Nineteen seventy nine saw Clive Sinclair and the NEB part company. Left, once again, to his own devices the electronics mail order king turned his attention to the home computer. The microcomputer had previously been the preserve of wealthy computer buffs and forward thinking businessmen, but Clive Sinclair was about to change all of that.... We continue our series of articles on the man who brought us the Spectrum.
Clive Sinclair has always been, and probably always will be, in the business of consumer products.
In the mid-70s, the National Enterprise Board injected capital into his pocket TV project, funding it on a fifty-fifty basis. The instrument side of Sinclair’s activities thrived, dominating the market — but soon after the NEB joined forces with Clive Sinclair they lost faith in his consumer electronics interests.
Admittedly, money was being lost by Sinclair on a number of fronts, and he had suffered a heavy loss on his black watch project. However, the whole consumer electronics industry was going through a bad patch, with cut-throat competition making life difficult for everyone in the field. The Japanese hadn’t entered the instrument market and Sinclair’s digital metering equipment led the field. The NEB wanted to concentrate effort on the range of meters.
Despite Clive Sinclair’s argument that his company had the winning technology and was bound to be successful with one-per-person electronic products, the NEB had no confidence in the future of consumer electronics. Eventually, the partnership was dissolved. In 1979 Clive Sinclair left with a nucleus to start a new company, leaving the NEB to make meters.
History proved Clive Sinclair to be right.
The MK14 microcomputer rapidly grew from a single board machine into a multi-board, modular system which included a cassette interface card, and a VDU module which permitted characters or graphics to be displayed on a domestic TV, and an EPROM programmer which allowed the user to transfer code to a chip and use the MK14 as a dedicated machine by replacing the monitor with a custom programmed EPROM chip.
The technology behind Science of Cambridge’s first computer was undeniably smart, but appealed mainly to electronic buffs, and was never really destined to be a major seller.
Home computers were still very much at the hobbyist level in the late 70s — relatively expensive complete machines were available, but tended to be the preserve of the businessman and the academic, rather than the ordinary person. Those computers that were to be found in the home were generally built from kits by dedicated wielders of the soldering iron.
In 1980, however, with the launch of the Sinclair ZX80, the foundations of the home computer market as we know it today were laid. Selling for £80 in kit form, but available ready built for £100, the ZX80 brought real computers within the purchasing power of most families — and you no longer needed to be technically inclined as well as a keen machine code programmer to run and use a home micro.
In combination, the ZX80 manual and the machine itself provided an easy way for the novice to begin programming in BASIC. The Sinclair keyword system of entering program statements from multi-function keys rather than entering them character by character from a standard typewriter keyboard was a major innovation — as was the built in syntax checking. Both these features were carried through into the ZX81 and Spectrum — not least because each successive machine to leave the Sinclair stable contained a chunk of its predecessor’s ROM.
The ZX80 had a 4K ROM containing the operating system and a BASIC interpreter. 1K of on-board RAM was included which was soon supplemented by a RAM pack which fitted into a slot at the rear of the machine. The ZX80 was very limited in its capabilities. Like its successor the ZX81, it wasn’t a colour computer and it couldn’t cope with much in the way of serious programming. You could only-use integers (whole numbers) and data for a program had to be stored with the BASIC, in the same file. The ZX80 couldn’t handle cassette files.
There were problems with the Sinclair RAM pack which tended to be rocked when pressure was put on the ZX80 keypad, and quite a few machines failed as the edge connectors at the rear of the machine shorted out. This problem gave rise to non-Sinclair add-ons, and a couple of well-known names currently active in the Spectrum marketplace have their roots in these days. Dk’Tronics, for instance, started making RAM packs, and Quicksilva turned out memory boards.
The ZX80, like the ZX81 and ZX Spectrum, started life as a mail order product. Straightaway, Clive Sinclair went for the person-on-the-street, not just the hobbyist, and he jumped into the market without any research: ‘It’s an absolute waste of time to take a personal computer along to people who’ve never seen one before and say ‘will you pay £100 for this plastic object here. It looks like a squashed telephone’, and people are bound to say ‘what would I want to know about a computer for? Life is far too complicated as it is’.
