>> Wafer memory... > "Wafer Scale Integration" I believe he referred to it as at the time. > An ambitious project back then (maybe even today) because it requires > flawless wafer sized pieces of silicon. Even though silicon is the > earth's most abundant (or second or whatever) element, the cost of > fabrication of silicon integrated circuits is related to the yield of > working chips from each wafer. Even some of those that fail to make the > grade can sometimes pass at lower clock rates and can be marked and > sold as lower spec. With wafer scale, it's all or nothing. The > rejection rate must be huge, and that is probably what ate most of his > investment in plant - the production of flawless silicon wafers.
> Bzzzz. Wrong. It refers to a bigger microdrive like device, > with magnetic tape. The same infinite tape trick as in microdrive, no > rewind needed. It was the alternative to floppy. (or the evolution of > classical tape). Bzzz - you are both wrong! First, it was not a tape. Do not confuse WM with something that was a tape, called Wafadrive or Waferdrive, which was made for the Spectrum. It had loop tapes similar to microdrives just much larger (they looked suspiciously like a cut down version of an US 8-track tape). As far as I know, they never made it to the QL, and that's a good thing too. As for WM requiring a flawless wafer of silicon, this is not true either. The whole project at Sinclir started when Sir Clive bought the patent rights for WSI from a rather strange individual by the name of Ivor Catt, and hired him as a consultant. Although Ivor Catt is a rather cranky and dissilusioned (to the point of paranoya) man who calls himself a 'technocrat', who has rather interesting ideas about transmission lines, he can very well be considered the worlds authority on WSI, in fact he wrote at least one book on the subject (I actually read it), along with several other books. The technology used by SRL would have been a patent called 'Catt Spiral' (the explanation can be found in the aforementioned book). However, all of this was in 1985/6, and we all know that a lot of rather bad things happened in the UK computer industry during that time, and in the several years that followed. Here are links to a couple of articles on the subject: http://www.electromagnetism.demon.co.uk/z005.htm http://www.electromagnetism.demon.co.uk/rh25cbl.htm Catt's patent addresses technologies used to circumvent flaws on wafers, which are not only common, but appear on 99.99% of all wafers, based on the idea to provide access points along the wafer edge, and have the chips themselves connect together using fuse technology similar to what is used in old programmable logic chips, once at least one working chip is found at the edge of the wafer. Strictly speaking, Sinclair Research had only a short time with this project before it was sold off to Amstrad, providing a 'venue' to employ the worlds first mass-produced wafer memory in, that being the QL. Typicall of Sinclair (and I dare say of the UK computer industy - of which almost none remains today as the eventual result!), the signifficance of a successful WSI project, after many multinational corporations already poured billions into failed attempts, was not recognized. At SRL it only went as far to create a new company called Anamartic. Eventually, the patent rights which were used as securities for Sinclair's loans reverted to the bank. Even so, after a comparatively small investment of about 30M$, they came up with a working product, which was used in machines by Tandem. Even so, as it's obvious today, wafers never made it to the mass market, but when I last looked (somewhere in the mid 1990s), they were made by Fujitsu (it is possible the patents were acquired through ICL who was one of the investors in Anamartic) - as solid state memory for special applications (read: military, black box recorders, etc). For a while they did offer 20 and 40MB solid state hard drives based on the technology (around about 1985/6), don't ask about the price! The wafer technology was of course based on chip technology and design and testing procedures of the time. Today it would actually be easyer to produce them as things like mid-process testing, late stage metalization, flash memory, entifuses etc. are well suited and well known technologies for wafer scale integration. Originally, Catt intended this to be a (parallel) computer on a wafer (in fact, his 'spiral' patent was later superceeded by something called the 'kernel machine'). Memory was initially intended as a 'back door' to peddle the concept and bring it to a mass market. Today, using it as solid state memory would actually immediately find a market in various devices that use flash chips. Nasta
