Decoding OS-9 floppy images.

All questions about how to use KryoFlux go here.
ieure
Posts: 11
Joined: Tue Apr 18, 2017 8:23 pm

Decoding OS-9 floppy images.

Post by ieure » Fri Apr 21, 2017 2:07 am

I'm attempting to archive floppies for the Fluke 9100, which was an early 1990s m68k based test instrument which runs Microware OS-9. All factory disks shipped on DSDD 3.5" floppies. There are images of many of the official media in TeleDisk and SamDisk, but the quality of these dumps is variable & many disks have not been preserved.

I have successfully duplicated disks I own (and written disks from images of ones I don't) with ImageDisk on a DOS PC, so I don't think it's a terribly exotic format. Based on the headers of the existing disk images, I believe they're 256 byte sectors with 16 sectors/track, 80 tracks, double sided.

My test disk before I try the factory media is a double-density floppy I created from a TeleDisk image. I wrote it on a high-density drive and confirmed that it works as expected on the 9100 hardware.

I created a stream image of the disk, then wrote it back to a blank media. I was able to boot the hardware with it, so I believe the stream image is good. Where I've been stumbling is trying to decode the stream to create a raw image.

Based on the headers in the TeleDisk image, I believe these are 256 byte sectors, 16 sectors/track, 80 tracks, 2 sides. Based on three bands in the scatter plot, I believe they're MFM encoded. I'm trying to decode with:

Code: Select all

dtc -fout.img -i4 -dd0 -z1 -n16 -m1 -fsd14/track -i0 -e83
This produces a "MFM: <mismatch>, *N" error for every track except 82 / 83, which are unformatted.

I've tried every possible sector size (removing the number of sectors/track when I do so), but haven't been able to get the data to decode.

Here's scatter plots of sector 0 on sides 0 & 1 of the disk:
Screen Shot 2017-04-20 at 5.32.40 PM.png
Screen Shot 2017-04-20 at 5.34.06 PM.png
Can someone suggest how I might get this image to decode?

ZrX
Posts: 291
Joined: Tue Dec 06, 2011 9:09 pm

Re: Decoding OS-9 floppy images.

Post by ZrX » Fri Apr 21, 2017 2:59 pm

The parameters seem ok, not sure if the order matters (it sometimes does).

Could you upload the dump or first few track somewhere to check?

ieure
Posts: 11
Joined: Tue Apr 18, 2017 8:23 pm

Re: Decoding OS-9 floppy images.

Post by ieure » Sat Apr 22, 2017 12:12 am

ZrX wrote:
Fri Apr 21, 2017 2:59 pm
The parameters seem ok, not sure if the order matters (it sometimes does).

Could you upload the dump or first few track somewhere to check?
Full stream dump is here: https://dl.dropboxusercontent.com/u/6120067/sd14.zip

brightcaster
Posts: 79
Joined: Fri Nov 08, 2013 10:48 pm

Re: Decoding OS-9 floppy images.

Post by brightcaster » Sat Apr 22, 2017 9:47 am

As you said before: Its a sector size of 256 Bytes! So it has to be changed in "file/settings/MFM sector image" (if you want to use the GUI). Yes and you have to set the "End Track" to 79 of to avoid adding the 2 aditional tracks to the image.

After that everything reads fine and gives you an image of 640 kByte...

David

ieure
Posts: 11
Joined: Tue Apr 18, 2017 8:23 pm

Re: Decoding OS-9 floppy images.

Post by ieure » Sat Apr 22, 2017 6:50 pm

brightcaster wrote:
Sat Apr 22, 2017 9:47 am
As you said before: Its a sector size of 256 Bytes! So it has to be changed in "file/settings/MFM sector image" (if you want to use the GUI). Yes and you have to set the "End Track" to 79 of to avoid adding the 2 aditional tracks to the image.

After that everything reads fine and gives you an image of 640 kByte...

David
That's very strange. It definitely doesn't work for me. The dtc command in my original post has the -z1 flag, which selects 256 bytes/sector. I tried every possible sector size, and got the same errors.

Could you share the dtc command which the GUI uses? If you launch it from a terminal with java -jar, it will be in the stdout.

I made a profile for the disks, configured as follows:
Screen Shot 2017-04-22 at 10.21.32 AM.png
Here's the main window:
Screen Shot 2017-04-22 at 10.47.18 AM.png
And this is the result; no tracks were decoded:
Screen Shot 2017-04-22 at 10.47.41 AM.png
It produces two zero-byte images:

Code: Select all

proton!ieure:~/Desktop/9100/9132/new-sony$ ls -l out*
ls -l out*
-rw-r--r-- 1 ieure staff 0 Apr 22 10:47 out_s0.img
-rw-r--r-- 1 ieure staff 0 Apr 22 10:47 out_s1.img
This is the command which the GUI runs:

Code: Select all

Command: /usr/local/bin/dtc -l15 -f/Users/ieure/Desktop/9100/9132/new-sony/out.img -e79 -g2 -z1 -n+16 -v250 -i4 -m1 -f/Users/ieure/Desktop/9100/9132/sd14/track -i0 -e83

brightcaster
Posts: 79
Joined: Fri Nov 08, 2013 10:48 pm

Re: Decoding OS-9 floppy images.

Post by brightcaster » Sat Apr 22, 2017 7:09 pm

It's the target rpm of 250 wich is misleading dtc! Set to standard 300 rpm decodes with no errors...

David

ieure
Posts: 11
Joined: Tue Apr 18, 2017 8:23 pm

Re: Decoding OS-9 floppy images.

Post by ieure » Sat Apr 22, 2017 7:15 pm

brightcaster wrote:
Sat Apr 22, 2017 7:09 pm
It's the target rpm of 250 wich is misleading dtc! Set to standard 300 rpm decodes with no errors...

David
Bingo! Thank you for the help.

ieure
Posts: 11
Joined: Tue Apr 18, 2017 8:23 pm

Re: Decoding OS-9 floppy images.

Post by ieure » Sat Apr 22, 2017 8:42 pm

One last question, the GUI gives me separate images for sides 0/1. I assume I can concatenate these to make a raw image, but is there a setting to have it output a single file?

ZrX
Posts: 291
Joined: Tue Dec 06, 2011 9:09 pm

Re: Decoding OS-9 floppy images.

Post by ZrX » Sun Apr 23, 2017 6:14 am

Remove the -g2 parameter and it'll give you single side-interleaved image.

Btw, the testdump shows some bad tracks for me.

brightcaster
Posts: 79
Joined: Fri Nov 08, 2013 10:48 pm

Re: Decoding OS-9 floppy images.

Post by brightcaster » Sun Apr 23, 2017 9:58 am

Yes I've seen that track 5 side 0 for example is mixed up. For me it looks as if there was an attempt to write to the disk with a drive having to low speed, as certain sectors are longer as they should be and overlap to the following ones. The long sectors are all filled with F6 by the way. So I assume that this is just a matter of the procedure the test-disk was undertaken (as decribed in the first posting of this thread).

Nevertheless, this is due to bad disk and not a matter of wrong format descripton. So I gues the factory disks will decode correct with the given parameters.

David

Post Reply