36: Historical Non-volatile Memory
So far, we’ve managed to manipulate and compute data if it is in RAM. How it got, there is another question. What we would like to tackle today is how one manages to store memory without the use of voltage. That is, we would finally like to store memory without the need for continuous power supply, resulting non-volatile memory, a type of memory that exists, without the need to attend to it with electricity (for modern computers), or anything else.
We know that whenever we wish to manipulate data on a computer, we need to load it into RAM, which then gets transferred into the internal registers of the CPU. Usually, this data comes from a non-volatile memory storage system that exists alongside the RAM, where certain vital files are stored, such as those making up the software that is the operating system. Note that with RAM, given that it consists of flip-flops or capacitors, requires power/voltage to be able to store memory. Non-volatile memory however, does not require this, which means we need to rethink the design of the hardware we are using.
Though we are presenting non-volatile memory far into the blog, that it may seem like a new technology, non-volatile memory is literally as old as history itself. After all, aren’t books, that are written on paper and papyrus, also a type of non-volatile memory, i.e. memory that doesn’t require any further attention from us to persist? If everyone in ancient Egypt went to sleep in the evening, would the messages on the papyrus (or carved in rock) still exist? Yes.
So let’s specify, and remember that we are talking about non-volatile memory as it relates to computers, computation, and automation, and not to stone inscriptions, papyrus or printing techniques. Scoping our history of non-volatile memory this way, we find that non-volatile memory started with looms in the industrial revolution. The first looms were created thousands of years ago, of course, and they have worked pretty well since then. But the 18th century saw the automation of patterns being made in carpets, so that people would not need to manually thread wool strings to create designs on textiles.
Of course, the creation of the technology did not occur in one step: One of the first people to build a mechanical loom for non-embellished textiles was Jean le Calabrais in 15th-century Italy, whose machine was eventually improved on by Basile Bouchon in 1725. Several more people had made improvements until someone had added a new feature to looms that cleverly used wooden planks to store bits of information.
This machine, one that is now able to automate pattern creation, is known as the Jacquard Loom, created by Joseph Marie Jacquard in 1801. It came to be revolutionary for textiles, but funny enough, had little effect on computers, even though it pioneered memory via non-volatile hardware know as punch cards (see above picture).
In essence, a thread was always either above or below the perpendicular-running other threads: “up”, or “down”, 1 or 0. Punched cards, at the time, were wooden cards that stored “data” via strategically-placed holes located on the card, a design that Charles Babbage would later borrow from to (in theory, at least) program his difference and analytical machines. This way, through some clever mechanics, holes decided which way the yarn would be threaded , hence, unknowingly at the time, creating a medium for non-volatile memory.
For each new pattern design, a large collection of punched cards could be tied together that would perfectly create a pre-programmed pattern in the textile. And this, perfectly for every single carpet or textile being created.
Because punched cards recorded information, they could be automatically used for other information processing machines as well. In the 18th century, when machines were automating computation and manufacturing, punched cards were the first tool to be used for memory storage and were widespread for about 150 years until the 1950s, when magnetic tape arrived on the scene. Amazingly, even though they were in continuous gradual decline and for the most part even unknown, punched cards (now made with paper) were still in use until the early 21st century in the US*.
Given this example, it is clear that magnetic tape did not instantly obliterate punched cards, as replacing all of the hardware of working machines would be time-consuming and, most of all, expensive. For instance, the UNIVAC I (created in 1951) uses punch cards to operate it, and no one was interested in buying a new, expensive computer just because it is the newest thing. When widespread magnetic tape usage became a reality, an extension device known as the UNITYPER was created at around the same time that magnetic tape accepted input, which transferred the information into the UNIVAC from magnetic tape instead of from punched cards. This was the “nail in the coffin” for punched cards.
Get 365 Days of Computer’s stories in your inbox
Join Medium for free to get updates from this writer.
