Im so glad you mentioned the applesauce, its a wonderful project that works very well and for many MANY systems. basically if someone joins the discord and asks about a format that we don't already have, within a night someone would have reversed engineered the disk and a beta version of the program will be out in the morning that can parse it. It's quite a great bunch of people who run the project.
As I recall, IPL stands for Initial Program Load, basically the initial bootstrap code that loads in more code to then actually start the machine. I used to enter front panel opcodes to boot various minicomputers like PDPs, HP 1000, Interdata / Perkin-Elmer, and early microprocessor based systems. Love the content, keep up the good work!
For the Control Data 6000 series it was called deadstarting and a switch panel was read to load instructions into a peripheral processor to read the system software from mag tape or disk. As you say, the PDPs used front panel keys to load the instructions.
I just remembered, while at work running a group of automated test stations based on the PDP-11, one of the support engineers had made a test fixture adapter with an EPROM in a socket. We only needed to enter a few bytes of data that told it to run the code in the EPROM, that performed the majority of the bootstrapping. Of course, these PDPs had magnetic core memory, non-volatile, so we didn't have to do it very often. But when we did, it beat the heck out of entering lots of code manually!
@@AureliusR Using an official term like IPL is more accurate, while "bootstrapping" is more casual, taken from the old saying "lifting yourself up by your own bootstraps". As a slang term it gradually was shortened to just "boot" or "booting", as computers became friendlier and commonplace. Kind of like how the name of a graphics editing program became an everyday verb: Photoshop, as in photoshopped, or "to photoshop an image".
Isn't it cool Eric's grand-dad worked for IBM on floppy-disk technology? Man those skilz run deep. And I cannot wait to see the 8" Floppy Adventures yet to come, no doubt you will get them to work eventually (as you say).
I find it tremendously fascinating to watch you guys reverse engineer the tech I used too back in the days, without knowing anything about all those different designs. Thanks Marc and wizard Eric!
I was just about to make the same comment… 😅 Primarily I used two 8” drives on SSLs very proprietary series G computer, controlling their (more) famous 4000 series audio mixing console. Although I did (at times do) pure IT tech and management too, the SSL computer was it’s own beast - we just never messed with it. There was this rumour that it was export prohibited under the Cocom rules because of the wast number of (I think) 10-bit ADCs for fader position sensing. Around 1994 our 8” drives were replaced with “modern” 3,5s, with a VERY expensive daughter board to make them compatible… Great work! 👍
I find it peculiar that there's such a massive difference between the Minnow disks and the ones that eventually became the commercial product. Typically all the storage media was a gradual evolution of the previous design, but this is a completely different thing that just happens to have the same shape
This brings back memories from the mid 1980s of my old single board CP/M system using a Z80 processor and 64K memory. It had a single 5.25" floppy built in and an external 19" rack mounted box with 2 VERY noisy 8" floppy drives mounted side by side. The floppies had a capacity of 1.2MB if I remember correctly. I mainly used the 8" floppies to archive all the shareware programs I downloaded via a 300bps modem from a local Bulletin Board System. The downloads took ages. Fun, Fun Days indeed.
Wow, so awesome to see these weird old 8" formats, the 70s and 80s was a golden era of computer tech. I actually have a disk here i would love to see you take a look at, it's a standard 8" DD soft sector disk but it doesn't use sectors, one track is one whole sector! In fact the system has no concept of sectors. It's from a little known machine that Alvy Ray Smith described as a 'beautiful realization of a paint program'.
Marc - I found and dug out my IBM 3600 controller disks and, unfortunately, regular soft-sector. Sorry. It must have been on my ancient Zenith Data Systems system that is buried in a storage locker next to a C= Amiga 2000 and a bunch of AT&T Unix PC 7300s. I can't remember the model number of the Zenith - I haven't looked at it in 35+ years. I do remember that the hard sectors were on the inner rather than outer. The disk looked exactly like a regular 8" - but with more holes that would show when you rotated it manually. As always, great job to you and Eric for keeping the electronic archeology alive.
