Magnetic Tape
The most well-known version of tape-based magnetic storage is the kind used for media. When tape-based recording was first introduced, it revolutionized the talk show and DJ-ing scene of the time (mostly post WWII) because it enabled shows to be recorded and played later, rather than live. Music recording tech already existed, but it required physical interaction from the DJ, so it wasn’t as hands-off as tapes were.
The second-most well-known version is the kind used for computer memory! Data is stored on the tape in the form of little magnetic ‘dots’ that the computer can read as bits. Before each pocket of data dots is a data marker that tells the computer how long that pocket should be, so it knows when one set of data ends and the next begins. The polarity of the dot determines it’s bit value, and the computer can then read all these dots as binary code.
This method of data storage was a massive breakthrough, and other mediums continue to use the format even today! Tapes are still in use for big stuff – parts of IBM’s library rely on modern tapes, which can now store terabytes of information at a higher density than disks and flash drives alike. Other memory types relying on magnetic domains include hard disks and drums, to name a couple. All that separates them is material and know-how: the better the magnetizing material on the outside, the smaller the domains can get. The better the insulation between the domains and regular old entropy, the more stable the data is!
Carousel Memory
Carousel memory was an attempt at shrinking the space that magnetic tape took, but to the extreme. Instead of one very long piece of magnetic tape on a bobbin, the carousel memory system uses several smaller reels of tape arranged in a carousel pattern around the central read mechanism. To get to the right info is as simple as selecting the right reel! This has some issues with it, as you might imagine. Moving parts add complications and an increased risk of mechanical failure to any device, but a device carrying thin, delicate magnetic tape on it is an especially bad place to start.
However, it wasn’t all bad. Carousel memory was actually quite fast for the time because it didn’t have to rewind or fast-forward as much to get to the right area of code. It could skip feet of tape at a time! This advantage declined as tape tech improved, but it still helped companies trying to squeeze the most life from their machines. The bobbins and individual ribbons were all replaceable, so the tape wasn’t worthless if it got torn or damaged. The carousel itself was also replaceable, so the many moving parts weren’t as much of a curse as they’d be on, say, the first hard disks, which had irreplaceable heads.
Core Rope Memory
Core rope memory featured magnetic gromets, or ‘cores’ on metal ‘ropes’, and then those ropes were woven into fabric the computer could read. In ROM (read-only memory) format, if a wire went through the core, it was a ‘one’, or a ‘yes’. If it didn’t, it was a ‘zero’, or a ‘no’. In this way, the fabric is physically coded into binary that the computer can use. ROMd Core-rope memory involved quite a bit of complicated weaving and un-weaving to get the cores in the right spots.
Core rope memory was chosen over tape memory for the Apollo missions, mainly for weight purposes. Tape was great, but not nearly dense or hardy enough for the mission yet, and neither were the other similar core modules available to NASA. A read-only core-rope memory module could store as many as 192 bits per core, where erasable core memory could only manage one bit per core. Where each core on the final module depended on reading the wires to determine the bit’s state, the erasable model (core memory) read the core’s magnetic state to determine the bit state, not the threads going through it. The final module sent up to get to the moon was a total of 70-ish pounds and read fairly quickly. Tape, core memory, or hard disks available at the time couldn’t have gotten to the same weight or speed.
Core-rope memory has its place. It’s very sturdy, and since it relies on the cores to act as bits, it’s possible to visually identify bugs before the memory’s even used, unlike core memory. Both are sometimes called ‘software crystallized as hardware’ because of the core system. It isn’t seen much today, since it is still incredibly bulky, but at the time of its use it was revolutionary.
Core Memory
Core memory is the older sibling of core rope memory, and it stores less. However, the people who got to work with it call it one of the most reliable forms of memory out there! Core memory works much the same as core rope memory, where the bits are stored in cores.
However, the formats are different. If core rope memory is like a binary-encoded scarf, core memory is more like a rug. Thin threads made of conductive material are woven into a grid pattern, with cores suspended on where the threads cross each other. The computer understands these threads as address lines, so asking for a specific bit to be read is as simple as locating the X and Y address of the core. A third set of lines, the sense lines, runs through each core on the diagonal, and this is the thread that does the actual reading.
When asked to, the computer sends a current down the sense threads and sees if the cores flip their magnetic polarity or not. If it doesn’t, it was a zero. If it does, it was a one, and it has been flipped to zero by the reading process. This method is known as ‘destructive reading’ as a result, however, the computer compensates for this by flipping the bit back to where it was after the reading. Due to its magnetic nature, the core then keeps this info even after power to it is cut!
