This may seem like a niche win, but I’ve been battling with a complex web of legacy hardware unknowns on my IBM 5160 XT build, and I wasn’t getting anywhere fast. I wanted to document the breakthrough here to save others a massive amount of frustration

The hardware:
IBM 5160 - running an intel inboard 386
Trantor T130b with bios
Fujitsu M2512A in an external enclosure
3x 2nd user disk form eBay

The software:
scsiworks 1.3 (Internet Archive)
Fujitsu fudd204.zip including fjfdisk 2.06 (theretroweb.com/floppydrives/1507)

The hardware guide:
https://www.bitsavers.org/pdf/fujitsu/Fujitsu_M2512A_LR.pdf

The Setup and Initial Failures:

I mapped the Fujitsu drive to SCSI ID 0. Because the drive was physically filthy inside, I did a deep mechanical cleaning. I cleaned under the internal prism that sits above the media paths, and carefully inserted an unpowered cartridge with the drive lid off to safely clean the laser lens assembly and the prism entry point on the optical pickup unit.

I fired up the system, loaded the ASPI layer (MA13B.SYS), and the drive mechanism detected beautifully.

However, when I ran the official fjfdisk.exe to format any of the three eBay disks, the utility would spin up for 5 to 10 minutes, fail and throw an exception stating the disk could not be found in very broken English.

No matter what DIP switches I modified on the back of the drive, fjfdisk stubbornly refused to initialize any of the cartridges. I was at the point of assuming either the drive's laser diode was completely dead or all three disks were toast.

The Secret Hack: The "Fixed-Disk" Dumb Pipe Override

To isolate the driver layer, I decided to trick the system into dropping all its intelligent removable-media safety checks. I reconfigured the drive switches to force Fixed Disk / Direct Access Emulation Mode and load Trantor's native raw hard disk block driver instead of the floptical drivers.

PHASE 1 DIAGNOSTIC DIP SETTINGS
(Hard Drive Emulation) SW1 Bank (Fixed Hard Disk Emulation Mode):
-------------------------------------------
Key 1: OFF \
Key 2: OFF |--> Sets SCSI ID to 0
Key 3: OFF /
Key 4: ON |--> SCSI Parity Enabled
Key 5: OFF |--> Synchronous Mode Disabled (Forces safe 8-bit Asynchronous timing)
Key 6: OFF |--> Direct Access / Fixed Hard Drive Emulation Mode (CRUCIAL OVERRIDE)
Key 7: ON |--> Spindle Auto Stop Enabled
Key 8: OFF |--> Normal Activity LED Mode

SW2 Bank (Asynchronous SCSI-1 Layout):
---------------------------------------
Key 1: ON |--> Internal Drive Write Cache ENABLED (Speeds up formatting)
Key 2: OFF |--> Reserved
Key 3: OFF |--> Macintosh Mode Disabled
Key 4: ON |--> Verify After Write Disabled (Let TFORMAT handle verification math)
Key 5: OFF |--> Forced to Legacy SCSI-1 Protocol (Bypasses 8-bit bus flooding)
Key 6: OFF |--> Default Data Pointer Behavior
Key 7: OFF |--> Standard Factory Operation Mode
Key 8: OFF |--> Standard Factory Operation Mode

I then modified my CONFIG.SYS to load DEVICE=C:\TSCSI\TSCSI.SYS immediately below my ASPI manager (MA13B.SYS).Upon booting cleanly to the DOS prompt, TSCSI.SYS successfully detected the unformatted drive as a standard fixed disk and assigned it drive letter D:.

I launched Trantor’s raw low-level formatting utility using the sector-forcing switches:
format d: /L /S512

The utility verified the target ID. When prompted for parameters, I entered an interleave of 1.

Thanks to the system speed, the low-level format pass completed in a blistering 15 minutes instead of the stated hour.

Once the format hit 100%, TFORMAT asked for partition boundaries. I selected S for Single Drive, and allowed it to execute the sector verification pass. It successfully mapped out a Large DOS (LRGDOS) master boot record. At the very end, it attempted to copy DOS system files and dropped an error because command.com wasn't in its immediate path.

This caused the drive firmware to lock the cartridge inside using its internal physical solenoid latch (a standard SCSI prevent-medium-removal safeguard).

