FAQ: 1. 5.25" drives come in 48, 96 or 100TPI versions. 3.5" based on the Sony mech are 135TPI. 2. I didn't go much on the concept of "Single Density" and "Double sensity." but there is no such thing as single density or double density drive or floppy disk. They are both just encoding methods that the floppy disk controller speaks. It's like saving a picture in GIF or PNG. One is just more efficient than the other, but they both do the same thing and you can save those files anywhere. The data rate of FM or MFM encoding can be 250kbps or 500kbs as the data rate doesn't have anything to do with single density FM or double density MFM encoding. All the disk drive sees is a stream of 0s and 1s, and it saves those onto a disk and then can read then back. The controller IC itself is what "speaks" MFM or FM, and it will decode or encode data to or from those encoding formats. This Wikipedia article has some additional information: en.wikipedia.org/wiki/Modified_frequency_modulation Things get a bit messy with High Density. I did mention that High Density is just double density MFM encoding used at 500kbps on 5.25" or 3.5" disks. It only applies to 5.25" or 3.5" as 8" disk drives always used 500kbps as their data rate and early controllers used FM and later controllers switched to MFM. The NEC APC wrote at 500kbps MFM double sided, but this was not called high density, it was always called double density because of the MFM encoding. IBM muddied things up when they made the 5.25" High Density drive with the non standard 360rpm and 500kbit data rate. Worse yet, in order to pack in twice as much data using 500kbps, the drive needs to use a different write strategy for 500kbps and the floppy disks themselves needed to be reformulated. Think of Metal Type IV audio takes versus Normal Type 1. The problem is HD 5.25" disks are physically identical (As far as the drive is concerned) do the 250kbps SD/DD version. So the drive doesn't know what to do. Sony fixed this with the 3.5" High Density disks later, but even those just use MFM at double the data rate for high density. It's still technically a double density disk as far as encoding goes, just double the data is packed into each track. I know it's confusing! Hope this helps a little! 3. It might be possible to use RLL encoding on floppy drives. This would give another 50% improvement in storage capacity over MFM using the same 250kbps! FM -> 50% better -> MFM -> 50% better -> RLL. This was done on ST-506 hard drives (often called MFM drives) which used MFM encoding primarily, I think at 5mbit. Switching to a RLL controller would give you that 50% capacity and speed boost. (Note that MFM drives almost always spin at 3600rpm, 10x what 8" drives ran at.) The only problem with RLL on floppies is their speed isn't as stable as a heavy spinning hard disk, so that instability might lead to problems decoding the data properly. A big part of how RLL is more efficient is how less clock data is encoded into the data, but that means that instability in the speed of the disk would result in data loss. 4: Hard sector disks. I didn't touch on this either and it's another wrinkle in the floppy drive universe. These disks are the same as normal 5.25" disks, but instead of one index hole, there are usually 16. (Or is it 18?) The index holes create pulses that the drive controller use as a reference to where to write sector data onto the floppy disk. These were needed on old primitive controller cards that did all of the encoding using TLL logic. Once highly integrated LSI controller chips like the WD1791 came out, they no longer needed all of the index holes as the controller only needed one index pulse to synchronize itself. This allowed for the use of variable number of sectors per track, as the controlled IC was handling all of that. Some floppy drive systems like the GCR encoding on the Apple II and Commodore don't use the index pulses at all. This is why you can flip those disks over to use the back side, while this is not possible with systems using these LSI controller chips. You can use a hard sector disk just fine in a Commodore or Apple II, as the index pulses aren't used, but you cannot use a hard sector disk in a system that is "soft sectored" as the controller will be immediately confused by the large number of pulses. You can use a Kryoflux to make disk images of hard sector floppy disks, using command line switches. It can then write back the images as well. DIsk drives themselves are not hard sector or soft sector. It is strictly the controller that requires one type of disk or another. (With one index hole per track, or many.)
When you have a spare week or so, maybe you can make a revised version of the video that includes these this info, as it's very good info that some may not realize is in this FAQ.
I've found this page to be really helpful too. It covers technical specs for both disk formats and drives/controller: en.wikipedia.org/wiki/List_of_floppy_disk_formats
I believe it is possible that you can do the reading and writing of wide tracks and other odd formats even with the standard PC floppy controller as it might have a manual mode that lets the CPU take control. If that is true the question would be if somebody already wrote the software? Or maybe somebody can conjure up a way with SDR and other tricks.
