Chris, After this section at 21:53 I gave you a standing applause right here at my computer screen in my lab...because your solution her is absolutely brilliant! I know first hand from my work recovering QIC Tapes and MFM drives in the last 6 years, that if you want the data bad enough, then this level of extreme research is well worth the effort. Draw the peaks by hand and compare with the CRC...absolutely brilliant!!!!
I'm a preservation expert on 3"/3.5"/5.25" Disks, and i have some informations to share and also react about this video. What is shown at 10:55 is basically the flux at the disk surface that has been displaced. It's due to a user fault : inserting a mouldy disk inside the floppy drive without cleaning the disk. Replace the flux can be done, but it's a very tedious task, and it can take 2 hours just for 1 track on the disk. The other possibility is that the user tried to clean the disk with som alcohol and use a cotton bud with round head or either a sheet with alcohol to clean the mould. This is enough to cause what you see on the screen at 10:55. The best solution to clean a floppy is to use alcohol 70% (not 90%, it's way way too agressive with the disk surface!) with a drop per drop flask, put 2 drops on the disk surface, use a floppy cleaner, and let the alcohol do the trick with the help of the internal coating. My experience and own success on the matter : I have extracted and preserved more than 3000 original softwares from floppies and also more than 2.000 original software tapes. This has only been possible with the help of my method. Thanks for reading me.
Future improvements can also be: look at what kind of data are you recovering, is it a picture, video or sound file or text file (ascii?), or is it a (part of a) program. The bytes that you are recovering can then even be predicted and so you can narrow down the search.
If even for my own notes, here's my recap of your account of Hector Martin's Frequency domain analysis and probabilities of a one versus a zero from section at 30:03 Custom Algorithm - aggressive linear phase low-pass filtering to avoid shifting the peaks around relative to one another - time domain peak discarding (where we know peaks are just too close together to be possible - out of range of drive capacity) - walking the peaks - looking at the timing between them for a zero bit or a one bit - if previous is early or late, adjust next peak with a certain percentage adjustment - Apply human eyes to failures after that
At 28:02, Chris, could you share the specific model of Mitsubishi drive that you used? I'd like to study this further as well. Like you, I prefer and use the TEAC series of drives, so I'd like to see if this "older" model you mention could help me in these efforts as well. I suspect you listed the model(s) of Mitsubishi drives you tried on your chart on the slide at 28:55, but it isn't quite readable on the screen/video capture. Thank you!
The Mitsubishi drive used at 28:02 isn't mine but I believe it's a M4853. Unfortunately, these Mitsubishi drives have a large variance within the same model number. The drive in question has a logic board revision NAMFM. This appears to have the older drive heads but a newer custom Mitsuibishi ASIC for the preamplifier. My Mitsubishi drives are both newer (MF-50x) and neither appears particularly special.
At 10:53 Chris mentions a "slightly aggressive algorithm to decide upper from lower". Can you point us to this algorithm, please? I'd like to do some experiments with it on my own "lost floppy discs". Thank you.
@@scarybeasts Also, thank you! This is quite logical, and I see exactly how you call it "slightly aggressive". Well said. BTW, did you receive my email today?
Greaseweazle is a terrible tool for Commodore disks, wasn't even made to handle them, can't handle half-tracks. Also doesn't implement the SCP format properly, practically useless for Commodore 64 stuff. Better option: get the Super Card Pro - this is the SCP standard, they designed it, does it all. As the old adage says: you get what you pay for, don't settle just cuz GW is cheap
Greaseweazle's also open source - if it doesn't do what you need, I'm sure Keir would be happy to accept a patch to fix it. Also, I've never seen Keir backdate a spec change - unlike Jim Drew's latest SCP change - or send legal threats to anyone archiving the software. Point being, Greaseweazle might not be perfect, but at least it's open source and can easily be fixed!
@@philpem wow, you're really clueless aren't you? Keir has NO interest in making his product more C64 compatible, none! And why the hell would I do all that coding work (which sounds like it's illegal) for free? He sells the product, he makes the money, I'm just supposed to rewrite his total crap for free? And oh joy, I get to get sued for my efforts because it's illegal? Yeah, that sounds like fun, not! Do you have ANY idea what happens if a company DOESN'T protect their commercial product? They lose it! You know Microsoft would sue you into OBLIVION if you tried to sell a product based on their code without permission, right? The same goes for IBM, or any other company, big or small. So you've not only told people they should break the law and sell products illegally, but you've also, to me, made defamatory comments about Jim - which could be actionable. Also, you idiotically suggest that anyone and their dog can produce commercial-grade software just because it's open-source. You seem to have no grasp on reality. This is why commercial companies charge money for products, this is why it takes years to develop something. Even Linus doesn't produce Linux by himself, not even the kernel, not a driver, nothing. So how about you create a product like Greaseweazle - that actually works - but build in the functionality of your oscilloscope into it so it does the analysis automatically? Oh, wait, right, you'll get sued by the oscilloscope company and any company who's code you pilfer to make it without permission. drat, those pesky laws. Everything should be open-source because open-source is always better! Open-source isn't always better, but that's why we have both. Very few people are real coders, it's exceedingly rare. And less than 1% of those are kernel coders, for example. People don't do what you suggest, because they can't. If you don't like commercial products, hey that's your choice, but don't go bashing them. You seem to love the oscilloscope which is a completely commercial product. And there are others in your video that you use that you have no problem with. But you seem to have no problem with copyright infringement, and that's a serious problem.
There's space for open-source and closed-source software. I'm sure SCP is great but my limited experience of the Greaseweazle community suggests it's open-minded, helpful and interested in helping people archive their data!