Sinclair and the 'Sunrise' Technology: The Deconstruction of a Myth (23 page)

The new machine, it was decided in late 1982, would now be based around the Motorola 68008 microprocessor chip. This was decided on not because of any of the inherent virtues such as power, speed or flexibility on which chip choice is normally based, but because of its perceived virtues as a market ploy and its attractiveness as a futuristic hip chip in 1982. Getting away from the ageing Z80 workhorse chip was presumably seen as necessary to presenting the image of pushing forward the frontiers of technology or some such marketable rationale.

The complex innards of processor chips are not of interest to our story, but a few comments are probably in order to illustrate this particular theme. Data in a computer is all numbers, and is transferred as on/off voltage patterns along ‘buses’, parallel sets of conductors.

Think of it as a road with a certain number of lanes, and you can see that the more lanes, the more traffic (in this case, data) can be handled. All old-style chips used 8-lane highways (an 8-bit bus), both inside the chip and to pass data in and out. The 68000 family of chips uses 32-bit buses internally, but different sizes of bus to communicate with the rest of the computer. The 68008 has an 8-bit bus, the 68000 a 16-bit bus. Other considerations aside, it is apparent that all data coming in and out of a 68008 will have to be chopped up into smaller pieces for transmission, and take longer, than would be the case with a 68000 chip. The performance penalty is serious, so much so that no other manufacturer used the 68008 in a micro.

When first chosen the 68008 was admittedly much cheaper than the full-specification 68000. By 1985 it was only a few dollars cheaper. If the project had been controlled by the ever-cost-conscious Sir Clive, the savings might have made some sense. In the event, in terms of both design and component costs, the difference is minimal. Choosing a limited chip rather than the full-specification 68000 version (now used by the Apple Macintosh, the Atari 520ST and other modern machines) is an example of Sinclair Research’s incapacity to get it right when it matters. It ended up paying more or less the same price for a processor that does less, more slowly, than the correct choice. In any case, the price differences were a matter of a few dollars, and not of major significance to production costs. However the QL saga illustrates the capacity of Sinclair Research to make not only single wrong choices and assessments, but a whole sequence of them.

When the R&D effort started in earnest, there were, to put it mildly, problems. There was no project director, and Nigel Searle was effectively running the show. Sinclair’s explanation for his own non-involvement in the development process centres around Searle’s sensibilities as an MD:

The reason behind that was because Nigel Searle was managing director at the time, and he really wanted a free hand, you know, he was very concerned not to have me breathing down his neck. (Interview, 6 November 1985.)

The notional concept, at the beginning of 1983, was, as we saw, for a portable battery-powered computer with built-in modem, twin Microdrives, the flat-screen display and business software. Those facilities which constituted whatever claims the machine had to represent technological advance got dropped one by one. First the battery-driven capability went, since although the Microdrives could easily run on batteries, the DRAM (dynamic memory) chips necessary would drain too much power. Then the flat-screen display was dropped, itself a blow to any lingering portability concepts. In the light of the performance of both earlier (calculators) and later (C5) battery-powered Sinclair products, one should perhaps be grateful.

The continual promise of the flat-screen display deserves a comment. When the QL development started, it was not in production. During the first half of 1983 the production problems were solved, or at least ameliorated to the point where the televisions could start rolling off the lines. Since the 2-inch flat-screen display of the Microvision has resisted all attempts to increase its size (the only reason the Microvision is viewable is a combination of electronic fixes for the beam path and a thick plastic lens), making a larger version for use on a computer was a closed pathway. A larger display would produce a more distorted image, beyond optical correction. Successful attempts were apparently made to produce a projected image from the small flat screen, one version using a mirror producing a real image hanging in space - a nice idea but a non-starter as far as engineering it for production went. ICL was reportedly none too pleased that the visibly innovative bit of its machine was unproduceable, but presumably consoled itself with the other promised features. ICL could get by with a black and white display (the OPD being dubbed by some ICL wit ‘Work Station Zebra’), since this was all they wanted, but whether the concept of the QL screen involved colour is not known. If it did, then Sir Clive was being more ludicrously optimistic than ever in his assessment of Sinclair Research’s capabilities, however much he could depend on the inventive engineering skills of Jim Westwood, already tried and tested in pursuit of Sinclair’s television visions over decades. Derek Holley’s comments, made in the context of the conventional tube of the first pocket television, are perhaps even more appropriate to the flat screen:

His argument was that once you developed the 2-inch screen and got it running it would be easy to scale upwards, a typical non-engineering approach to it. You take something and double the size of it, he thinks it’s easy, but it isn’t. In fact, it’s as hard to double the size as it is to halve it. (Interview, 13 November 1985.)

