Anecdote: Imaging an Amiga Hard Drive.

[Grey Badger]

A personal experience with imaging an old Amiga hard drive, and attaching the image to WinUAE.

I have an A1200 equipped with a Blizzard 1220/4 accelerator card and a Seagate ST9100A 85MB hard drive. Back in the mid 1990s the power supply popped and released the magic smoke. With university looming and lacking the resources to effect repair, it was consigned to storage and largely forgotten.

Fast Forward to the present with the Amiga retrieved and the hard drive extracted:
Armed only with ignorance and luck, I connected the drive to a USB to IDE connector and performed a block copy using the linux dd command. I configured a WinUAE machine as closely as possible to the old A1200 and attached the generated image as a hardfile using the default settings.
To my surprise and utter delight this worked perfectly, and the WinUAE machine behaves almost identically to the old A1200.

I hold a firm belief that if something is working perfectly, I should fiddle with it until it isn't.
Accordingly, I decided that I would prefer to attach the hard drive image to the emulated A1200 IDE controller rather than the excellent uaehf.device. This also seemed to work perfectly, however after some experimentation I found that I had issues accessing data stored on the third partition, with read errors starting from block 41516. These errors are not present when using uaehf.device.

When importing the image file into WinUAE, the following data is presented:
664/4/63, 167328/167552 blocks, 81.7MB/81.8MB, RDSK...@zzMd.... [5244534B 00000040 7A7A4D64 00000007]

When using the A1200 IDE controller emulation, it is these values that are recovered through HD Toolbox:
(Cylinders=664, Heads=4, Blocks per Track=63).

When using fsuae.device, these are the values recovered from the drive by HD Toolbox:
(Cylinders=748, Heads=14, Block per Track=16).

The ST1900A installation manual lists the following as the default logical geometry for the drive:
(Cylinders=748, Heads=14, Sectors per Track=16).

By inputting manual geometry into the WinUAE hardfile settings (Surfaces=14, Sectors=16), and using the emulated A1200 IDE controller results in a perfectly working hard drive. The 748,14,16 values are recovered from the drive by HD Toolbox.

I know very little of drive geometry, or the inner workings of WinUAE. I hope that someone more knowlegable than me will come along to offer further insight.

After further reading, it seems that I was extremely fortunate to recover an image from such an elderly hard drive using a USB-IDE bridge. (It is difficult to believe that the ST1900A speaks LBA properly). The USB-IDE bridge is part of a fairly old adaptor branded as a Safecom SUSB2-F25CA, which also features a Firewire port. The PCB features an NEC branded chip labelled D720130GC 0416X3A76 which appears to be the USB-IDE bridge. There is a second chip covered by a sticker which I assume is the Firewire-IDE bridge.
Data Sheets for the D720130GC are easily found, and the specifications suggest it is capable of PIO mode.
Quote: Compliant with ATA/ATAPI-6 (LBA48, PIO Mode 0-4, Multi Word DMA Mode 0-2, Ultra DMA Mode 0-4)

It is my hope that this post offers some help to someone, somewhere. I also posses a much more modern USB-IDE adaptor, based around an Innostor IS611. If anyone is interested I can try imaging my old drive using this. (This may take a while, as work is looming. Writing this post is proving an excellent distraction).

[rutra80]

Well what I did is copying all the files from Amiga to PC via SMB and it works in directory mode. It's also easier to tamper Amiga files on PC side that way. Isn't directory mode faster than image mode anyway?

[Jerry]

I think it's always a good idea to start with images of drives especially when dealing with old drives with probable errors. dd_rescue can effectively skip bad areas which can greatly speed up salvaging the good parts of a faulty harddrive.

It should be noted however that you can directly mount some FFS Amiga partitions on any linux (see https://www.kernel.org/doc/html/late...tems/affs.html)

[Grey Badger]

A bried addendum to this thread:

I took the opportunity to attempt imaging the old Amiga hard drive using the newer Innostor based adaptor. Unsurprisingly, it failed. Symptoms were similar to those others have previously observed: nonsensical or missing drive data reported and an inaccessible drive.

Just for confirmation I then imaged the drive again using the older NEC based adaptor, and this was again successful. However this is only true when the adaptor was attached to a USB2 root hub. The adaptor would not work when attached to a USB3 root hub. (lsusb did not list the device).

I would like to express my gratitude to Toni and all other contributors to WinUAE. The emulator is excellent, and has allowed me to revisit the good old days of the Amiga!