Interestingly, by the time magnetic tape had taken the role that punched cards occupied in the United States, it was previously used for decades in Germany, where it was invented in 1928. Because of the history of the country until 1945, the technology only arrived in the rest of the world after that, which meant that while the newer technology did exist, computer engineers didn’t have access to it until much later (even if American research at the time independently invented magnetic tape storage).
Data was saved and read to and from magnetic tape using the idea of magnetic hysteresis as follows: Let us construct a long line of consecutive, insulated paramagnetic (in this case, iron) lines and ensure that none of the insulated metal units are magnetized. We can now change the magnetization of each unit by having a magnetic field pass through the metal, which gives it either + or — magnetization. Because of hysteresis, the magnetization remains on the metal, which means we can encode binary information into the tape, by differentiating their dipole moments (north and south)
To be able to do this, we need a very fine conductive needle through which a current passes through. When we bring the needle close to a unit, the magnetic field lines pass through the storage medium, which magnetizes it, i.e. changes its magnetization. Then we simply need to move the needle to the next unit, where we can save another bit of information in the same way
Magnetic tape, while not metallic per se, isn’t only metal. It only has iron particles strewn over the surface of the (plastic) interior, which captures the metallic magnetization sequences on top of it. The details of the exact makeup of the tapes depend from one brand and technology to another, but the principle is the same.
In wider society, magnetic tape was also widely used outside of storing computer data, such as in recording images (polaroid cameras), “movies” (VHS (Video Home System) & Betamax), and music (cassettes). Film, for example, the tape used to record “movies”, is still in use from time to time, depending on the filmmaker’s needs/desires. In computers, many different formats of magnetic tape storage have emerged, which reflect the improving tape technology, though they always came in the form of disks or spinning units. This way, one could spin a disk and have a needle pass over the disk, without having to move the needle.
The first floppy disks came out in 1972, called “floppy” because the tape came in a soft, plastic case that was, well, floppy. It made a funny sound when you shook it. Haha.
Later on, in 1976, the more widely used 5 ¼ inch floppy disk emerged, until in the 1980s, one of the most well-known floppy disks emerged: the 3 ½ inch floppy disk. This disk was technically still floppy, but it was now protected in a plastic case with a shifty metal thing that would be pulled back when the computer wanted to read the data on it.
Because this was when software began incorporating images into their user interface (UI), the floppy disk seemed like the most logical image to use when trying to communicate that one wanted to “save to disk”, i.e., save to a physical medium.
If we go beyond the physical floppy disk, we begin entering into the computer era I grew up in. Non-volatile storage was still being used as moving parts at this time: VHS and cassettes were still around (they spun the spooled tape to expose sections of it to a tape reader), but compact discs (CDs) and its cousins, DVDs and Blu-Ray, took more and more of the market share of data storage over time.
All three, and their many other cousins, could save data on polycarbonate disks (carbon discs), in which little grooves would (literally) be burned into using a laser. Then one could read 1s and 0s, depending on whether there was a short groove or a long one. Hence, the expression “burn to disk”, was nicely accompanied by a little burning CD icon.
Of course, CDs are still in use, so placing them on this list seems odd. CDs are also still sold (unlike VHS tapes) and are still used. I was even given a bunch of CDs to copy as part of my job, just a few days ago. Either way, the elegant CD is on its way out. If it is being used today, it tends to be because the data on it has not been copied to a more modern storage system.
That’s about it for historical non-volatile. There are plenty of different contemporary ways to save data that we haven’t touched upon just yet, but will in post 39. We are still far away from the computer of today, or any general-purpose computer for that matter. For this reason, we will need to be satisfied with this basic overview for data storage. And while CDs and DVDs can store up to ~0.5–15GB, we will find out what is being used to store data on the order of terabytes (1000s of GB) without sacrificing more space than a small, handheld box.
Footnotes
*I rue the day I was introduced to the style-outdated forms of the US government. What a mess…