You Sir, are a whole gift to the entire retro computing world! I recently used the TRS-80 diagnostic ROM you co-authored to resurrect one of my Model 3s. Your channel motivated me to finally get working on them, and I’ve started a video series on the process!
The 3" drives you mention for the "Amstrads" are Hitachi 3" mechanisms. They run at 250kbps, and just use the standard signalling. They are configured as single sided units, but the disks can explicitly flipped.
I distinctly remember i had a 3" Drive hooked up to my 386 back in the day. The low storage made them almost useless, but I used them as boot disks for different configurations. Until I figured out config.sys menu commands, that is
@@TSteffi The Amstrad CPC machines that used these drives were simple Z80 based machines with 128k of RAM, so the capacity was fine for those. Can't imagine them being much use on a 386, though.
@@fattomandeibu To expand, the Amstrad CPC range came in 3 memory configurations 64K (464 / 664), 72K (472 which was Spain only, due to a wierd tax thing applied to machines with less than 64KB when imported), and 128K (6128). The 664 and 6128 were the "base" models that came with the Disk drive (the 464 and 472 were cassette). There was also the "plus" range, but the + versions still came with the same storage mediums as the "normal" models (although they had an additional cartridge slot for the ROM at loading) Going from memory, the disk held *up to* 180K (2K directory, 178K actual usable storage) and was double sided (so a total of 360K per 3" disk)
@@gangrel_eu The even funnier thing about the 72K CPC machines for Spain was that the 8K extra ram was just a faulty ram chip soldered onto a separate PCB and added inside the case and not even connected to anything electrically. But the machine did indeed contain 72K of ram, so it was able to be imported 🤣
I would be interested in writing Commodores 8050 1080kb disks. Quad density GCR. I think the Amiga uses this format as well. I have a bunch of 1mb Amiga 5,25" disks...
Hi Adrian! I can confirm what you said on the video: You can connect a 3" CF2 drive like the ones used on the Amstrad CPC, PCW or ZX Spectrum +3 to a PC and dump the disk contents with a tool called CPCdiskXP and you can also "upgrade" those machines to use a regular 3.5" drive (or even a Gotek)
When you upgrade to a 3.5" drive, does it physically fit inside the machine? That would probably be the upgrade I would do simply because 3.5" disks are ubiquitous and easy to find. I will probably never end up with a CPC with a disk drive though. Rare as hen's teeth in the US. :-)
@@adriansdigitalbasement you have to file out plastics but I used a 3,5" drive in a cpc. This is a MUST upgrade, fuck those 3 inch disks... Thry are probably all dead by now. I knew a guy who had a box of 3000 non working disks. What a sad sight. Just get a 3,5 drive, you need 2 switches. One for it being able to accept a 2nd drive. The other switch is to switch side a/b of the drive. Get cpcdskxp, tape over the hd hole on the disk and there you go. I think i used cpcdskxp with a usb floppy. Very easy way to get disks to a cpc with no expensive gotek. Also you can get more data on a 3,5 disk because the floppies are not shit 3 inch disks. Something like 280kb per side is more realistic on a good 3.5 disk.
Addendum: you need the drive a/b switch, to boot from an external 3 inch drive if you really need to run a game for testing. Most games just assume you only have drive a. Also the command |cpm only runs off physical drive a:. This is a boot command alot of demos use. The wiring is simple and you can find it on the internet. Most drives work that you can change to ds0.
@@AmstradExin I paid £50 for a gotek already mounted in a 3D mount for the Spectrum +3 last year (2021) on eBay so it's not that expensive - gives a far better look to it without cutting up a vintage machine to fit a 3.5" drive into it.
The Greaseweazle sounds absolutely amazing, glad you are getting your money's worth! If only I wasn't stuck with an Iomega ZIP drive shot down by the Click of Death holding my Amiga backups, including some seriously vintage porn. :( Not even the Weazle can save me here.
I remember we always used TEAC drives back in the day in any PC our family set up. I actually had a combo drive in one of my towers, it was both 3.5" and 5.25" in a single half-height bay, I thought it was just amazing. I dont think I've ever seen another one ever since either.