But want to know, they did. Especially when the ZX81 was launched early in 1981 — half a million were sold within a year.
The ZX81, amazingly, was a better computer than the ZX80 but was sold by Sinclair for £30 less! Sinclair has a policy of passing on savings made on production cost over to the consumer. The ZX81 was designed to use a total of four chips, rather than the twenty one chips (plus voltage regulator IC) used in the ZX80. It was cheaper to assemble, and was priced accordingly.
The ZX81 could cope with floating point arithmetic which was achieved by a sub-interpreter, crammed into the ROM, and written very compactly in a FORTH type language. Square roots, for instance, were calculated using only seven bytes — but the code took an awful long time to run! The Spectrum is so slow on floating point arithmetic because it uses these very same routines, that first appeared in the ZX81 ROM.
So how did Sinclair get rid of seventeen or eighteen chips in one go?
The ZX81 had a ROM chip, a CPU chip, a 1Kx8 RAM chip (the 4118) and an Uncommitted Logic Array (ULA), which improved upon the work of the eighteen chips from the ZX80 which it replaced.
A ULA chip is basically a relatively cheap way of producing a custom chip quickly — the ULA manufacturer puts a mask designed to the customer’s specification over a standard component, rather than designing an entire chip from start to finish. The resulting chip is not as efficient as a custom chip, built from scratch. The big snag with ULA’s is that you can’t be sure that your mask will actually produce the chip that you want until you’ve actually made it.
Sinclair was fairly lucky with the ZX81 ULA — it was pretty much bug free, but the machine suffered from overheating because Mr Sinclair stretched the ULA to its limits. According to Ferranti who made it for Sinclair, you were only supposed to use 70% of the logic gates on the ‘blank’ chip when you add the mask. Sinclair used all the gates, so things tended to get a bit warm!
Fortunately for Clive Sinclair the ZX81 board had been designed to take on 4118 chip (a 1Kx8 memory chip) or two 2114 (1Kx4 memory) chips. Fortunately, because there was a chip famine shortly after the ZX81 hit the streets, which meant that most ZX81’s ended up with five chips. Two 2114’s replaced the originally-specified 4118. This famine also had a lot to do with the final collapse of NASCOM whose computer used a lot of the 4118 chips.
April 1982 saw the launch of the ZX Spectrum, Clive Sinclair’s first colour computer which came in two versions with 16 and 48K RAMs. (And we won’t bother to explain what a Spectrum does in this column).
Once again, a ULA was used in the Spectrum, and like the ZX81, the Spectrum ROM borrowed from its younger brother. (Machine code freaks can find a tiny 24-byte subroutine hidden away in their Speccy RAMs at address 04AA. It checks for ZX81 program names).
Sinclair wasn’t so lucky with the ULA used in the Spectrum — a fault was found at a very late stage, which meant that the CPU could misread the keyboard half the time. On Issue I Spectrums this fault was remedied with the ‘Dead Cockroach’ — a small circuit board stuck next to the Central Processing Unit, upside down. Issue II Spectrums had a different ULA to Issue Is and they had a different bug, which caused the machine to select the ULA when it shouldn’t have done. This was Kludged too, with a transistor inserted over the top of the Z80 chip, which stopped the CPU and ULA from interfering with each other.
The Issue III Spectrum had a redesigned ULA, together with a redesigned circuit board which allowed extra memory to be added more simply. It was cheaper to make, because the trimmers which fine tuned the colour display on earlier versions no longer had to be set by hand.
Incidentally, the Issue III machine set three unused bits on the eight bit keyboard port differently, placing different values to those generated on the three redundant lines on the Issue I and Issue II Spectrum. Some gumby programmers had written code which read those three unused lines for no good reason, and relied on finding the expected values (when they shouldn’t have). When they found that their code behaved erratically on the Issue III Spectrum, they blamed Clive Sinclair!
By January 1983, Clive Sinclair had sold his first million computers. In November of that year his flat screen TV was unveiled, and he was knighted, becoming SIR Clive Sinclair in the Queen’s Birthday Honours List. Next month, as we didn’t quite get to it this month: interfaces, microdrives and the QL.