Old Memories 2: Magnetic Boogaloo!
The Dance of Magnets Moving To and Fro has Enabled Information Storage Since 1898.
This article is coming to you because of moving magnets. In spite of impressive advances with solid-state memory, the large storage arrays of “The Cloud” (including Substack’s cloud I would reasonably guess) still rely heavily on traditional hard disk drives, which operate by magnetizing very small areas of a thin coating on a rotating platter substrate. Those areas act as magnets themselves, creating a field that can be detected by a flying “head” that skims ever-so-close over the rotating platform these magnetic sources ride upon. Billions upon billions of tiny little magnets, spinning around in unison, forming the vast majority of the world’s stored electronic information. (Gee, I suddenly feel like Carl Sagan)
When I think about things in these terms, it seems like a pretty strange way to store information, and perhaps a bit precarious, considering how heavily we rely upon it. But this Magnetic Boogaloo has been going on for some time, and today we’ll dive into the early days of spinning storage, and a few other weird things from the early days of computing.
Those Few Other Weird Things
In a previous Mad Ned Memo, we had a great discussion about the (very) old days of mainframe computing, when the shift from mechanical systems to electronic ones necessitated faster mechanisms for storing data than the punched-paper approach that was prevalent at the time. In that article, I talked about strange electronic storage systems, like the Delay Line Memory and the Williams-Kilburn Tube.
My readers had a couple of contributions of their own to this list. Martin Cohen commented about using UCLA’s SWAC computer with Williams Tube memories dating back to 1949. He was programming it in 1964, so the machine was already very obsolete but appreciated that he could get time on it. I especially liked hearing about how this machine supported binary punch cards - I had not thought of that combo, all the punch cards I ever dealt with were just pure text. But it makes sense, given paper tape is essentially binary. Anyway Martin you win the award I would guess for oldest memory system programmed by a reader!
Reader EricV mentioned a pretty unique use for ferrite cores, in the EGRET instrument of the Compton Gamma Ray Observatory which launched in 1991, to study high-energy photons. The cores were apparently used to digitize signals coming from the gamma-ray detecting ‘spark chamber’. I didn’t see that detail in the wiki article, but EricV designs high-energy physics experiments so seems to know what he is talking about. Way cool, to see an old technology like core proving useful as late as the 1990s!
My favorite contribution though was from Tom Thornton, who mentioned a super-obscure memory system I had not heard of called Rod Memory, which was created by NCR in 1962 for their NCR 315 series of computers. Rod memory used tiny little rods to store information magnetically, rather than the doughnut-shaped cores of core memory. This geometry apparently allowed for some automation of assembly, as the rods could be magnetically aligned to fit into a matrix without manual intervention. I think though the significant note in the article was, the design avoided NCR having to pay royalties to IBM, who held the patents on core memory.
Thanks for all those weird contributions! Several readers though did point out a glaring hole in my storage article, that being my omission of memories involving moving magnets. Core memory, while magnetic, does not qualify for this category, but disks, drums, and tapes certainly do. I left these important things out because it was just such a big category to delve into, it would have blown up the article to an unwieldy length. I’ll try to make up for that, now.
Oberlin Smith, Long-Lost Father of Magnetic Recording
When you hear an old Edison-era recording from the 19th century that is full of scratches and pops, you either assume it has those problems because it is so old, or maybe because the old-timey people who recorded it were just happy to record anything in the first place, and didn’t really worry too much about the low quality.
Turns out that neither is the case. Those old recordings had pops and scratches even when they were new, due to the nature of what happens when a mechanical stylus travels over a medium with sound vibrations encoded on it, but which also contains flaws of some sort. Anyone who has played their Dark Side of the Moon vinyl LP too many times and/or left it outside the dust jacket on a non-clean surface (looks around guiltily) can confirm this phenomenon.