I used 5.25 and 3.5 disks and drives starting in the late 80s and 90s but I have never really understood the details behind the drives and disks. This was so informative and cool to watch. Thanks Adrian.
Wow that was quite a deep dive into the world of FM and MFM floppies! I’d love a part 2 going into GCR floppies and perhaps also the variable speed 3.25” floppies. I recall Commodore uses fewer sectors on the inner tracks to maintain a more uniform data density on the magnetic surface, while Woz used different RPM for some tracks to achieve the same effect. Would love to learn more about this stuff!
You could store more data on the larger tracks by one of two means. You can keep the same data rate, but spin the disk more slowly for the outer tracks, or you can keep the same spin rate, but increase the data rate for the outer tracks. The early Macs did the former, while later computers did the latter.
Well this is nice and timely, I was putting off on building a new floppy extractor for my library of NABU diskettes. I've bookmarked this to review when I take another crack at them with IMD next week.
I cannot believe that I watched all of the video, but I am sure that this will become a REFERENCE VIDEO that many retro enthusists will consult as a vital tutorial. WEll laid out and considered. Thankyou for your hard work!
This video really does boil down the different drive and disk formats that we retro enthusiasts have to deal with, and all in the one video. Whenever I need to explain this to someone in the future I won't need to fumble around with my own words trying to explain, but rather just point them to your video here. Thanks Adrian! Love your channel, one of my favourites BTW!
I'm so happy you're on here, making such an incredible effort to document old tech, and especially highly detailed vids like this. We cannot let all of this important history become lost. Thank you a million times over for what you do, your knowledge, effort, and these videos!
Thank you for a VERY INFORMATIVE video Adrian! I have watched it twice and will be referencing it in the future as I work with 5¼" disk drives. I also find the sections on the IMD and HxC software to be very helpful. Thank you again and please keep up the great work.
I'm only about 16 minutes into your video, and you've covered so many details of the various formats that I remember having to fight through! The Tandy 2000 with 720K quad-density 5.25-inch drives, the oddball HP 150. I don't blame you for leaving out the hard-sectored floppies and things like the DEC RX-50 floppies. Back to watching the rest!
One modem correction: 33k was reached by just using improved encodings, but 56k actually _did_ require phone system upgrades. In fact, part of the initial modem negotiation was talking with the _phone system_ in order to switch to that improved hardware (so that the phone company could save some money).
It required ISDN technology at the oposite end of the link. 56K was not available from person-to-person but rather person-to-service. Thats why you ended up with 33K one way and 56K the other.
Absolutely fantastic video! I deeply appreciate your attention to detail, along with how you distill the most important parts into easily understandable chunks, while touching on the edge cases so we know where to start investigating if they're relevant to our situation. Thanks as always for the love and care you put into the videos you make. I watch pretty much all of them, and yours is easily my favorite retro computing and electronics troubleshooting resource on RU-vid!
The fundamental difference between the media of 48TPI and 96TPI drives is the oxide used. 48tpi disks use ferric oxide, and 96tpi disks use cobalt oxide, which has a much higher coercivity (600 oersteds, vs 300 oersteds.) This means that the magnetic head write current must be twice as much, in order to induce a flux transition.
I think you're talking about high density 96TPI media, not double density 96TPI media. I've always found that I can use regular 48TPI disks as 96TPI double density disks without issue... but using a high density 5.25" disk as double density doesn't work (or vice versa.)
Today I repaired my first VIC 1541 floppy drive and FIRST bad chip in my bad chip box. Super stoked! Had a bad DOS ROM. Thank you for helping us out! Now to play Jump Man.
This is a fantastic & educational video. Mandatory viewing for any & all retro-computing enthusiasts. I thought I new a lot about "classic computer" disk drives and such, but I learned so much and amazed with the new tools that are available. Another great ADB video! And wishes for a wonderful new year, Mr. Black.
Fantastic video, Adrian. I knew some of these facts, but I didn't know the details. It is really awesome that floppies, in essence, haven't changed since the Shugart SA400 drive.
Once again, Adrian, you never cease to amaze with your videos! Awesome!!!! I've been a user of Dave Dunfield's ImageDisk for 7-8 years now...so it's perfect to see you using this tool in a dedicated workstation like this.
