Strange Request

[customukr]

Hi,

One of our clients wishes to obtain an old JPG of his sons football team from an old redundant floppy drive.

There are two in total, one with a pzp archive (with the jpg embedded), the other one has the single jpg file.

Both the floppy drives have bad sectors, therefore attempting to copy out of the drive onto desktop will fail with both. The drives and files can be loaded into HxD, but as you scroll down the data stream it will lock up. Of course tools like R-Studio etc can't get around the bad sector issue.

I need to clone the device, but have a lack of available tools (apart from DD). What do you think will be sufficient? I don't want to try DD until I know the best course of action as i'm worried the amount of stress the device will receive will leave it unusable.

I appreciate your time.

Thanks,

Chris

[hhddrec]

Good advice Spildit

Hi customukr.

You must clone tools, if you do not know this tools contact with a pro.

[fzabkar]

If you have a very old pc, like a IBM Ps2 you will have better resoults reading damaged floppies at the low speed it uses.

Your statement made me curious, but I was unable to find any reference to any speed other than 300RPM or 360RPM. In fact, AISI, a reduced rotation rate would result in a lower signal amplitude. That's because the EMF induced in the head coil is proportional to the time rate of change of flux, and that is in turn proportional to the RPM. Therefore ISTM that reducing the RPM would make it more difficult to read a marginal diskette.

[Vorian]

But it is true that usb floppy disk drive are *very* sensitive and give lots of errors when reading old floppies

[dick]

It might also help to use one of the old floppy copying utilities in pure MS dos. So copy from source to target floppy.

For example I used to use a little copying utility called 'exact diskcopy' well I think that was what it was called. This was very useful for making identical copies of floppy disks with sectors outside the usual parameters and would also copy any bad sectors. That way we were able to back up copy protected disks for the old Atari games consols and the like.

Do you have old floppy disk drives to test if they are working? As has been mentioned over time the disks and drives seem to go slightly out of specification which means there might be an element of luck riding on this. Sorry to say!

[fzabkar]

@Spildit, I don't for one second question your data recovery experience with floppy diskettes. However, I do take issue with your explanation, and I do that as an electrical engineer. I also have some practical hands-on experience with the PS/2 Model 25 (720KB 3.5" FDD) and the PS1. I did once have the complete set of IBM's technical reference manuals for the Model 25, but I can't recall any specification data for its FDD. However, I do still have the IBM PC/AT Technical Reference Manual.

In fact here are the schematics for the original IBM PC/AT HDD/FDD MFM controller:
http://www.users.on.net/~fzabkar/PC-AT/FDC_HDC/

AISI factors that could explain the PS1's slow FDD copy speed would be track-to-track skew and head skew. In the ideal case the maximum copy time at 300 RPM would be ...

80 tracks per side x 2 sides per diskette x 0.2 seconds per revolution = 32 seconds

The IBM PC/AT standard allows for programmable head load/unload times and stepping rates, among other things. If PC DOS or OS/2 were to opt for more conservative values for each of these parameters than does MS-DOS or Windows, then it could result in the target sector passing by the head before it settles on the next track, which in turn would necessitate an additional revolution. This would significantly increase the copy time.

In fact I once played with a DOS utility that enabled the user to tune these FDD settings, including skew.

http://www.qnx.com/developers/docs/qnx_ ... ormat.html

As for RPM, AFAICT the 720KB and 1.44MB 3.5" drives both spun at 300RPM just like regular PC drives.

See http://retrotechnology.com/herbs_stuff/drive.html

The early 1.2MB 5.25" drives were dual speed (360RPM and 300RPM), but the latest ones were 360RPM single speed drives.

The data separator on the PC/AT FDD controller could be configured for 250kbps, 300kbps, or 500kbps data rates. If you were to slow the drive down, then the data separator's PLL wouldn't have enough range to lock onto the bitrate. That is, unless the PS/2 were to use something different.

In fact the following device appears to be able to accommodate all of IBM's standards, including 2.88MB FDDs, so it does appear that the PS/2 was no different in regard to bitrate.

DP8473 Floppy Disk Controller:
http://www.datasheetcatalog.org/datashe ... 009384.PDF

"This controller is a full featured floppy disk controller that is software compatible with the uPD765A but also includes many additional hardware and software enhancements. These enhancements include additional logic specifically required for an IBM PC PC-XT PC-AT or PS/2 design."

"This controller incorporates a precision analog data separator. ... This provides optimal perform-
ance at the standard PC data rates of 250, 300 kb/s and 500 kb/s. It also enables optimum performance at 1 Mb/s."

