"Imaging" a MFM drive?

[Steven Hirsch]

On Tue, 6 Oct 2009, Joachim Thiemann wrote:

>> There are a few things I'd be worried about here...
>>  - You need to know what parameters were used when formatting the drive.
>> Usually these will be the ones printed on the drive label (or in the
>> instruction book) but sometimes e.g. the cylinder count is reduced to get
>> around bad cylinders, and so on.

That's only part of the picture.  Many of the Xebec MFM/SCSI adapters had 
their own reserved area on the drive for configuration and bad sector 
maps.  I wouldn't bet money on the raw track/sector format lining up with 
anything an ISA bus MFM controller would understand.

> Yes, dd would be nice, but that assumes a 32-bit machine with a driver
> for the card.  That is my concern.
>
> Hmmm, maybe an old Minix system could talk to the card :-) od the raw
> device to the serial port, then capture the octal data and reassemble
> the image file...

Given the above proviso, that's definitely the best bet.  If you can 
address the drive as a raw device it should be possible to figure out 
offset to start of "real" data.

For extra credit, it would be fun to see if a modern SCSI adapter can be 
made to communicate with the command syntax supported on early Xebec 
boards.  The earlier ones were only kinda, sorta SCSI-1 compliant.  I 
rememember seeing a Linux kernel driver for talking to the Xebec 1410 
series, but it was probably more than 10 years ago.

Steve



--