I don't know why but I saw the title and thumbnail and thought you were talking about making the physical disks themselves and not just writing data onto disks due to the fact that nobody manufactures them anymore. That would have been a really fascinating video.
When imaging a disk in the section where you added the comment I recommend entering a description of the disk's format. Sides, Sectors per Track, FM or MFM, Data rate etc. Others who have done this have saved me a lot of time when I create disks from their images.
I will probably never use the knowledge in this video, but I'm still watching it. Guess that's a testament to the production quality... who doesn't love mesmerizing bouncing balls over 'fancy drawings', though for me it's the always present partitioning. Not only does it allow quick access if one wants to skip to a particular segment, it always guarantees the video follows an enjoyable progressive flow. Since I only ever sometimes play with emulators, disk images is all I need. But for that, I need people like you, maybe, who will use the knowledge in this video to be able to upload those images for the plebs like me to download and use for our leisure. I would like to say: thank you, both to Mr. Black and his community for taking care of the larger community, the trickle down effect is particularly visible ( and enjoyable ) here.
The 'width' aspect of writing to 40 vs 80 track drives was only true for a few very early PC drives. In general both types of drives used the same heads, and many of the non-IBM early 5.25 drives were all 80 track devices with BIOS coded double-stepping to reduce manufacturing costs, meaning they could fleece us as consumers by charging more for 80 track devices at no extra cost to themselves :) When I was a BBC micro games developer I used the Watford Electronics DFS ROM which had commands *40 and *80 to switch modes and do double-stepping on an 80 track drive, but I created a modified EPROM version that read the track count from the outermost track when the directory loaded and auto-switched, so I could insert any disk and have it auto track-select. Fun days! These videos are always an interesting trip down memory lane, but they really make me feel old!
I personally haven't dealt with any of this in years this is good refresher. I have an old at server that has all these drives in it but that hasn't been running years either that goes back to my it days 😂😂
Don't consider myself the sharpest tool in the shed. Yet the way you explained the disks made perfect sense. Just goes to show it's true that the teacher makes a major difference to the students ability to learn, even in my 40's. :) . Thanks Adrian.
"You set both to drive select 1" ✌ "Connect drives 1 and 2 to this port and drives 2 and 3 to this port." 😁 Making both the zero-index and one-index people happy I see.
1:45 The ZoomFloppy's driver suite (OpenCBM) actually just uses the generic emulator format (D64). In fact, the device can only read/write "raw/stream" (NIB) files with a 1571 or a modded 1541 with a parallel port, because a stock 1541 isn't capable of doing it -- and you need to download additional software (NIBTOOLS). So at least in the Commodore 8-bit world, the ZoomFloppy should have no problems writing out any old D64 that you download from the Internet. This is a fantastic video you've put together for MFM disks, and other systems! -- JC
As one author of the floppy user guide I congratulate to the great presentation of this topic! I never heard about the 100 tpi format before, amazing. :) There is one thing you should mention for anybody who does not use this specific floppy controller: Various modern floppy disk controllers do not work with FM. The older, the more likely they are to work.
Thank you for this. This is heavily relevant to the next project, getting as many disk formats on one Pentium 4 machine as possible using a board from Tindie as a way to archive all the 5.25 shareware disks I have.
This video was so satisfyingly useful... The weirdest thing is that I don't own any vintage hardware with which I could actually make practical use of all that's been explained, but still feel it's very helpful. Am I out of my head?
