Sinclair and the 'Sunrise' Technology: The Deconstruction of a Myth (25 page)

At launch date minus a few weeks, the stance was still that there would not be an inbuilt version of BASIC on the QL, in line with the decision to go for a ‘business’, rather than a hobbyist, machine. The only set of instructions provided in ROM memory, apart from the QDOS operating system, would be a ‘bootstrap’ BASIC with no instructions other than those needed to enable the Psion software packages to be loaded and run. Jan Jones had been recruited for, and had started writing, an advanced and structured BASIC for the proposed, but now dropped, SuperSpectrum. When the Super Spectrum project got the chop, she continued to develop the BASIC in her own time, while working on other QL software for Sinclair. Since there was most of a 68008-based advanced form of BASIC in existence on the premises, the temptation was to use it. No Sinclair machine since the MK14 had been provided without an inbuilt BASIC, and the computer freak/hobbyist market made much use of BASIC (when they weren’t playing games), as was attested by the continuing popularity of BASIC program listings in the magazines. To commit to the business computer market and have no product for the mass of loyal Sinclair enthusiasts to upgrade to must have produced a distinct and disconcerting feeling of going out on a limb. The Sinclair success with computers was based on a certain type of machine: ones that had BASIC built in, could be plugged into a television, powered up, and be ready to go. To move into a new area of the computing arena, even at the bottom end of such a market, would require both an impeccable machine and a great deal of confidence in the product and the market. This latter would appear to have been lacking, while as yet the machine could hardly be said to have emerged as an operational reality.

It is only such considerations that can explain the gradual watering down of the QL concept in certain respects. Adding a television outlet was as much a response to this as the fact that Sinclair Research hadn’t produced a monitor to go with the QL, or arranged an OEM deal with someone who manufactured a monitor. The late decision to hedge the bets yet again and include a BASIC was not only a failure of nerve in the concept, but productive of more problems.

The operating system had been allocated two sockets on the board, each for a 16K ROM chip, into which the operating system and the ‘bootstrap’ loader for the Psion programs had to fit. The operating system commissioned from GST was not finished (the designers realistically couldn’t do much to finalize it until there was some definitive hardware) and in any case the OS was designed to occupy most of the 32K available. QDOS was more compact, and more complete, but with the unfinished Superb ASIC needing some 22K, there was no way in which the OS plus the BASIC would fit within 32K. Tony Tebby, on learning of the launch, promptly handed in his notice, to take effect as soon as working machines had been produced, on the grounds that the launch was misleading to the consumer:

There was never any possibility of launching a machine of which there was not a working prototype ... [the launch] was ... commercially foolish, and brought no benefits. (Interview, 14 October 1985.)

The QL manual handed out at the launch was a stop-gap construct leaning heavily on the Psion package’s documentation, since at least the user’s actions and their consequences could be described with some accuracy, even if they were not yet converted so that they actually worked on the QL. The SuperBASIC section of the manual was a confabulation of existing facilities, hoped-for additions and some straightforwardly inventive writing.

At this point, we arrive at the launch date. There are two views as to why the launch was allowed to go ahead so prematurely. One has it that the launch was designed to upstage the Apple Macintosh computer. As the Macintosh would predictably be a more expensive machine, and would not be competing directly with the QL, this explanation makes little sense as a marketing decision. Given that the Macintosh is, however, noticeably more innovative, flexible and user-friendly, as well as faster, than the QL, the need to get in ahead on the media attention may have played a part. Far more likely is the suggestion made by, among others, Guy Kewney in Personal Computer World that it was a desire to get some funds in before the end of the financial year (March 1984), to enhance the sales figures for the potential shareholders who would be invited to invest in British innovation on the basis of this year’s figures. Given the investment costs of the flat-screen television production line, and the continuing problems that prevented even the fairly sluggish demand for the £100 Microvisions being satisfied, plus cessation of the high-profit mail-order market for computers and the Spectrum price cuts, this would seem as good an explanation as any. Christmas sales were not as great as had been hoped, and Sinclair was holding £7m of stocks at the end of the year - none of it in QLs!

