Floppy Emu for Lisa progress report: I now have a firmware version that sort-of-works on the Lisa, when using Macintosh disk images, and running Macworks Plus on the Lisa. I can read from an emulated 400K Macintosh disk as well as write to it. From the limited testing I’ve done so far, it looks like the data is read and written correctly. But it’s not really usable as-is, and there are still some significant problems to overcome.
Bit Slip
The reason it wasn’t working yesterday appears to have been a problem with the way Floppy Emu generates something called bitslip markers. The data on a floppy disk is just one long string of bits, with no framing information to show where one byte ends and the next begins. The floppy controller only knows that disk bytes always have 1 as the MSB, so if it reads a byte where the MSB is 0, it means the framing must be wrong. It slips one bit, and tries again. To assist the floppy controller, disks normally contain a series of special 10-bit markers called bitslip markers, with the binary value 1111111100. After five consecutive bitslip markers, it turns out that the floppy controller will always end up with the correct framing, regardless of what framing it started with in the bit stream.
The trouble was that Floppy Emu wasn’t generating 10-bit 1111111100 markers, it was generating 8-bit 11111111 markers. This is one of those cases where you scratch your head and wonder how it ever worked at all, even on the Macintosh. Once I fixed the bitslip generation (I hope), Macintosh disk images started working on the Lisa under Macworks Plus. Woot!
Slowness
The biggest problem is that it’s slow. S… L…. O….. W…… After inserting a disk image through the Floppy Emu interface, it takes about 20-30 seconds before the disk icon appears on the desktop. And file copy operations take so long, you may as well go eat a sandwich while you’re waiting, even for files in the 10K size range. I’m guessing it’s still encountering a ton of low-level errors, causing the floppy controller to retry over and over, until it finally succeeds. So something is definitely still wrong.
I’ve been corresponding with a Lisa owner who’s got a Floppy Emu and Basic Lisa Utilities already set up, and he’s run a few tests for me. What’s strange is that this new firmware version is actually much worse than the old version, according to BLU. Despite the fact that this one actually works on the Lisa and the older one didn’t, this new version shows more disk errors in BLU, and more types of different disk errors. Hmmm.
Stepping Out
The other problem is that Floppy Emu on the Lisa is stepping to the wrong track for the desired sector… I think. On a 400K floppy, there’s only data on one side. There are 80 tracks, and each track has between 8 and 12 sectors. Each sector is 512 bytes. So 512 bytes per track times an average of 10 sectors per track times 80 tracks = 800 sectors and 400K. I’ve written a simple utility program where I can type in the sector number I want to read, and it fetches the data from the floppy and displays it, along with any error code returned by the read operation. With a 400K disk, if I request the last sector (number 799), it should step to the last track (track 79).
I can see the current track number on the Floppy Emu LCD, so I can verify whether it’s doing the expected thing. And when tested on a Mac Plus or a Mac SE, the Floppy Emu with a 400K disk image does what it should: it steps to track 79. But on the Lisa, under Macworks, when requesting sector 799 the Floppy Emu consistently steps to track 35 and returns bogus data. What the heck?
It turns out that track 35 is exactly where you’d expect to find sector 799, if the disk were a double-sided 800K disk. So maybe the Lisa is somehow confused, and thinks there’s a double-sided disk and drive present. But if this were the case, reading and writing files from the disk through the Finder shouldn’t work at all – it should be hopelessly broken. Yet as far as I can tell, it works, just very slowly. Could there be something about my testing program that’s flawed? I just don’t know.
It’s also possible this stepping weirdness is some kind of problem with the particular Lisa I’m using for testing. From what I’ve learned, the Lisa originally supported 400K floppies only. Later an 800K upgrade kit became available, which included a new 800K-compatible ROM chip for the Lisa’s IO board. I’m running Macworks Plus on the Lisa, which is supposed to make the Lisa act like a Mac Plus, which had an 800K drive. But if this Lisa doesn’t have the 800K ROM chip on its IO board, maybe using Macworks Plus will cause this kind of strange stepping problem.
Download
If you’ve got a Lisa and a Floppy Emu, and want to test this out, you can download a few different firmware variations here. There’s a single CPLD firmware file (firmware.xvf), and a few different AVR firmware files (femu.bin), each of which uses a slightly different strategy for bitslips. In my tests, the “address to data gap size 20″ AVR firmware doesn’t work, but all of the others one do. But they’re all very slow. If you try it, please let me know how this firmware works for you, as well as the details of your Lisa hardware and software configuration. Thanks!