Floppy Disk Recovery

Updated 19-Apr-2026 21:24 -0400

INTRO

Several weeks ago, @BrokenPipe found a set of 17 8" disks on eBay which looked interesting. Some were labeled CP/M-86, others MS-DOS, and one Seattle Computer Products MS-DOS 2.0 which was sadly marked "bad". I already had images of the MS-DOS 2.0 "OEM Adaptation Disk" from my work restoring the Gazelle held in the VCF collection (this is discussed here).

I bought them anyway, and it turned out to be both a good and very interesting purchase, and not all purchases are created equal.

THE BITS

Here are the disks. An interesting lot for sure. A mix of SSSD and DSDD, and as we discovered, several "flippy" disks with three index holes. Flippy disks are DSDD disks that can be flopped over and used as two single-sided disks in an SSSD or SSDD configuration (like the SA-800, for example). This will prove to be an important fact to know later. "PN" means "part number", and the 800810 disks are "flippy" disks.

Dysan 3740/2D PN800810 (image)
IBM DisplayWriter 1 (image)
Dysan 3740/2D PN800810 (image)
IBM DisplayWriter 1 (image)
Dysan 3740/2D PN800895 (image)
Memorex DSDD "MS-DOS 2.0 DSDD 9-21-86 #1/MS DOS Seattle Computer (Gazelle)"
Dysan 3740/2D PN 800895 "Digital Research CP/M-86 1 of 2" (image)
Dysan 3740/2D PN 800895 "Digital Research CP/M-86 2 of 2" (image)
Memorex DSDD SCP-labeled "2.0G WNHD" (image)
Disk with no label or sleeve (image)
Memorex "SYSTEM DISK/work disk" "MS-DOS (Gazelle) 4/84" (image)
Dysan 3740/2D PN800810 "MS-DOS #1" (image)
Dysan 3740/2D PN800810 (image)
Dysan 3740/2D PN800810 "MS-DOS #2" (image)
Dysan 3740/2D PN800895 (image)
Dysan 3740/2D PN800810 "MS-DOS #3" (image)
Dysan 3740/2D PN800895 (image)

Disk 6 was not recoverable, although work is ongoing to recover it.

THE BYTES

For imaging and analyzing, several tools are key:

Good floppy drives. 8" drives are exceedingly rare and when found, very expensive. I have two Shugart SA-851 drives in SA-850 mode, and two Qume QT-242 half-height drives. The Shugart drives are the "gold standard" for work like this, but the Qume are also acceptable but experience more issues with excessive head pressure or stickiness of the head solenoid. Proceed with caution on those.
Disk imager. I use the Greaseweazle v4.1 and the built-in command line tools.
Disk image viewer. I use HxC for that.
50-pin adapter. This adapter is needed to connect the 50-pin drives to a standard 34-pin controller or the Greaseweazle. Although there are a few, I use the Floppy Adapter from John Wilson's DBit. This is NOT the same as the usual 34to50 adapter found on Github or elsewhere. Those are designed to connect a 34-pin ("standard") floppy drive to a 50 pin (S-100) floppy controller. If not carefully configured, it will ground several output drivers on the 8" floppy and does not handle the TG43 write-current adjust.
Various cables.
Floppy disk cleaning kit. 8" ones are rare now, but if you can find one on eBay, grab it. I can't overstate the importance of keeping the heads clean because any build-up of oxide can scratch subsequent disks.
Imaging "flippy" disks is a unique process because not every floppy drive interprets the multiple index signals properly. For example, on the Shugart drive, it was able to handle the multiple index pulses fine. On the Qume drive, it reads both pulses, messing up the images. So, depending on the drive you have to cover the SD (middle) hole with an IR-opaque sticker to ensure only one set of index pulses is used. Using the disk as a flippy disk in an SSSD drive is fine. For the sticker, I used a write-protect sticker. When doing this, however, you have to ensure the adhesive doesn't contact the disk itself. I put a piece of Kapton tape on the back of my sticker, and it worked quite well.

What's interesting about this disk collection is that there is some randomness of the collection.

Disks 1, 3, and 5 appear to be an OEM Adaptation version of MS-DOS for the Gazelle with the Fulcrum Computers OmniDisk floppy controller. On the SCP page is a copy of the related system boot ROM. Disks 3 and 5 are similar to Disk 1 but lacks the compilers. Disks 12-17 are similar disks, but have files spread among the six disks. A full comparison of the files has not yet been done.
Disk 2: Unknown
Disk 4: Unknown
Disks 7 and 8 -- two CP/M disks appear to be OEM distribution disks for the Intel iSBC86-12 Multibus board. Introduced in 1978, this board was basically a "reference design" using Intel chips, some of which are now very rare and had only limited distribution.
Disk 9 is a Seattle Computer Products OEM disk. The SCP-written legend is "2.0GWNHD" which I interpret as v2.0G "with no hard drive", so no HDISK.BIN driver installed. The original hard drive was a Quantum Q540 5MB MFM drive. The system-side interface is unknown, but on mine, a Morrow HD-DMA controller is used.

Another interesting fact about Disk 9 is that it's marked as "Bad" by someone. This is where @BrokenPipe and I got deep into analyzing the flux data. The disk didn't image properly but upon closer inspection, it seemed it was originally written in 1983 at 300 RPM and someone in 1986 accessed the disk on a system with a drive that was running at 360 RPM. This access is what trashed the sectors, and subsequent accesses failed due to the errors, and thus deemed a "bad" disk.

A slower RPM increases the recording density by about 20% in comparison to 360 RPM. To my knowledge, there are no 300 RPM 8" drives. Original 8" full-height drives used an AC/line powered synchronous motor, so the RPM rate was fixed (1800 for 60Hz, 1500 RPM for 50Hz) which was then "geared-down" by changing the drive pulley and belt to get 360 RPM. Once they made 8" half-height drives, AC synchronous motors were no longer used, instead relying on standard DC motors with a speed control tuned by a potentiometer.

Once the image was adjusted for the proper bit cell width, every thing came into focus, and the data became recoverable. The disk was not full, so there would be no reason to increase the bit density, especially since the format had not changed from the standard 77/8/1024 for that type. One possible, although remote, explanation is that the disk was made by a system running on 50Hz current, but as it was an SCP master disk, that would be unlikely.

Years ago, I imaged a similar OEM disk for one in the collection at the Vintage Computer Federation. In comparing the disks, this disk is dated about one month prior to the other disk. Also, the directory structure is slightly different, segregating the ASM source from the OBJ output and finally BIN. It was writing to the BIN directory in 1986 that wiped out two data sectors and part of the directory sector.

Disk 10: Unknown
Disk 11: Unknown

Home

Copyright (c) 2006-2026 Richard A. Cini, Jr. (rcini at msn dot com) All Rights Reserved. All copyrights of any third parties referred to herein are hereby acknowledged. There is no warranty, either express or implied, relating to any of the content contained herein. The site maintainer shall in no event be liable to anyone for damages, including any loss of profits, lost savings, or other incidental or consequential damages arising out of the use or misuse of the information contained on this Web site.

You may use the information contained herein for NON-COMMERCIAL purposes only and AT YOUR OWN RISK. Batteries not included. Contents may settle during shipping.