So the downgraded ZX83 project lurched along in what one source called the ‘disorganized shambles’ that was Sinclair Research at the time. The absence of a project leader, a board acting divisively and throwing up conflicting views masquerading as decisions, and the lack of co-ordination all compounded each other, and combined with the absence of Sinclair from the R&D scene to produce a fiasco. Whatever Sir Clive’s competence or lack of it in other respects, his capacity to provide drive to anR&D team’s efforts are not in question, nor is his decisiveness. His lack of direct interaction with the project, which however uninteresting to him personally was vital to his own company, cannot be explained solely by his interest in Sinclair Vehicles. Since he attended various of the board and steering committee meetings, and since the QL was the major project, the progress of the machine should have been both of concern to Sir Clive personally, and the subject of report. Whether because of over-optimism, personal and departmental strife between the directors, fear of showing incompetence and triggering Sir Clive’s infamous temper, or whatever other influences were at work in the boardroom cabals, the true situation appears to have been concealed from him. He certainly seems to have known little of the QL’s problems until it was too late for him to resist the pressures driving the company to launch the machine prematurely. Not that our readers will by now think that Sir Clive is averse to premature launches, or overly concerned with the problems of production, only that he has a history of at least producing a prototype of some kind before launching a new idea on the world. His comments to us, if accurate, show that even this charitable view is incorrect:

I knew what was being done. If I’d felt that it was very wrong I’d have said so pretty firmly, so I can’t say I got it right, I didn’t. (Interview, 6 November 1985.)

As development work proceeded on the separate sections of the machine - the Microdrives, the main-board hardware, the software, the case and keyboard, all the concern of separate groups - it became apparent that all was not going as it should. Part of the trouble was that it was an all-new machine, with the necessary need for well-organized development that that implies, rather than a development of a prior model, as with the ZX80-ZX81-Spectrum series. The plans were also based on an inflated view of Sinclair Research’s capabilities, ensuring that time and effort would be wasted in trying to produce features then abandoned, such as the flat screen. Over-optimism in respect of flat-screen displays can be laid firmly at the door of Sir Clive, who first announced in 1980 that the ZX80 would be linked to a flat-screen display’, and additional pressure must have come from the ICL agreement, which specified the flat screen as part of the deal.

Other aborted lines of research wasted more time, changed the specifications and delayed the project. Even David Karlin, the designer of the main QL board and hardware, in a staunch display of Sinclair PR hyperbole in Personal Computer World while the customers were still waiting for sight of the QL, had to admit that ‘there was a spec, but this was modified almost every day’. His comment that ‘it was the machine I wanted to build, although it went through various permutations en route’ makes one wonder about his criteria, if they were not based on simple pragmatism, since they were still trying to get bits working when he made the comments (around March of 1984). As our story of the Sinclair team’s travails continues, the inaccuracy of his comments about this period will become apparent:

Karlin says that the development of the QL was almost trouble-free. In fact, the most difficult problem the team encountered was how to assemble the case! (Personal Computer World, April 1984.)

The PR style of bland and placatory statements delivered straight-faced by Sinclair personnel intent on retaining a veneer of ‘professionalism’ and credibility seems to be absorbed by association with Sir Clive, or else is a factor in his choice of employees. Whatever his PR utility, David Karlin was valuable as someone who knew the microchip manufacturing scene well from his days at Fairchild Semiconductors running a production line. Because of his insider’s knowledge of the industry, it fell to him to replace the absent Sir Clive in the role of cost-cutting component purchaser. Unfortunately, this role diverted effort from the work on what was his first major piece of electronics design. The tendency of Sir Clive to recruit talented people and place them on tasks new to them, on the grounds that they are then not cluttered with old ideas, may give them an interesting challenge, but also increases the likelihood of inefficiency in the ways they go about the job.