But it is also untrue that everyone was happy about it, even in 1898. Enter Oberlin Smith, industrialist and owner of a large metal press business that supplied parts to everything from rifles to Model T’s. Smith is described as a “tinkerer”, who could not look at some gadget without thinking of how to improve it. I would say he meets the modern-day definition of hacker, who among other things invented a remote control phonograph system so he could sit in a chair and select a record without getting up - pretty much the first jukebox.1
Smith seems to have been somewhat of an audiophile as well, because upon hearing Edison’s foil-cylinder sound recordings during a visit to his famous laboratory in 1897, thought the quality could be improved, and decided to do something about it. His idea was an early entry in a classic tech playbook: substitute a mechanical process with a non-mechanical one. In this case, getting rid of the vibrating stylus following grooves on a cylinder in favor of a magnetic thread reel and pickup coil.
Not a complete elimination of all mechanical action of course — that innovation really didn’t come until the solid-state memories of the 21st century. But merely eliminating the mechanically-encoded grooves of Edison’s recording medium in favor of a magnetic version that could be read electronically was a huge idea. One that would power the information age to come.
Unfortunately for Oberlin Smith, he was much too busy with his other business concerns to either patent his idea, or to commercialize it, and at one point just put it out in the public domain. Maybe it was just that he liked to hack and didn’t really want to pursue it further, I don’t know. He was in fact already rich and didn’t really need the royalty income.
Whatever the reason, Danish engineer Valdemar Poulsen ended up credited with inventing magnetic recording, even though he had previously seen and essentially copied Smiths’ ideas. To give Poulsen his due though, he was the one who came up with a practical magnetic wire recorder, whereas Smith never got anything out of the lab. But Smith went unrecognized for his contributions to magnetic recording until his original disclosure article was rediscovered in 1985.
Let’s Go To The Tape
After that, a bunch of other guys figured out how to bond metal oxide to tape, and we got magnetic tape recording, along with BASF, magnetic media giant. Vacuum tube amplifiers were on the scene by 1910, and the magnetic audio recording era was in full swing. It wasn’t until the 1950s though that anyone managed to use magnetic tape with a computer - something called UNISERVO for the Univac I Computer. Although it offered a pretty low information density of 128 bits per inch, it must have seemed like an enormous treasure trove of storage compared to other formats like punch cards.
The tape format lives on to today. IBM’s TS1160 tape drive for instance was released in 2018, holding up to 40TB per tape at a transfer speed of 400MB/S. That is not super-fast, when you do the math about how long it will take to record a full tape. But tape systems never were speed demons, and never had to be. Their mission was to provide removable storage capacity, which could be scaled just by buying more media, and allowed for things like archiving.
I still have a roll of DECTape - Digital’s “personal-sized” magnetic media developed for their PDP series computers. I bet if I could find a TU55 drive somewhere I could read all 256K or so of it, even today.
But let’s also be honest. Tape drives are also a pain in the ass. They make for great sci-fi movie props, but the rapidly spinning wheels on those soon-to-be-destroyed-by-Godzilla control center computers were probably rewinding, or seeking. Both of which were necessary evils in the tape game. As an owner of several cassette tape storage computers in the 1980s, I can tell you of my troubles locating a program on a cassette that I used to store multiple programs on, manually searching for a program start point back and forth on a tape. (yeah I know that’s what the counter is for - if you remember to use it!)
Cassette storage for early PCs was not even a truly electronic storage system — it basically operated as an audio recorder, storing modem-like sound output from the computer. This was really not super-efficient or fault-tolerant, compared to other commercial tape drives of the day. But it allowed companies like Radio Shack and Commodore to use off-the-shelf audio cassette recorders as a storage device. (Probably more importantly, to rebrand off-the-shelf audio cassette recorders as computer storage devices, with a very profitable markup.)
Drum Machine 1.0
This drum machine above did not provide the beats for Breakin 2: Electric Boogaloo — the ZAM41 pre-dates that 1984 classic by a few decades, and the general computer idea of a drum machine goes back even further, to the 1930s.
Drums were the precursor to hard disk drives, and were simpler to design because the heads did not have to move, and the tracks on the rotating drum were all the same length, unlike the concentric tracks of a disk. I was surprised in my research to learn that one of the earliest drum machines on something called the Atanasoff Berry Computer worked on capacitance rather than magnetics, and the drum consisted of banks of them that stored the data.
Magnetic drum storage started as a main memory solution for computers and were pretty small in capacity - under 100K bits. Eventually, they found a place as secondary storage systems, and drums were in regular use up until the 1970s, and if you consider the frighteningly outdated systems of the US Military, beyond that even. But in the mid-1950s, a new storage technology enters the chat.