Adrian, Also don't forget there were 5.25" ED 2.4MB floppy drives as well. There were also 3.5" ED 2.88MB drives too. The 5.25" ED drives I could only find in IBM Workstations. Great video as always sir! :D
I want to call attention to a typo in the overlay text at ~20:48 where Adrian lists various 3.5" drives - the Commodore drive is the 1581, not 1851... oh, and I LOVE the use of the Boing Ball demo for the chapter sections. :)
Excellent 5.25" drive analysis Adrian, i started my entry into the world of disk drives started with the ZX Sprctrum +3 3" discs that i thought were very robust that i still have today working for both Spectrum & Amstrad CPC 128 8 bit computers, i then moved onto Atari STe with the double sided 3.5" , then in 1993 i built my very first IBM compatible PC- Intel 486 DX2 66Mhz, 4MB 30pin ram ( 2MB I removed out my 4MB STe , 1/ because it was an overkill amount on ST just to play games and 2/ save money for my new PC build as ram was quite expensive back then) 1MB VLB Trident video card, 80MB 3.5" HDD ( im sure it was a Conner branded drive) Sound Blaster Pro2, but by this time in 93 every software was being released on 3.5" media ( all the games and all the magazine cover discs in UK ) so i never owned a 5.25" drive or disks and never had the pleasure of using said drives n disks. About the speed of the drives, so a disk written at 250k can't be read back at twice the speed , is that correct?, I'm glad that's not the case with 3.5" drives, I have a USB SMARTDISK FDUSB-TM2 floppy drive that reads back at 2X speed, Also a Sony Mavica MVC FD200 Digital Camera USB that can read back at a blazing 4X speed that is an insane floppy transfer rate . I don't know if you've seen it but "CATHODE RAY DUDE " has a great video on drives that can read faster than 1X , it's tilted " READING FLOPPY DISKS ? GOTTA GO FASTER" , Check it out Adrian if you've never seen it, and once again many thanks for all your hard research and excellent video, have a great day buddy.!!!!!!
In the past year, I put together a bare-metal DOS machine for old games. I went through a lot of HD floppy drives to get a pair that actually worked. For my personal collection, my drives are a 3-inch and a 5 inch. A have several containers of old floppies from my youth, and thankfully a vast majority of them have been completely readable. I haven't been reckless though, before I actually use the old installers on the disk, I use a DOS program called FIRM, to rip an image to a backup directory on my drive C. FIRM doesn't have the options of ImageDisk, but my needs are simple and are limited to DOS disks. Whenever I have a disk image that wasn't from a physical disk, I can skip the step of physical floppies with a DOS program called TurboImage, which simulates a drive and mounts it on a letter. I don't really have a pressing need to write the DD 5-inch disks, so I'm basically happy simply being able to read and rip them in the HD drive.
Adrian's explanation regarding Amstrad 3" drives at 19:28 is absolutely correct: 3.5" drives can, for example, be plugged into the back of an Amstrad CPC6128 with a slightly modified floppy ribbon cable with the 5.25" / shugart edge connector.
Great Video! When DiskImage reports a deleted sector, it probably refers to the sector data mark in the IBM track layout indicating that a sector has been deleted. Normally this data mark is 0xFB, but for deleted sectors is 0xF8. Presumably this allows to quickly delete (and restore) sectors or files on a floppy disk if the OS or software supports it.
I am impressed with your knowledge of the different drive systems and configurations... it certainly is complex, I am not a retro computer guy but watch your videos as I find it interesting non the less.
Fibre optics has had the same kinds of leaps in modulation technology as modems back in the day, by discovering techniques for multiplexing light in different frequencies. I think it's interesting that bandwidth is basically doubled with each leap in all areas. Like DDR RAM - DDR 4 has twice the bandwidth of DDR 3 and DDR 5 is twice that again. Same with PCI Express, etc etc etc. Maths and physics, man, it's incredible.
thank you for making this Video very informative , I just want too Add that the Commodore 1571 Disk Drive which ( for those that do not know ) was created specifically for the C128 for the Use of its CP/M Compatibility . The Disk Drive can Run all formats mentioned here in this Video : MFM(double) FM(single) and of Course GCR.. we had a 1571 with my C128 when my Parents bought a Commodore C128 Computer back in early 1986 I had used a Program called BIG BLUE READER to Read IBM / and Kaypro Disks and Transfer TXT Files from one IBM to Commodore "vice versa " using the IBM Compatibles in my High School and bringing the IBM disks home to transfer my school work on a Commodore formatted disks to use in my C128 , The 1571 I.M.O. , is most versatile Disk Drive of the 1980s reading IBM CP/M and Commodore formats . not sure if there was any other versatile multiple format reading 5.25 Disk Drives like the 1571 ..
An amazing video once again. I think that a lot of folks may not have been aware of the differences that can exist even among the physical components of the drives shown. I've had my own adventures with putting a machine together only to find I need a software tool or different type of drive to get the data I wanted to use. A+
Thanks Adrian. Terrific video I have been using computers all types since the 1970s. This was like a trip down memory lane. I had forgotten that all these things existed. Enjoyed it very much.
