Hi, The aforementioned issue in the Discferret was fixed in early January. We found it after putting the Discferret through some heavy testing and noticing discrepancies. Hey, it happens. Anyway, so far I've been pleased with the quality of the Discferret hardware, and Phil's friendliness and desire to work with others. As with many other open source projects, there's an IRC chat room on Freenode, #discferret, where a lot of development work gets done. The hardware, I can say, is extremely solid. The software has a little ways to go, mainly decoding wise. We're working on it, though anyone is welcome to help. :) Feel free to come and ask questions, even if you don't have a board! I do have a Kryoflux board too ? I got it mainly for testing purposes. The Kryoflux team does not seem interested in working together with interested people at the same level as Phil ? their actually-useful analysis software is very expensive and even dongle protected, according to the FAQ at http://forum.kryoflux.com/viewtopic.php?f=3&t=3#p1241 . The basic DTC software is closed-source and can be rather temperamental, as is the firmware. Though Christian says the Kryoflux "never had such a problem," I have never seen an independent analysis or source code or anything confirming that. I may end up doing this myself. Without source it's definitely more work to verify it. An overview of the engineering of the Kryoflux board indicates it was designed mainly with low cost in mind. The level buffers on it are very ESD-sensitive and are not designed for cables (so a long floppy cable may cause issues), and there's lack of protection overall, especially in the power supply. Anyway, the real beauty of the Discferret is that it can sample at more than 25MHz. The Kryoflux is fixed at a 24MHz sample rate, but with the Discferret I can do 50MHz or even 100MHz. The provided 512KiB of SRAM has plenty of space for multiread ? for a 1.44MB floppy disk, 10 copies of one track at 25MHz, 5 at 50MHz, and two at 100MHz. Of course this slows it down, but the additional data can be used to reconstruct weak and damaged sectors. For doing things like this, 100MHz is absolutely needed: http://chrisfenton.com/cray-1-digital-archeology/ . I have a lot of DEC removable media and I'd like to image it, and I will probably end up using the Discferret. The same can be said about MFM and RLE hard drives. There's no way the current Kryoflux design ever will be able to image high-RPM media. As for Linux support: I've seen several complaints lately on the Kryoflux forums about libusb and glibc issues. Take a look: http://forum.kryoflux.com/viewforum.php?f=3 . Even though they're somewhat responsive, it's still relying on them for support. I've never had a problem compiling the Discferret software, and Phil accepts good patches. :) There's also the issue of the Kryoflux software lagging behind. Currently the Mac version (which is what I use) is a few months old, and lacks many of the features I would like. When I asked about this, I got brushed away, saying that the person who usually builds that version is busy. It's been nearly two months since I asked, and I'm getting irritated. The Linux version lagged behind even more, until it got updated not too long ago. Anyway, feel free to email or join the IRC channel if you have any questions! :) --Dave