Lol, I thought this was an Usagi Electric video from the thumbnail, and when Marc started his voiceover I had the most incredible experience of cognitive dissonance. :D
The company I worked for in the 80's had a 370/158 that used 8" floppies to load the micro-code. The drive was under the console CRT. The system also had two built in Integrated Storage Control for 3330 disk drives that also used 8" floppies. It took a while to get the system IPL'ed from a cold power on. I hadn't thought about that part of the system in some time.
Vydek word processors used those 32 sector disks @ 2 min mark. I remember the missing corner to prevent using the "wrong way". I think the hard sector holes were on the outer edge, but the envelope had a corner notched to prevent inserting the wrong way. I still have 8 inch floppy accessories: binders, loose-leaf sleeves, smoked plastic locking cases and D-I-Y reinforcing hub rings! And labels! Lotsa labels!
Vydec was an interesting WP company. Their engineers did everything possible trick to create an incompatible disc format and so "lock-in" their customer base to prevent migration to other brands. They even rotated the media "backwards". They were horrified when they learned that a company from Australia had created a system that could read Vydec word processor discs, since Vydec never sold, but rented their WP systems...
Ahhh, the old days - when floppy disks were actually floppy, and there were more formats in heaven and earth than were dreamt of in my technical library! My first floppy drive was a Memorex 651 of uncertain provenance, back when the disks for it were still available. I built a controller for it based on something described, I believe, in a BYTE article - based on an SDLC controller chip with a strange Signetics one-shot for data/clock separation. I should dig that out of the archives and refresh my memory. I didn't realize the Intellec floppies used a nonstandard format. My memories of those involve (1) the head loading and unloading for each sector, and (2) the drives needing to be aligned to magnetic north for reliable operation. Cool stuff as always!
CuriousMarc You forgot the 32 Hard Sectors 8" FFDs. Like the "3M 740-32 SS,SD" Single Side, Single Density 8 Inch. Those have 32 index holes at the "normal"place as the SS diskettes. Use f.i. the Shurgart SA801 drive. Its just a SA800 drive with special settings to read those 32 Hard sectors. Great thing is, you can convert that drive to behave like the SA800 to read the 1 index disks. Thanks for showing those special format 8"diskettes. You see lots of people/companies try to figure out a system of their own. Like the "DATAPOINT 1800" format 8". a like M2FM or MMFM format , same like DEC RX02. Never able to read those disks back on a PC due to mix of FM and MFM that a PC can not handel.
Sound and HD are working now, it's good to be a bit, or several bits (stupid IT joke🤦😁) late sometimes😊 I had a double 8 inch disk drive next to my Z80 based computer rack. The system used 100x160mm PCBs with one CPU card, one memory card, one disk controller card, etc. When you started the disk drive the ceiling lights flickered and the heavy drive almost jumped on the table. The "Eurocard" (I think that was what the size was called, we had our own product called Metric Card) rack with the actual computer looked tiny next to the massive double drive😂. I worked as pre and post tech support for industries like Volvo, Electrolux and others where the Z80 racks were controlling robots in car manufacturing and such. I remember the joy when I got a colour screen and a colour graphics card. The base system was just connected to a serial terminal. The amazement when I could create graphics on pressure, angles, etc was almost overwhelming 😁 Ahh, the 80s 😁
Love the vids and all of you guys as well, absolutely amazing work that yall do, i think its especially cool that you guys actually use the old stuff you fix to repair and fix other old stuff. Bravo 👏 and thank you especially for all the awesome ad free entertainment you guys provide folks like me. Much appreciation ✌️
Original floppy had no sleeve on it. I saw one in Helsinki Industrial Fair 1966. It was on a dictaphone, but I bet someone used it for data storage rightaway
I always remember a story about the 8" floppy that an administrator called up the tech support about the system could not boot up. Then he was told to reboot it again, so he took the floppy out from his shirt pocket and unfolded it. 🙂
Yeah. This brings back memories. Even with 5 1/4 floopies it wasn't simple as the number of tracks and/or sectors could vary. This is why it was an extremely good idea to write those formatting numbers onto the floppy label Single sided / double sided wasn't that big of an issue. If the floppy was capable but the drive didn't have a 2nd photoelectric for dealing with the mirrored hole in the sleeve, .... cautious punching a second set of holes into the sleeve "doubled" the capacity by enabeling to turn around the floppy :-)
We had several machines on the factory floor being controlled by these 8 inch floppy diskettes. It was a big task to replace these disk programs with a Modicon PLC. 😎 Thanks.
