Photography by Tony Hutchings
96 PCW MARCH 1987
If Sir Clive Sinclair's new computer works the way he plans it to work, then he's back in business. Once again, he's doing it on his own. Starting this month, Sinclair's new company, Cambridge Computer, will run mail-order adverts offering the Z88, a lapheld product to be avail- able in May. Sinclair expects a wait- ing list, despite the delay. The price of £200 and the weight of two pounds make this 'super- portable' 8-bit business system a breakthrough in lapheld computing. The closest on both points is the Tandy 102, which is twice the weight and priced at £299. Sinclair's machine has far more versatile software, with an integrated spreadsheet, editor, diary and calcu- lator which make it a widely useful design. It also includes BBC Basic. And it has vastly more memory capacity. With a theoretical max- imum size of 4Mbytes of memory, it's expandable - in one way at least - as far as people are likely to ex- pect during its lifetime. That is, at least, the theory. In order to prepare this report for re- lease at the same time as the machine was due to be launched, at the Which Computer? Show (17-20 February, NEC), we had to take a lot of the detail on trust from the de- velopment team. What I actually saw of the machine itself was a wood model. The soft- ware was working and (very largely) debugged, but it was running on a Z80 second processor attached to a BBC Master. Some segments of it were running on separate machines, not integrated (there wasn't room) and the memory expansion system obviously requires the kind of elec- tronics that the BBC second proces- sor doesn't include. For these reasons, although every attempt has been made to put the machine through its paces in accord- ance with standard PCW Benchtest requirements, there wasn't the slight- est point in running any Benchmark programs. The hardware of the final lapheld will run at vastly different speeds, and the designers reckoned that any estimate of overall perform- ance - based on the BBC second processor - would be out by a fac- tor of five or more. Nonetheless, interest in Sinclair in the UK - and the world - is such that it was obviously necessary to satisfy curiosity by producing an analysis of the machine, designed to coincide with its launch. For that reason, I present here a preview, rather than a Benchtest, with the objective of updating it with perform- ance figures and a full review as soon as a finished machine can be obtained. Technical overviewThe standard lapheld machine has a |
Z80 central processor chip, with 128k of permanent programs and operat- ing software In ROM, plus 32k of RAM. Inside, there are four major custom VLSI chips handling all logic connections and I/O interfaces, plus a small loudspeaker. Display is an 8-line by 100 charac- ters wide liquid crystal unit from Epson. The keyboard is full type- writer style, slightly reduced in size to fit on the Z88's A4 shape. Around the edge are sockets for serial I/O, memory expansion and system expansion. And underneath, there's a slot for batteries. BackgroundThis machine is not just a flash in thepan, an opportunistic attempt to sell another bit of electronic wizardry. This is the culmination of a dream. Sir Clive Sinclair has been trying to design the ultimate lapheld portable for years and years, starting before he left the National Enterprise Board which was running Sinclair Radionics. Students of history will know that the original Sinclair Research design, the ZX80, was derived from the New- Brain computer which the National Enterprise Board gave to Newbury to turn into reality. That was intended to be a portable battery-powered system. So was the ZX80. So too, when it was first conceived, was the QL. Indeed, according to some histo- rians, the Sinclair flat-screen TV was originally designed to be a low-cost display for a portable computer. The flat screen was not usable in a port- able computer. It was simply too small, but Sinclair took a very, very long time before he gave up this idea. I saw 'Pandora', a briefcase-sized prototype, about a year ago. It used the flat screen and magnified it by using a concave mirror. Holding your head uncomfortably steady in front of this box, for all the world as if watching an old-fashioned 'What the Butler Saw' peepshow on Brighton Pier, you saw green letters, 20 char- acters per line on 12 lines, floating eerily off at an infinite distance. The software was vaguely sup- posed to be CP/M and Spectrum compatible, and there was supposed to be some way of plugging it into a full-scale display. There were no disks, and the idea was to use an Astron card - credit card sized, with electrical connectors and embedded memory chips - instead. This idea has been retained, but only in outline. It has been modified in the actual machine because of the problems anticipated from the card's exposed electrical connectors. 'We wanted protected contacts,' said Sinclair, 'so we had to devise our own package for add-in chips.' The correct word to describe the original Pandora was 'disaster'. It |
was heavy, much too thick, and the display was unusable. The software was never finalised. Fortunately for the world, the costs of liquid crystal displays were dropping like lead on a neutron star. By the time Sinclair Re- search was sold to Amstrad, it was obvious to everyone that the econo- mic advantages of the flat screen were disappearing, and no-one would have to justify dropping the strange design. When he bought Sinclair Research, Alan Sugar also bought the rights to Pandora. I asked him if he was going to launch it. 'Have you seen it?' he asked shortly. 'Yes,' I said. 'Well, then.' Free from the need to justify years of hard work on the flat screen, Clive decided to buy an Epson 100-column LCD display and build his system around that. His two main objectives were price and weight. 'Computers at present are a compromise,' he said when describing this new machine. 'They are either not portable, or not comprehensive. This one is highly portable, and it should do everything you want a personal computer to do. That means you have to do a rethink - you can't put in disks because that slows you down and increases the weight.' HardwareThe machine is the size of an A4sheet of paper, and at 2cm is pretty thin. It has a standard qwerty keyboard, described by Sinclair him- self as AT standard for reasons that have more to do with wish than ful- filment. However, he's right about one thing: a portable has to have a quiet keyboard. With four or five people in a lecture clattering away on standard keyboards, the device would quickly be banned from public events. Unfortunately I didn't get to see the prototype or pre-production keyboard, so I can't tell you whether this objective has been reached. Nor do I know if anything was sacrificed to provide it. Above the keyboard is a display made from 'super-twist' high- contrast liquid crystal. This is not backlit, but is very readable in nor- mal light. It is divided into four win- dows, with the central 80-column window providing a conventional character display. On the left is a menu window of 106 characters; on the right is a map of the whole page, showing 64 lines of text. One pixel represents each character of text. Further to the right, the last window shows machine status. Underneath the space bar, under a smoked acrylic panel (at least, that was the plan) are three slots. 'These give the function that disk drives nor- mally give,' said Sinclair. Well, sort of: they are actually memory expan- sion, but since you can plug in permanent memory and erasable |