For those unaware of the normal progression of microcomputer hardware design, you start with schematics, essentially a paper design. Any standard chips, such as the 68008, can be considered as ‘black boxes’ where the innards don’t need to concern you overmuch, since the chip specification will tell you what comes out for a particular set of inputs and conditions. The design work, after these chips are chosen, proceeds to a definition of the logical circuits necessary to control and manipulate the system. Some portions of these circuits can utilize standard off-the-shelf chips, others will be incorporated in custom-designed chips. Usual practice is then to build TTL (transistor-transistor logic) prototypes of the custom chips, and then of the full machine, using the lego blocks of electronics design, simple chips forming logical elements, to replace the symbols of the schematic. This gives you the capability to test out the correct operation of the circuits and their interactions, and revise as necessary before commissioning custom chips. The next stage is the design of the printed circuit board on which the chips, plus sockets, resistors, and all the other discrete components are mounted, and which provides the interconnecting ‘wires’ as metallic tracks laid down on the insulating board material. Prototypes can then be built with the custom chips, and they can be tested. Several reiterations of complex chips will be necessary for various reasons - logical errors and the like.

This is a fairly well-established development process, and originally must have been plotted out on a timescale by Sinclair Research despite the statement in Rodney Dale’s account that:

It was part of the Nigel Searle management technique never to prepare any sort of schedule showing who was going to do what and by when. Such an approach, he averred, leads people to take more time than they should. (The Sinclair Story, p. 136.)

If this was the case, it would have made for unusually vague ‘management’. Apparently Nigel Searle, at least towards the end of the project, was engaged in not only defining timescales for particular activities, but manipulating them rather arbitrarily. Tony Tebby, working on system design and software, recollects the process:

For instance, I was asked how long it would take to do the Microdrive software, and said four weeks’ actual working time once the work started. That went down in the minutes as ‘The Microdrive software would be complete within two weeks.’ First, the period was halved, and then it was changed from actual to elapsed time, by Nigel Searle. When the minutes went to everyone else, [and] in two weeks I haven’t done the Microdrive software because I’m doing something else, everyone who gets [the minutes] says I’m falling down on my job ... I think he thought he could make things run faster by generating external pressure. (Interview, 14 October 1984.)

Slippage in the timescale and the design flaws resulted from a combination of factors. Shortage of technical staff, forced and unforced changes in the design, problems with the technology, lack of management decisiveness allied with the divisiveness of tactics such as the above, all emerge from the story. As pressure to complete the work was exerted by management the problems got worse, with mutual recriminations the order of the day at the steering committee meetings.

Grandiose ideas of what could be achieved by as small a group as the Sinclair R&D team, however talented individually, led to the contemplation of an additional computer project in 1982. Sinclair Research was going not only to produce an all-new machine, the ZX83/QL, aimed at the business market, but it was also going to recycle bits of the hardware into a new machine for the Spectrum market. Although the Interface I and Interface II were still not ready, despite the efforts of Martin Brennan and Ben Cheese (Brennan had originally joined to do artificial intelligence work, but technical staff were so thin on the ground that he ended up doing logic design for the interfaces), plans were in hand for a 68008-based SuperSpectrum. This would have all the add-ons incorporated: ROM cartridge, joystick ports, microdrives, serial interface and network. A faster processor and more memory, plus a new BASIC, would give the loyal Spectrum freaks a machine to upgrade to, it was hoped, since the QL was supposed, in both price and concept, to avoid this area. Jan Jones was recruited to do the new BASIC, and the project started, including an enhanced keyboard design, which replaced the soggy rubber keypads of the Spectrum with hard plastic for a firmer feel. Then, at the end of 1982, the project was dropped, because the hardware design load was already overstretching the capacity of the limited technical team. The Spectrum market was also holding up well, and the Interfaces would give it a boost when they finally appeared, so there was no immediate need for a new hobbyist machine.