Absolutely super video on disk formats. I wish you would have spent a little time on the Amiga Disk Format. Many years ago, I was able to adapt a couple of Dell double density drives to work as external drives. The information was published on one of the Amiga user group sites (no longer in existence) I still have the drives, and they both still work. Within the last year, I saw information online, that would allow a person to modify some specific IBM type 3 1/2 inch drives for the Amiga. I have found some of the right type of drives, now I will have to find the information to modify them to work.
For us Amstrad users, 3 inch drives are mostly single sided and we flip the discs physically to use both sides. When we use 3.5" or 5.25" drives we either upgrade our ROMs to cater for 80 track drives or... put a side switch on the drive so we can use them as 2 single sides of discs. You can use your switches as a side switch so you can treat a disc with 2 single sides as such but also treat a double sided disc format as such. Some drives we can also put a 40 track/80 track switch, but a bit less common.
Hi Adrian. Amazing video and highly instructive. Another one of your videos to keep track and use as documentation while recovering some of my FDDs. Thank you so much for this large amount of information !!!!
Just wanted to add another shout-out for John Morris' AppleSauce. Not only is the hardware really, really good, but John is also excellent at creating UI. Mac only, but beautiful!
Most excellent Adrian. 👍 I wish I had this video 20+ years ago when I first started figuring out disk drives. 😅 This will no doubt be useful as we go through and attempt recovery of our old floppies. Thanks!
When I got into this hobby and bought my Tandy 1000 SX, I had no 5.25" floppies of any kind, and no way to create system disks, so I literally had to buy some OEM sets of MS-DOS off of eBay, just so I had something to get started with. I bought them, made copies of them, (on the 1000 SX because it had has two drives,) and sold the OEM disks back on eBay! Then I was able to buy and XT-IDE CF Reader and was able to format it.
Great Video Adrian !. The Amstrad CPC 664, 6128 and 464 "with the DDI-1 Interface" used the NEC765 floppy disc controller. This enabled you to choose FM or MFM recording modes. These computers used the 3" CF2 Compact Floppy Disc, made by AMSOFT, MAXELL & Schneider. The drive hardware was made by Hitachi, Matsushita, Maxell, Teac & Amstrad.
If you work with Macintosh disks, there's also the AppleSauce Floppy Disk Controller. It interfaces directly with a variable speed Macintosh floppy drive, enabling you to easily read and write Macintosh floppy disks. In addition to Macintosh floppy drives, it will also interface with Apple II and standard PC floppy drives, making it the only headache-free archival option for macOS for working with nearly any format.
Excellent video. Lots of important information that’s needed that will handle most disks out there. I didn’t even know about 100 TPI drives before today. I would love one of those Texelec controllers once they come out. I just need to figure out how to store disks so they don’t degrade in the south Florida humidity. A couple of points some might be interested in. With the 5.25 HD 96 TPI disks, they are commonly used in some of the Japanese computers that are not PC compatible, like the NEC PC-98 (which does run MS-DOS) or the Sharp X68000. Another quirk of Japanese computers is the mode 3 3.5 disks, which are 3.5 HD floppies written at 360 RPM with only 1.2 MB of formatted storage. Again the PC-98 or X68000 3.5 inch drives use this, as does the FM Towns, which only supports 3.5 inch disks along with the built in CD-ROM. Not sure why the Japanese computers have such a strange format, but they do. These disks can be written on a PC with a special 3 mode 3.5 inch drive, and BIOS support. If you ever wondered what did the BIOS option for 3 mode floppy was, that was it.
Such a great and timely video! I'm working on an old Monroe OC8820 and need to create some disks for it from td0 files. This was a great help! This particular computer uses Micropolis single sided quad density drives. Luckily they are 96tpi.
Another thing to consider is whether the disk uses soft or hard sectors (given by the number of holes in the physical media) that pass under the hole in the sleeve (for 5-1/4” disks) where they’re picked up by an optical sensor on the drive. One type has only one hole, so if the disk spins at 300 RPM, the sensor will send 5 pulses per second (300 RPM / 60 sec/min). The other kind has multiple holes in the media, so the sensor will send way more pulses per second, for the same spindle speed (300 RPM.) 🤔
I was in the process of thinning out my stack of 3,5" and 5,25" drives and was looking for some guide on what to leave and what to recycle. Thanks for directions, and Happy New Year!