Returning to Native Removable Floptical ModeI

I powered down the external enclosure, inserted a small tool into the front panel's emergency mechanical eject hole to pop out the disk, and flipped SW1 Key 6 back to ON to restore native Optical Removable Mode. Crucially, you must leave SW2 Key 5 set to OFF (SCSI-1 mode) for the final removable driver configuration.

Forcing the drive down to legacy Async SCSI-1 drops its vendor ID string to "NECITSU" and strips out the advanced SCSI-2 command negotiations that flood and desynchronize the limited 8-bit ISA bus during system boots.

PHASE 2 FINAL OPERATIONAL DIP SETTINGS
(Hot-Swappable Removable) SW1 Bank (Optical Removable Mode):
----------------------------------
Key 1: OFF \
Key 2: OFF |--> Sets SCSI ID to 0
Key 3: OFF /
Key 4: ON |--> SCSI Parity Enabled
Key 5: OFF |--> Synchronous Mode Disabled (Forces safe 8-bit Asynchronous timing)
Key 6: ON |--> Optical / Removable Media Mode Enabled (RETURN TO NATIVE REMOVABLE)
Key 7: ON |--> Spindle Auto Stop Enabled
Key 8: OFF |--> Normal Activity LED Mode

SW2 Bank (Asynchronous SCSI-1 Layout):
---------------------------------------
Key 1: ON |--> Internal Drive Write Cache ENABLED (Massive day-to-day performance boost)
Key 2: OFF |--> Reserved
Key 3: OFF |--> Macintosh Mode Disabled
Key 4: ON |--> Verify After Write Disabled
Key 5: OFF |--> FORCED TO LEGACY SCSI-1 PROTOCOL (The missing link to clear "Invalid Media Type")
Key 6: OFF |--> Default Data Pointer Behavior
Key 7: OFF |--> Standard Factory Operation Mode
Key 8: OFF |--> Standard Factory Operation Mode

I edited my CONFIG.SYS back to loading the official Fujitsu removable media driver:

DEVICE=C:\SCSI\MODISK2.SYS

I booted the machine with the drive empty. MODISK2.SYS cleanly mapped a placeholder drive letter without any boot-time geometry conflicts. Once sitting safely at the C:\> prompt, I pushed the disk in and launched fjfdisk.exe one final time.Because the underlying physical track geometry was now fully compliant, fjfdisk accepted the disk, I selected the Super Floppy Format layout, assigned it a standard volume label (DISK1), and exited.

Changing to drive D: now works flawlessly. The disk is fully readable, writable, and behaves as a true hot-swappable 217MB floptical volume under DOS and Windows 3.11

Why Did the Initial Formats Fail?

It turns out that many surplus 230MB MO disks sold online were originally formatted on vintage Macintosh or Unix workstations, which burns a physical 2048-byte sector layout into the substrate.

PC-XT architecture, standard ASPI layers, and DOS utilities require 512-byte sectors. When fjfdisk encountered those alien 2048-byte physical blocks while in removable mode, it lacked the low-level permissions to overwrite them, causing it to panic and drop the session.

By forcing the drive into fixed hard disk emulation and utilizing Trantor's raw TFORMAT utility, we created a perfect "dumb pipe" that bypassed the OS driver blocks.

This allowed the physical laser assembly to safely crush the old Mac boundaries and format the substrate back down to a standardized 512-byte PC layout, reviving media and hardware that otherwise looked destined for the bin.

Two out of my three eBay disks processed through this assembly line perfectly (the second disk turned out to have severe physical concentric scratches from an old head-crash). A 66% survival rate on my 2nd user disks.

As a bonus the Fujitsu driver also handles nicely my zip100 drive so I got to free some additional conventional memory also in the mix. Double win!

Anyway I figured I’d document and share as it may just help someone else with the question of is it the drive or the media and perhaps get an old MO drive working fully with an 8bit card.

I appreciate my setup might be slightly niche but these learning curves also can sometimes offer up answers to others.

As an aside I also have hooked up an external ls120 via parallel and the tformat could also low level format disks in the drive which also saved a disk destined for the bin with bad sectors, so where native dos tools like format couldn’t detect or wipe the disk tformat sent low level scsi commands to the aspi layer and formatted a dead disk marking the bad sectors and recovered the media.

A great little tool that got me onto trying it with the MO disks.

Anyway hope that may help someone trying to keep those vintage drives and media going for another day.

Do note I got AI to help clean up the text above but I’ve manually amended it to read nicer. Hopefully that captures the technical details for those that need it.