As up to 500 orders for the QL poured in each day, there was certainly no hesitation in cashing cheques, despite an absence of product. Indeed, it must have been clear that there was no hope of shipping before the end of the financial year in March, let alone honouring the ‘28 days delivery’ promise. That this was not only apparent, but recognized as a fact within Sinclair, is confirmed by Tony Tebby:

I discovered they were going to launch it about one week before Christmas. However, in the press release I was shown the day before the launch, it estimated delivery as ‘end April’. The press release that went out had ‘end February’ - a stroke of total idiocy. They said the ads said 28 days delivery, so the press release couldn’t say different. But, I said, it’s totally untrue, we don’t even have a complete working prototype! (ibid.)

Whether the over-optimistic promises to the public were rooted in the desire to help the faltering cash flow or not, this was the result. The cash figures for the year end were a massive £8.5m, and apparently included the £5.5m ‘trust fund’ set up by Sinclair (after the end of the financial year) to hold the mail-orderers’ money when the flak started to fly in the press. However useful in accounting terms, the publicity subsequent to the initial euphoria of the ‘launch’ was uniformly negative.

The lack of euphoria within Sinclair Research itself can be illustrated by a story that subsequently emerged. Your Spectrum magazine, in its December 1984 issue, asked Nigel Searle what he would like as a Christmas present:

What I’d really like to have is the name of the person who sabotaged my chair at the QL launch. Sitting in my cushioned chair waiting for Clive to finish his introduction so that I could kick off the proceedings I became aware that the chair was absolutely soaked. Someone had filled the cushions with a few gallons of water, so that it looked perfectly all right before you sat in it, but as soon as you did - well, need I say more? When I stood up to make my speech I had rivers of water pouring down my legs.

The world and the press waited in vain for QLs while the Sinclair team battled with ‘finalizing’ SuperBASIC and QDOS, getting the Microdrive interface working, and solving the logic errors in the custom chips, all while notionally gearing up for production. Inquiring journalists, anxious for review machines, were told that delays were caused by ‘development problems’. This was at least more honest than the letters sent out to all who had placed orders in response to the full-colour ad campaign in the Sunday supplements, which blamed the delays on the fact that, ‘The demand for the QL has been phenomenal from the day we launched it.’

This non sequitur posing as a ‘reason’ for not having produced a single QL was backed up by the offer of a ‘free gift’ for those who waited patiently while their money accrued interest for Sinclair Research. So those patient souls who didn’t demand their money back would receive a free RS232 cable. Normally priced at £9.95, it was repriced, as an added virtue, for the hopeful customers at £14.95 to increase their ardour. The complaints went to the ASA, and Sinclair itself changed the ads to say that delivery ‘may take longer than 28 days’.

The computer press, especially those having to fill QL supplements with notional reviews, rumours, speculative comparisons with other machines and details of the 68008 chip, was starting to complain. Dave Tebbutt, a friend of Sir Clive and a Mensa member among other things, got hold of one of the first working models, some time in April, and produced a positive review in Personal Computer World, despite the unfinished BASIC and QDOS. The less-privileged journalists were shipped off in the Sinclair Research black Mercedes to use four machines on show. The results were not what Sinclair had hoped from this PR exercise. Firstly, the fact that more than 32K of ROM was needed for the BASIC and operating system meant that hanging out of the back of the QLs was what came to be known as the ‘kludge’ - an extra 16K of storage sticking out of the ROM cartridge socket. Other aspects of the machine also proved disappointing:

The bad news is that QDOS and the bundled software’s current implementation is what one of Sinclair’s engineers described as ‘flakey’. Even basic operations like retrieving specific bytes from microdrives brought the system down. Several of the bugs thrown up in the session seemed new to Sinclair, and were noted with bemused interest. (Practical Computing, June 1984.)