Wow, lots of good info! Adafruit managed to do a GreaseWeazle style thing on an Apple floppy drive about a year ago, which is an interesting wrinkle. Picked up a GreaseWeazle a few months ago, and subsequently realized I had no 5.25 inch PC drives... Hoping there's still one hidden in my parents' basement, I used to have so many, and a Pentium II with one installed in it for floppy transfer purposes....
Big hurdle on the GreaseWeazle is the software. So many issues, crappy OS support (No proper OSX version), various hang-ups, impossible GUI's with tons of usability issues. So it might feel cheap and handy, it will cost you a few K in lost working hours.
Brought back memories of a misspent youth, part of my first job in '87 was to write MC6809 assembler code to convert all kinds of weird floppies back and forth. One physical property of the floppy that wasn't mentioned is the different magnetic coatings for DD/HD. This goes for both 3,5" and 5,25", some drives can be a bit sensitive about this. Also, if you're working with NEC PC98 stuff, most of their 5,25" and 3,5" drives run at 360 rpm. Early 90's I designed equipment that had to read/write PC98 disks and 3,5" drives running at 360 rpm was tricky to get hold of outside Japan. The PC98 drives followed the original formatting schemes proposed by IBM more closely than the drives in the IBM PC world ended up doing, IBM jumped through a few hoops to cram in a little extra data.
I have had my "play" with a few floppy drives. In fact, I still have one loose 8" drive, brand Siemens and one installed in a Bering har disk / floppy combo. I never have had the Siemens installed and running. The Bering unit floppy has 1.2 MB capacity and 5 MB hard disk, which I partitioned to 4 1.2 MB size units to support backing up each partition to a matching floppy. Oh yes, the system was used on a HP-87 computer, which also comprised of a dual 5.12" floppies, each track consisting 16 sectors of 256 bytes for a total disk capacity of 286,720 bytes. Organized on two sides, 35 tracks per side. Then I have 800 kB, 5.25" floppy disk(s) used on a ZCPR3 operating system by Richard Conn. Those were the days of dial-in bulletin boards. I also have blank disks, as well as full of data ones. 5.25 floppy drives in double and HD, and 3.5" unused Teac drives. I forget if they are HD or just double side. I used to buy my floppy disks in 100 quantities at a low price. Actually my first floppy disks were for playing after hours on TRS 80 computer the company bought for inventory control purposes. I got for it a Miller MicroSystems FORTH operating system that allowed me to study the ROM Basic. Sort of elementary reverse engineering. But more important, I wanted to gain and keep some sort of understanding of the FORTH language that our process controllers started using. And that brings me to two more 8" drives, part of the development system that was used for the first generation of our controllers. I eventually purchased that development system, figuring I might need to support the control systems for some time after our vendor gave up. Never actually needed, but as a result I still have the two included 8" drives. Thanks for educating and fun material!
This is such a great video! one basic, very handy way around the floppy issue is using serial and xtide and booting from "floppy" as an image streamed across serial from a host pc or copying files off the image. (obvs PC only!)
There was a meme phrase back in the day: "On a clear disk, you can seek forever." This is talking about an unformatted disk. And by "unformatted", I'm talking about the low-level formatting, which many people are not familiar with. On a floppy disk, you don't need to worry about finding the tracks, since that is done by a stepper motor. You just push the head all the way to the track 0 stop, and then you seek to a given track by stepping the head forward the proper number of steps, and the hardware is calibrated to find the right track. But seeking to the right sector is a different matter. If you think about it, when the disk is spinning and the computer is reading it, all it sees is a stream of 1's and 0's. It needs to be able to figure out where is the start of a sector just by looking at the bit stream that it is reading. It does this by writing a special bit pattern that is not used for normal data, but instead marks the start of a sector. After this is written the ID of the sector (and perhaps the ID of the track as well), and after this is written the actual data for the sector. In addition, each sector needs to have a certain amount of "guard" space on either side of it. The purpose of this is to allow drives that are slightly slower to write over the sector without "bumping" into the next sector. On the Amiga, they did away with the guard space between sectors by always reading and writing entire tracks at a time (which is why they hold more). You can think of that as having just 1 big physical sector per track, and it still needs some guard space at the end of it. I can also write a bit about FM, MFM, and GCR, which are just different ways of keeping the bits properly timed. When writing a bit stream, you need a way for the system reading it to keep track of the start of each bit even when it might be operating at a slightly different speed from the system that wrote it. There are different ways