This is not about Taito’s excellent platform game. This is about bubble memory! More precisely, about magnetic bubble memory technology. Invited in the ’70s, bubble memory was supposed to be the best thing since the invention of sliced bread. And indeed, on the paper – and put back in its time – the technology was promising. First, it was non-volatile. Take that DRAM! Second, it had no moving parts, which is always a good thing – especially in space applications.
Take that hard drive! To top this, bubble memory is working fine at extreme temperatures (- 40C to +100C for example). But, except for the happy few, I am sure that you probably never heard about this technology, right? So how comes such a great marvel made such a flop in the 80s? Well, in one sentence, I would say that while the competing technologies didn’t forget to evolve over time, magnetic bubble memory had also its own share of limitations.
I first interacted with the technology while using the Sharp PC-5000 computer. If you dig deep into this collection, you will find a dedicated post with a bunch of pictures of this really nice machine. But even before, as probably many of us, I was drawling affront of articles about the GRiD Compass 1101 and its bubble memory modules providing this prodigy a whopping 384 KB of storage space (that’s 3 x 128 KB, which makes sense since it had three modules). I am still looking for one of these rare gems… Prior to talking about the causes of this epic fail, let’s see how the technology work? First, there is no liquid or foam in this memory, so no risk of leakage (not like with that rubbish LCD… Ok, I stop joking, LCD doesn’t leak :-)).
The term bubble comes from the visualization of how the magnetic field in a thin garnet layer is twisted to store the data (bits). The magnetic information – the bubble – is created and moved in the 2D space using magnetic fields. The bubble itself is the magnetic property of the material perpendicular to the surface at a given position (north-facing up or down). As-is, the material has natural regions polarized the same way (imagine the black patterns on a cow and you get close to what is happening here). By applying perpendicular magnetic fields, one can shrink the size of these patterns until they have the size of a bubble. Hence the name. The idea is then to organize these bubbles into loops and read and write the information at a given point in the loop. The 2D movement comes into the picture when you consider that, to be read, for example, the bubbles need to progress thru the entire loop until they reach the read position (in a single file). So, the further out the bubble of interest is in the loop, the longer it takes to read it! As you already guessed, this is one of the main limitations of the technology. Compared to minuscule capacity hard drives, this was acceptable, but it really was unsustainable as drives improved in performance and capacity. Also, non-volatile technologies such as flash memory became redoubtable competitors. Intel manufacture in the early 80’s one of the most common bubble memory, the 7110, a 1 Mbit bubble memory package. Note that besides the Sharp PC-5000 bubble memory cartridge, all the pictures of this post are of the Intel 7110A-1 (the first spin).
This package provides 128 KB of memory organized in 256 loops of 4096 storage bits. It is operating at 100 Kbits/s with an average access time of 40 ms. The package is really a beast! Just think about it, you had to screw it onto the motherboard because of its weight. Intel also had a chipset to ease the development of bubble memory applications: the iD7242 – the sense amplifier –, and the iD7230 – the current pulse generator. There were few others, but I don’t have them. I am sure that you are as curious as I was about what is luring in this package, so I went ahead a tore one module down as far as I could. Enjoy!