So AISI the 720KB 3.5" drives would be operating at a bitrate of 250kbps, and the 1.44MB drives would be running at 500kbps. This would make them no different to ordinary FDDs.

As for why the PS1 drives gave you a better recovery rate, I can only imagine that they may have had better alignment tolerances, or perhaps their heads were more sensitive, or maybe the head coils were slightly wider. Whatever the reason, I can't see how it could be related to copy speed.

[fzabkar]

Guess that my IBM PS/1 Controller and drive are as good as the Kryoflux lolololol

I'm guessing that your PS1's FDD controller would have been nothing special. The Read Data signal is TTL, so all the analogue-to-digital conversion would already have been done on the drive. All that the controller would need to do would be to extract the clock and data from the MFM stream. If you still have the controller, I'd be very curious as to which chip(s) it uses.

As for KryoFlux, it appears to be primarily a software solution. Even though it has its own controller, it still depends heavily on the quality of the drive for its success. It bypasses BIOS, but that's its only real advantage, AFAICT. It can read the actual GAP and ID fields, and it would be able to set up its own data rates, so that would enable it to handle non-PC formats. But IMHO, if you wanted a genuine data recovery solution, then you would need a specialised floppy drive. For example, I would think that a micro-stepping feature would be desirable. This would allow misaligned diskettes to be recovered. I would also want a drive whose read gains could be adjusted, thereby circumventing the drive's AGC.

[fzabkar]

What would you suggest for the OP to do in orther to get a better chance to recover the data from the floppy ?

Send it to you. :-)

[fzabkar]

What would you suggest for the OP to do in orther to get a better chance to recover the data from the floppy ?

There is a tool called Bad Block Copy for Windows. It works like ddrescue, but it works on individual files. You can use it to automatically reconstruct a single file from multiple corrupt sources.

http://alter.org.ua/soft/win/bb_recover/

In the old days I used DiskEdit from the Norton Utilities. It would identify all the sectors/clusters belonging to a particular file. You could then manually copy those sectors to a target drive.

Before doing anything, I would examine the diskettes for physical damage, eg circular scratch marks and mould, as you have already said. I lived in Singapore for 2 years where it is humid every day, so I know that mould is a very real problem.

[fzabkar]

And by the way, if the floppy that you are trying to read is a low density one (only one window to write-protect and no window on the other side of the floppy) modern USB floppy drives will not even read those !!!! (The majority will only read HD floppy)

I haven't come across those myself, but I wonder what would happen if you chose 720KB for your drive type in BIOS.

In fact I have an old 486 machine running Win95 DOS. It has a 3.5" FDD. If I choose a drive type of 1.2MB 5.25" in BIOS, then I can still read the data on a 1.44MB diskette without error, ie ...

copy a:*.* nul

However, if I choose a drive type of 720KB 3.5", then I get an Abort/Retry/Fail error.

The reason is that the 1.2MB and 1.44MB FDDs use the same bit rate for the data separator, namely 500kbps. The former rotates at 360RPM while the latter spins at 300RPM.

15 sectors per track x 6 revs per second = 18 sectors per track x 5 revs per second

15 x 6 x 512 x 8 = 18 x 5 x 512 x 8 = 368640 data bits per second

[fzabkar]

I prefer to think of it as a discussion.

[hhddrec]

Hi fzabkar and Spildit, you are two gurus.

thanks 4 your coments

I thin your two ways are good advices. sometimes maybe one baetter than other but all recovery cases are diferent and those two ways are very pretty.

[fzabkar]

It would be nice to know if op got the photos from the floppy at the end...

I'd like to know if it was a 5.25" or a 3.5". All he said was that they were "old and redundant". I still have both floppy drives in my 486 machine, so I wonder if that makes me old and redundant, too. :-)

[customukr]

My apologies been inundated at work and had to put this to the side for a day.

Seems I've struck up quite a discussion, and many many thanks for the points laid out by all of you.

I will upload a photo of the floppy drives once i get to work tomorrow morning. Incidentally I was with the client this morning completing an SBS installation and I stated there are two ways to try this, either you outsource it, or I give it a go. He was willing for me to try and doesn't mind if it becomes unrecoverable. It still holds sentimental value so it's worth a shot but not the end of the world.

I was attempting to read the floppy through an external USB floppy reader (which I believe is quite recent) so first thing to try would be the older floppy drive.

Sorry for taking a while to respond.

Chris