This link here is an excellent, interactive diagram of the system.
Even though this improved the bit-to-space-taken ratio, core memory still aged out of the market. With the price of bits decreasing rapidly, core memory got smaller and smaller, but the nature of its assembly means it was almost always done by hand – all competitors had to do was match the size and win out on labor. Soon, its main market was taken over by semi-conductor chips, which are still used today.
Magnetic Bubbles
Magnetic memory has had strange branches grow off the central tree of progress, and magnetic bubble memory is one of those strange shoots. One guy (who later developed other forms of memory under AT&T) developed bubble memory. Bubble memory never took off in the same way other magnetic memory styles did, although it was revolutionary for its compact size – before the next big leap in technology, people were thinking this was the big leap. It was effectively shock proof! Unfortunately, better DRAM chips took off shortly after it hit the market and crushed bubble memory with improved efficiency.
Anyway, bubble memory worked by moving the bit to-be-read to the edge of the chip via magnets. The magnetic charge itself is what’s moving the bits, much in the same way electrons move along a wire when charge is applied, so nothing is actually, physically moving within the chip! It was cool tech, and it did reduce space, it just didn’t hold up to semi-conductor memory chips. They saw a spike in use with a shortage, but they were so fiddly that as soon as DRAM chips were available again, they went out of style.
Semi-Conductor DRAM – Honorable Mention
DRAM chips are a lot like core memory, in that the device is reading the state of a physical object to determine what the bit readout is. In Semi-conductor chips, that physical object is a tiny capacitor, hooked up to a tiny transistor, on semiconductive metal-oxide material. Instead of determining magnetic state, the device is instead checking if the capacitor’s discharged or not. No charge = 0, yes charge = 1. These chips aren’t technically magnetic, but since they’ve killed so many of the other options, here they are!
DRAM stands for Dynamic Random-Access Memory, and it means that the memory can be accessed randomly instead of linearly. As long as the computer knows where the data’s stored, it’s able to pull it without pulling other files first. They’re still being sold today!
Magnetic Disk (Hard Disk Drive)
Hard drives work more like tape than core memory. A Hard drive is a platter (or a stack of platters) with a read-write head hovering above it. When you want to save data, the hard drive head magnetizes areas in binary to represent that information. When you want to read or recover that data, the head interprets these areas as bits in binary, where the polarity of the magnetized zone is either a zero or a one.
The zones of magnetization are incredibly tiny, which makes hard drives one of the more demanding memory forms out there, both now and back then.
Early hard drives could suffer from ‘de-magnetization’, where a magnetic disk’s domains were too close and gradually drew each other out of position, slowly erasing the information on the disk. This meant that the disks had to be bigger to hold the data (like everything else at the time) until better materials for data storage came along. Even though they held more capacity at launch, they were passed over for smaller and more stable stuff like tapes and core memory. The very early drives developed by IBM were huge. Like, washing machine huge. They didn’t respond to requests for data very quickly, either, which further pushed reliance on tape and core technology.
Over time, hard disks improved dramatically. Instead of magnetic zones being arranged end-to-end, storing them vertically next to each other created even denser data storage, enough to outcompete other forms of media storage entirely. Especially small hard drives also come with a second layer of non-magnetizable material between the first layer and a third layer of reverse-magnetized ‘reinforcement’ which keeps the data aligned right. This enables even more data capacity to be crammed into the disks!
Some time in the 80s, hard drives finally became feasible to use in personal computers, and since then they’ve been the standard. SSDs, which don’t have any moving parts whatsoever, are beginning to gain ground in the market, but they can’t be truly, irrevocably erased like hard drives can due to different storage techniques. Hard drives are going to stick around a while, especially for the medical and military industries, as a result!
Sources:
https://spectrum.ieee.org/tech-history/space-age/software-as-hardware-apollos-rope-memory
https://www.apolloartifacts.com/2008/01/rope-memory-mod.html
https://electronics.howstuffworks.com/vcr.htm
https://www.apolloartifacts.com/2008/01/rope-memory-mod.html
http://www.righto.com/2019/07/software-woven-into-wire-core-rope-and.html
https://www.computerhistory.org/revolution/memory-storage/8/253
https://www.rohm.com/electronics-basics/memory/what-is-semiconductor-memory
https://cs.stanford.edu/people/nick/how-hard-drive-works/
https://psap.library.illinois.edu/collection-id-guide/audiotape
https://en.wikipedia.org/wiki/Carousel_memory (all primary sources regarding carousel memory are in Swedish)