Well, I downloaded the python code but my only contribution so far is noticing that there is a (kind of) pattern: quite a lot (but not all) of the hex consists of 4-byte chunks beginning with 0xA - so I just printed a newline before each of those keeping the hex on the same line (instead of 1 byte per line). And track 0 seems to have duplicated sectors - maybe for redundancy. Haven't looked at the other tracks yet. I did work with a 370/135 in the dim past, but I never saw any microcode docs - secret IBM stuff probably.
It is like stepping into another world, turning backwords floppies, halfsteps and awkward places index holes. What's next? Some tracks with different sector sizes?
The Amiga floppies crammed 880K onto what everyone else would consider an 800K disc by rotating the disc slower for outer tracks and faster for inner tracks, a bit like CDs do. You generally need the special Amiga drives to read them, unless you can control the rotation speed yourself.
Had no idea they used a floppy for IPLing - guess it's more durable than paper tape. And faster than a bank of switches. For someone who works on modern mainframes this is wild as hell - these days you IPL from a separate machine mounted in the rack above the processor drawers.
I love how the old hdd sounds.... next: old hdd!... and who knows, perhaps we can find something interesting in old data disk from a university or militar industry!
Ooo. I have a bunch of hard sector 8" disks, but none of them have the sector holes on the edge of the disk like that. Rather, they are arround the hub in the same place as the soft sector disk index hole. Alows you to use a hard sector controller with a standard sughart drive
When I was a field engineer in the late '80's I was looking after CPM based micro's in the UK. They were using a custom format on standard 5.25 drives to cram over a MB on a 720kB disk. I don't know how it was done, but alignment was critiical. We carried a scope and a calibration disk to align the drives on the customer sites. Even the calibration disks were cross checked with each other as the tolerances weren't good enough to guarantee alignment. It was a real pain and scopes were heavy in those days!
I remeber those. They where placed on a chair (cooling issues) behind the factory control computer.. Sometimes a reboot was necessary :) And they were very floppy
For MFM see if you can find a WD 1791 chip and documentation. For FM get the WD 1771. With the 1771 you can work with any (reasonable) sector size that is a multiple of 16. I used that feature to squeeze a very reliable extra amount of data off my SSSD 8" floppies. {^_^}
My approach to floppy disk archiving has been snapping up the bitstream with a logic analyzer, saving in SuperCard Pro file format, then having a look at it with some tool like HxC. HxC doesn't natively take the more exotic sector-formats, but the data visualizer is enough for me to see if things look fine or not. It can be time-consuming, but on the flip side I go around a lot of issues which can become a roadblock for more conventional methods.
It feels great to be among nerds here in the comments. One thing that was easier in those days than they are now: you easily remembered to make a backup, because you never knew if your data would come off the disk.
There's also the interesting 3" floppies beloved of Amstrad/Schneider on their PCW and the +3 ZX Spectrum. I've always wondered if you could read on on a PC with appropriate adapters.
I have one of those AppleSauce controllers on a modified Disk ][ drive I used to transfer data to/from my Macs to Apple // floppies. The controllers are made in batches and not often in stock. It'll be a waiting game.
I went to process the data from CSVs - using only the stream of reading head (since index holes are intended to assist writing or disc jam). Made an edge detector and outputted a stream of nanosecond peak to peak times. Divide by 250 as the acquisition gear seems to use 4 MS/s (which is plenty of oversample even for higher speed rotation), my zero bit times are about 32 T which correspond to 8us, which is correct-ish if the drive spin at 360 rpm instead of 90 rpm. After bit decoding of >24T being 0, otherwise 1 + ignore next period, I am able to convert these peak to peak times to bitstream. Reversing it, then filtering out the sole 1 glitch at sector boundary near index marker, I can split the data into sectors. There are about 3 rotations for each track, and the data looks same for second/third read. I am able to directly see the incrementing CFDA - sector address on first byte, but from second byte and later bytes dont have a 1 start bit marker (when I split the sector to 10 bit chunks by length). Also the number of bits per sector varies, from about 2485 to 2505 with few outliers. Anybody to help out with the framing or decoding method?
