Preserving a floppy disk with a logic analyzer and a serial cable

[czietz]

I tied together two pieces of hardware I already had (logic analyzer and USB-to-serial cable) with a small Python script to preserve floppy disks at magnetic flux level, a bit like e.g. a Kryoflux would do. You can read the details here:
http://www.chzsoft.de/site/hardware/pre ... -analyzer/

[ijor]

Amazing! A very strange and unusual way to perform this job. But shows an amazing creativity. In a way, it reminds me of the works done by Vicent Joguin's and his Disk2FDI hardware. Nice retro geek stuff

[Brume]

Interesting
Does the SCP file you generated work with emulator?

[czietz]

I loaded it successfully into Steem SSE, yes.

[mikro]

Great hack in the true meaning of the word.

[joska]

Great project I see the Kryoflux is quite expensive, but judging from your experiment a €5 Raspberry Pi Zero should be plenty quick enough to handle this.

[czietz]

Basically you need a device that is able to sample two digital signals (index and read data) consistently at a sufficiently high sample rate (min. 10 MHz, preferably 40 MHz). I'm don't know if a Raspberry Pi alone can do this. Often, embedded Linux boards have issues with such hard real-time requirements.

[joska]

You don't have to run Linux on the RPi, there are plenty of alternatives ranging from "bare metal" libraries to RTOS-solutions.

[stormy]

Perhaps you could be the guy to finally make a successful copy of Crown of creation?

[czietz]

Perhaps you could be the guy to finally make a successful copy of Crown of creation?

My setup isn't more powerful than, e.g., a Kryoflux. It just has the advantage (for me) that I already had all necessary hardware.

Also, I don't have that game, otherwise I would of course try.

[ijor]

Perhaps you could be the guy to finally make a successful copy of Crown of creation?

czietz's method (and I have to insist, a very creative one ) can be used only to dump a disk. Write back is a little more complicated.

Btw, Christian, you might want to store a little more than "just" one revolution because some disks have so called data under the index. It is not strictly necessary to store multiple revolutions, not for this purpose, but at least store up to the next trailing edge of the index pulse. This would provide at least some overlap.

[czietz]

Btw, Christian, you might want to store a little more than "just" one revolution because some disks have so called data under the index. It is not strictly necessary to store multiple revolutions, not for this purpose, but at least store up to the next trailing edge of the index pulse. This would provide at least some overlap.

I have to see how this can be mapped to the scp file format. As far as I understand it always stores complete revolutions.

[ijor]

I have to see how this can be mapped to the scp file format. As far as I understand it always stores complete revolutions.

Yes, you are right. If you want to store it as SCP then you can't use fractional revolutions. You would need to decide if it's worth to store two revolutions, use other formats, or just ignore the issue of data under the index.

For ST disks it happens mostly (probably only) on copy protected disks, and even then, not on every copy protected disk, of course. If you want to consider other architectures, especially some using 5.25 disks, then it is much more common. E.g., Atari 8-bit drives didn't have an index hole detector at all. So all user made disks had data under the index.

There is a possible workaround that is making sure that the stored revolution can be used as a circular buffer. You would need to analyze two consecutive revolutions, perform some pattern matching, and then you might need to remove or to add a couple of transitions from the next revolution. This is because the index pulse has some jitter. But this is a bit of a kludge. Not every tool would know how to recover the data under the index from a single stored revolution.

[czietz]

I have modified my script to store a second revolution in the .scp file. Resulting image loads fine in Aufit, the HxC software and Steem SSE. Now I just have to find a floppy with "fuzzy" bits.

[ijor]

I have modified my script to store a second revolution in the .scp file. Resulting image loads fine in Aufit, the HxC software and Steem SSE. Now I just have to find a floppy with "fuzzy" bits.

Nice!

Weak bits aren't necessarily different across revolutions, although they do are different in most cases. In either case you don't necessarily need multiple revolutions to detect weak bits, but it does make the processing easier and more accurate.

[JimDrew]

I just saw this post for the first time! I have been a long time Saleae beta tester. I actually used the Logic device when developing SuperCard Pro to capture the signal that the board controls.