPI or 135TPI. I've considered the possibility of building a system around the Amlyn drive (an early crude jukebox floppy drive with analog track seeking).
Because of the fortunately limited hardware choices, it has been fairly straightforward to build or modify equipment to look at what is on tracks on disks in those drives.
Because the NEC765 (and all subsequent PC FDCs) has a multiple sector read instead of an actual track read, some of my equipment is built around WD 179x FDCs. Other systems are based on intercepting the raw data between the drive and the FDC. The Central Point "Option Board" does a crude job of that. Once the physical parameters of the format are determined, then the rest of the work can be done using sector reading software on the PC.
>Initially, pass a "SQUID" across the surface of the media and record the magnetic field variations it detects. Use image-processing tech- niques on the resulting "map" dataset to isolate the expected/known positions of data tracks for the type of media you're working with (whereupon you can probably THROW AWAY the rest of the map data). Trace the magnetic variations "along" each track, and run the appro- priate (FM, MFM, etc.) decoding algorithm over them to extract a bit- stream. Use similar methods, along with either known specifications or inspection-based reverse-engineering, to reconstruct the byte- stream comprising the track data, extract the "raw disk blocks," in- terpret these as a filesystem, etc. etc., up to whatever level of abstraction you desire.
A very large number of systems use FM or MFM, and use the old "IBM" sector header system, as implemented in both the WD and NEC type FDCs. Therefore, at least the MFM ones of those are possible to read with PC hardware. I've concentrated on those that I can implement with the existing hardware. Mostly I just try to interpret file system issues.
Hard-sectored and GCR disks can not be read with the PC hardware, Even ordinary "single density" FM requires modification to most PC FDCs.
>As I said, in THEORY, at least, it seems almost EASY.
With adequate samples to study, it's merely a matter of trying to decipher the file system data structures. With what we USUALLY get, there are some challenges.
--
Fred Cisin (510) 436-2663 Computer Information Systems Department
Merritt College 12500 Campus Drive Oakland, CA 94619
Xenosoft http://www,xenosoft.com
______________________________________________________________________