In April, the first ‘production’ models (complete with kludge) were shipped to customers. Estimates of the numbers varied from 89 (curiously precise) to 1000, but were probably closer to the low figure. This enabled Sinclair to declare that it had started shipping machines. The press got the long-awaited review machines shortly after. Their gratitude at finally having something about which to churn out words at NUJ rates muted the criticisms somewhat, but the reception the QL got could still best be described as mixed. The keyboard, the most obvious part of the machine, was criticized by many:

The keyboard is not what it’s cracked up to be and don’t let anyone tell you otherwise ... The worst factor of all is the key action itself: squashy. Anyone using the wordprocessor will have very tired fingers and wrists after a day’s keying in. There is no springback on the keys and they have to be depressed a considerable distance to function. (Electronics and Computing Monthly, June 1984.)

The keyboard also belies the QL’s image as a business machine.

The keytops are expensive, classy items, but underneath there’s the same old membrane, and we found them unpleasant and difficult to use. (Personal Computer News, 26 May 1984.)

Others were less scathing:

Sinclair are pushing it to describe the keyboard as of ‘professional quality’. It’s certainly adequate for programming and so on but doesn’t compare with something like the BBC. (QL User, July 1984.)

Some reviewers even liked it, but it’s worth commenting that reviewers don’t have to do wordprocessing for a day at a time, as would a dedicated business user. Nobody of course rated it as highly as Sir Clive, since nobody else would feel driven to defend a dubious ‘innovation’ as obsessively:

The mechanism inside the keyboard is an immense investment in tooling and is a very precise system ... We are very proud of the keyboard. (Personal Computer News, 16 June 1984.)

Emphasizing the investment in, rather than the utility of, a keyboard, especially since a better one could have been bought in, seems to miss the point. However, there were more cogent criticisms of the first ‘production’ QLs. To avoid the tedium of repetitive quotes, we’ll illustrate them by a summary that appeared in Your Computer in July 1984:

Those criticisms covered all aspects of the QL: it was slow, had an unfriendly editor, the microdrives were prone to lose files and data, there was no documentation other than for the Psion packages, the network would not allow integration of Spectrums, the RS232 interface had bugs in it, Microdrive files on a well-used cartridge would take an age to load, the keyboard felt a bit clattery with a sticking enter key, and so on.

The ‘and so on’ covers a multitude of computerish sins. We should perhaps note criticisms of the Psion wordprocessing package as being excruciatingly slow to rewrite the screen and of the microdrives as being very slow. All the above criticisms refer to version ‘FB’ of the QL BASIC software. (There was a whole sequence of them, issued and unissued, all identified by two-letter names. Rumour had it that these referred to the programmer’s initials, but in fact the first two were named after cab drivers’ initials, and subsequent ones after girls who worked at Sinclair Research.) The same reviewer went on to say that the newer (AH) version of the ROMs ‘
is Sinclair’s answer to most of the problems, but it does not present a cure for all the QL’s troubles, and cannot make any difference to the hardware faults
’.

Subsequently, this ‘final’ version was replaced by ‘JM’, and then by ’JS’ in production models. Sinclair got rid of the unsightly ‘kludge’ hanging out the back around version ‘AH’ by putting 32K EPROMs in one of the sockets inside the machine.

Actually, the hardware was going through some hard revision processes as well. By the time the ‘AH’ software arrived they were on issue 7 of the main board layout! There were firstly some relatively simple problems to resolve, such as the fact that the early machines had the high-frequency PAL television oscillator circuit right next to the head amplifier of Microdrive 1, effectively ensuring that this Microdrive was inoperative and you had to use Microdrive 2. There were interactional problems that appeared in the early machines, such as the fact that the act of turning on a Microdrive altered voltages in the circuit, and it started to oscillate. The result looked like a signal coming off the Microdrive head amplifier, as it stopped when the motor stopped, but was in fact garbage. This was solved by soldering a capacitor across the head amplifier. Gradual improvements were made - rerouting power-supply tracks, modifying the custom chips - and the machine improved over time. Some problems, such as the fact that QLs were supposed to be able to communicate over the network with Spectrums, could be solved just by ceasing to claim that it was possible.