Insert Drive Pun Here
I’m on deadline, so I will leave it to the readers to come up with a good disk drive topic header pun! “The Drive to Success?” … “Baby You Can Drive my Commodore?” I don’t know.
Hard disk drives came on the scene in 1956 when IBM introduced the IBM 350 storage system, a complete beast of a disk drive containing 52 double-sided platters and weighing one ton. Super cumbersome, but the mid-century programmers using the RAMAC system they literally built around this drive then had the pleasure of five whole megabytes of storage to work with, which must have seemed like a galaxy. (A funny thought I had was: this article might not fit on it)
You got to hand to it IBM when it comes to thinking big. No disk drive at all had ever been commercially sold when they built RAMAC. So you would think they would start with like one platter, or maybe a couple. But no, IBM said 52 platters is what we are going with. Ballsy.
My favorite part of the article about it though is the part about how IBM’s management initially killed the project because they thought it would be a threat to their punch-card business. Which was of course correct, but I guess it took them a while to embrace the you-must-eat-your-own-young world of technology creation and obsolescence.
Speaking of beasts, a personal favorite old mainframe disk drive was the Librascope Disk File, only two of which were produced. This two-ton monster featured a four-foot platter, spinning at 900RPM, and holds the Guinness world record for largest disk platter. I seem to remember a story from a visit to the Computer Museum (when it was in Boston), about some mammoth hard drive like this on some old computer breaking loose from its mount and going across the room. I would think the angular momentum on some of these mega-drives would be killer. But I could not find any accounts of this type of failure - reader contributions also appreciated here!
In the case of the Librascope, it was a head crash that constituted the catastrophe, wiping out at least half the disk on one of the units. As an aside, I think it’s pretty interesting that the heads of a disk drive need to ‘fly’ over the disk surface, at a perilously close distance. And if for some reason the power fails or something goes wrong with the servo controlling them, they can crash into the magnetic “ground”, destroying the drive.
That’s something that in theory can still happen even today, but modern disk systems have many more levels of protection that minimize the odds of that happening. In fact, if we move forward in time from these early drives, there are a lot of things that are essentially the same. Smaller drives, higher density, better electronics. But still magnets, moving in circles.
Even though almost every metric, today’s storage solutions are better, I still think we are missing out on the excitement of designing and using these unwieldy storage beasts. For instance, the sound that Librascope drive made when the head crashed on that four-foot platter must have been incredible! And imagine getting to drive a forklift to install your next disk drive. Also exciting was the profit margins on storage back in the day - IBM leased the RAMAC drive at $3200 a month to customers.
$3200. A month. For 5MB. Come to think of it, I guess I’ll stick with my 512GB SSD. Even if I could fit this article on a RAMAC drive, I don’t think the current Mad Ned Memo income could pay the leasing fee.
Explore Further
Next Time: The Victorians created a vast network of copper cables and switching stations, allowing for instant realtime audio communication and signaling between distant points. Sound like a Steampunk novel? Nope, just POTS - the Plain Old Telephone System! Invented so my friend Dave and I could make crank calls in the 1970s. How Ma Bell was just swell coming up in: Wardialing and other Phoney Stories.
It’s a hard sell getting people to sign up for more email, I know. But think about it: What do you have to look forward to in your Inbox? It could be the Mad Ned Memo! Get your weekly dose of nerdy computer stories and discussions, nestled right in there between your credit card bill and UPS shipping delay notification. The Mad Ned Memo is ad-free and cost-free, and you can unsubscribe at any time.
The Mad Ned Memo takes subscriber privacy seriously and does not share email or other personal information with third parties. For more information,
click here
.
on a clear disk you can seek forever
A while back I became fascinated by the Burroughs 205, the wall of blinkenlights featured in Adam West's Bat Cave, and also in Lost In Space (the 60s TV show, kids). Imagine the shock of realizing that that spinning drum was not like a disk drive, but was in fact a Fully Operational main memory! That's probably the main reason the poor thing topped out at 75kHz clock cycles. I haven't had the time to dig into the docs, but it seems a safe bet that the 'D' register was really a shift register to interface to the nibble-stream I/O with the drum ..