So now looking for a start-bit, I can see how the bytes are delayed (maybe to take into account the timing of the machine?) I end up with same variable length sectors as the Python script. Definitely not a format described in Model 145 docs.
The IBM manual that documents the sector format and the microcode for the 370/135 is apparently SY33-0032, "3135 Processing Unit Theory and Maintenance Manual". Any old IBM guys here with a copy? ;-) The sector format of the data Eric dumped seems more similar to the 3830 than the 370/145, or at least interpreting the data as 3830 format, track 0 comes out with no parity errors. Unless that IBM manual shows up, I guess we're stuck.
About 1979 I saw at the Northeast Computer Show in Boston, a diskette drive by Innovex where the magnetic medium was affixed to a sheet of solid plastic, and the entire drive head assembly (!) rotated on a spindle underneath it. Never saw that anywhere else, I wonder if anyone has a clue what the model might have been?
On Eric's first breadboard setup, there was a very unusual square component with multiple leads plugged into the breadboard. It almost looked like the ceramic substrate from an IBM SLT can. Does anyone know what it was?
@@TubeTimeUS So cool! IBM's Solid Logic Technology has always held a special place in my heart since I was a young kid lucky enough to tour a 360 system. What an amazing leap in technology it was. At the time, I could hardly believe they could make and mount such small transistors -- and here we are today!
Would it be theoretically possible to make a scanner that can read any disk at, say, 1000tpi and lots of horizontal resolution, dump the digitized magnetic readings to a (huge, I know) file and then feed this magnetic scan to software that reconstructs the data? Or is there something specific about magnetic media that makes this impossible?
I had a hard sector with the index hole in the normal place.... I don't remember as well as I should but I'm sure it was an 8".... but it might have been a 5.25"
Hello. if you can move the head by micro steps you don't need the right speed or right head position, just scan the magnetic image, exactly like a photo scan but read with a (fine) magnetic head and a sufficient definition to have redundant information allowing to reconstitute the chain of 0 and I. then, apply a mask for read the track and sector you want. that is going to take time ? yes, but you are not using these floppies in production, the purpose is only to read the file once. when you have the zero position and step parameters, you can make a specific custom controller.
I have an IBM Minnow drive (23FD) complete with documentation and about 25 diskettes. I'll send you detailed pictures if you want. Just give me an email address. Read only, 64K bytes, single sided. It was the first production floppy. My disks have S/370 Model 168 firmware (12/16/1975). The controller for the 3330 disk drives also had a Minnow drive for its firmware. There was an earlier floppy prototype that looked more like a phonograph. The disk was backed with foam rubber, about an eighth of an inch thick. This prototype was on display in IBM Building 004 in Poughkeepsie NY for many years, but I don't know what happened to it. I've worked with many 33FDs, called Igar. They were read/write, and they were "soft sectored." I've never seen the one you show (31FD). Enjoy!
This is like modern day archaeology, preserving the information of the past before it is lost forever. In 20 years time you will be reverse engineering a CD-ROM drive!
I remember 8 inch floppy I also remember 10 inch and 12 inch floppy The 5 1/4 was a strange size I never understood why they came up with that. They could have done 5 inch or 6 inch Just strange they came up with 5 1/4
Yup, only 360p available - it's still processing the HD versions... come on youtube, you only have about an infinite number of cpus available - process it faster
You looked at it too soon, it's fine now. RU-vid initially puts out a very low resolution (360p) version of a video while they are sending the higher resolutions to all of their content farms across the world. You were unlucky enough to watch this before the hi-res versions showed up in your area. A lot of people are complaining "no sound". This is another side effect of sending out a low-res version first.