of looking at this. One way is to imagine that you're writing a clock signal and a data signal combined together in some way. The other is to imagine a scheme that guarantees you won't have too long a sequence of just 1's or just 0's, since after some point you'll wonder "was that 7 zeros I just saw, or was it 8?" Either way, you need to guarantee a certain number of transitions per time, while still figuring out what the actual data is. FM uses an extremely simple scheme that almost interleaves the clock and data bits (it's just slightly more complicated than that). It allows a maximum of 2 zeros in a row. MFM allows a maximum of 3 zeros in a row. GCR will encode a certain number of data bits (say 6) into a larger group of bits (say 8) by only choosing the bit patterns in the larger set that are "legal" to write out (ie, using them won't write out too many 1's or 0's in a row). A special GCR value also needs to be chosen to indicate where the group boundaries are. This can also be used as the sync mark to indicate the start of a sector. As an aside, copy protection on the Apple II typically worked by writing out "illegal" patterns that were possible to write but not read (normally). This would make the normal DOS think it was reading bad data, but a custom DOS would expect to see it; also, normal DOS would never write those patterns out. As well, there were certain different patterns that you could write out that would all be read in as the same pattern, but the timing would be slightly different, so code could be written to verify the expected timing. Again, normal DOS would write a pattern that resulted in a different timing. There were many things you could do when you had such low-level control of the disk hardware. So getting back to that meme, an unformatted disk will just look like a bunch of 0's (or maybe 1's), and a sync mark indicating the start of a sector will never be found.
Floppy drives are pretty low speed, so termination is probably not critical, but if were, you wouldn't want one of the cables to be unterminated. You always want the end of the cable to be terminated. Usually that means you put a single drive at the end of the cable, and it provides termination, or if you use two drives, the one in the middle of the cable is unterminated. And in either case, the controller is terminated. However, in this case the controller is effectively in the middle of the cable, and you want the cable terminated at both ends, ie the further away drive on each cable, and the controller would be unterminated. I don't know if the card has provisions for that, and again it probably barely matters at the speed floppies run at
Oh that brings back memories now you remind me! (Well- almost PTSD and I may not be remembering exactly, but certainly triggered some memories of trying to work around this issue) : I used to build custom machines commercially for people and sometimes needing to run 2 cables to 2 floppies with a single controller to make the layout of the case and cable runs work, and needing to be able to “unterminate” the controller that was effectively in the middle was one of those “will this design work” moments as not all controllers supported doing that - and like you say, you could kinda get away with it if you had a following wind and luck, but IIRC you might get slower performance due to disk read errors and the OS needing to retry etc - and I used to run benchmarks on the systems to check for issues like that before delivering them to customers. On a similar note - I cringed seeing the mess in his PC build! I used to spend ages planning cable runs and drive layouts and case options so everything would be neat inside - making a quality build for the customers so it would be easy to work on later, but also to ensure optimum air flow for cooling etc, even considering things like the gaps in the case where it would draw in dust over time to cause issues for floppy drives eventually (as I also had to fix people’s old computers and learned how badly put together cases and internal layout of components would eventually cause issues when a machine had been stuck under a desk acting like a vacuum cleaner for years). Anyhow / thanks for the reminder of that detail :)
Love the setup, being a network tech I personally would have added in DOS file sharing support instead of transferring data via floppy disk, but love the build and use of this system
Great video Adrian. I worked at both Qume (think daisy wheel printers, but they also made the first half height 5 1/4 floppy) and Micropolis, the inventor of the 96/100 TPI drives. Yes, they also made floppies and then went into hard drives afterwards. You failed to mention hard versus soft sectored floppies. I only remember Vector Graphics who used hard sectored floppies, but I think there were others too. I usually save the drives from old computers when I scrap them, so I have a few in my collection. Best regards for a Happy New Year and lots more videos. Richard Ravich
One time the guy at the local TRS-80 club had made a bulk buy of floppies, but he accidentally got shipped a case of hard-sectored floppies. I put one of those disks in and started format, and it went rat-a-tat through the whole disk in seconds! But hard vs soft sector is determined by the disk controller, not the drive. The drive just passes the index pulses on to the FDC.