MARCH 1987 PCW 97
permanent memory as well as RAM, you can think of them as being vaguely disk-like if you prefer. The 'disks' themselves measure .2cm square by 1cm deep. Normally such tightly packed memory chips would overheat, but at the low pow- er consumption of this machine, I don't expect that kind of problem. Under the display there's a socket for an RS232 plug. There is no para- llel printer socket, so the RS232 must be used for modem communications and printer operation; obviously this rules out the idea of downloading text from a bulletin board direct to the printer. This port is also used for a pseudo-disk operation called PC Link On one side there's an expansion plug, providing third-party builders with the opportunity to build odds and ends like parallel printer drivers, mouse drivers, microdrive interfaces and other whimsies. A very hard-to- locate reset button is buried deep in the heart of the machine, reachable only with a strong, thin piece of wire. (Sinclair really doesn't want you to restart the machine once it is going.) Inside are the Z80 processor, the custom logic chips, a minimum of 32k of static CMOS memory, a tiny loudspeaker, and a big-capacity one- Farad capacitor to keep the machine going for two minutes, when you change batteries. Underneath is the socket for four pencell batteries, pro- viding 20 hours work. That doesn't mean you have to change them ev- ery day - you have to use the Z88 for 20 hours. SoftwareThe main program was devised by aCambridge outfit called Protech, under the name Pipedream. it is a spreadsheet with rows and columns usable as text pages. Underneath Pipedream, or perhaps side by side with it, are other utilities, all built into 128k. They include a diary/ calendar, a note storage (database) system, and a programming lan- guage - BBC Basic. They will also include a crucial extra - PC Link. Extra programs can be loaded physi- cally by plugging in cartridges con- taining the code. Here we come to two unknown factors. One is the operating system; and the other is the problem of keep- ing it going when people plug things in. First is the question of why the system runs other programs. Until independent software houses have seen the machine operating system (MOS) and attempted to create their own programs to work with it, there is no way of knowing how powerful it is. Nor is there any way of knowing how reliable it will be - and it really |
|
98 PCW MARCH 1987
do want to see the thing working. As far as I know, no-one has done any- thing quite like it before. Most machines warn you, very earnestly, to turn off the power before plugging in cards or devices. Most of us have tried saving time by plugging in car- tridges and mice and joysticks in the middle of a program, and have had reason to regret it. The closest I've come to the Sinc- lair approach is using my Tandy 102 lapheld, which has an off switch but also has an automatic switch-off function. If you let the Model 102 turn itself off, it will turn on exactly where it was when you left it. But if you operate the main switch, it starts up in the operating system with a menu of functions and files. From experience, I can tell you that the machine is quite tolerant about having things plugged into it - prin- ters, modems, and so on. But I can also say that one time in ten it locks up solid, getting a power surge or flicker from the plug to the printer, which confuses it. And any PC user will tell you of extraordinary effects on printers caused by plugging the cable in after printing has started. Sinclair's design team is fairly con- fident that they have thought of all the possible disasters that can be caused by having someone plug in a card while the machine is operating, but they hadn't tested it at review time, and it's just hope and guess- work which suggests that it will work. But when the machine is sleep- ing, or when it is running an applica- tion, all other applications go into stasis. They cease operation, they do |
not run, and you take things up from where you left them, exactly. PipedreamThe key to the operation of the wordprocessor/spreadsheet is the screen map on the right of the display. No- one in the development team pre- tends you can really do useful work if all you can see of your page is eight lines of 80 characters. It would be like trying to read your newspaper through a magnifying glass - better than no newspaper, but no way to search for interesting things to read. There isn't any way of getting a big enough LCD screen to show more without using up far too much pow- er. Nor is there space for a bigger display. And smaller characters, even if desirable, aren't possible - liquid crystals have a minimum size. Sinclair's solution is to give a 'map' of the whole page, showing your eight lines (as on the main dis- play) highlighted. All other text and spreadsheet information is shown, unreadably, with one pixel for each character. Amazingly, this gives a very helpful idea of where you are in editing or working. Apparently (according to researchers at Triumph Adler, where I first saw a similar no- tion) the human recognition ability finds the shape of a document memorable, and letters can be filed and recovered by their appearance without the words being readable. As a spreadsheet, Pipedream looks pretty powerful. In many ways it looks very like Lotus 1-2-3, add the designers say, modestly enough, that it has several enhancements over 1- |
2 3, It can, they add, load 1-2-3 data files and manipulate them, but obviously there are lmitations to this compatibility. On the other hand, they also say they don't have Lotus, don't use it, and haven't tried to im- itate it. As a text editor, Pipedream uses the same concept that Lotus intro- duced of having spreadsheet cells which are infinitely stretchable. This gives an important new feature which few text editors have - the ability to work in columns. It also makes it possible to put spreadsheet data right into your notes, letters, re- ports, and so on. For example, by setting the first cell as being 35 col- umns wide for wordwrap, and the second cell, next to it, you can have two columns of text, newsprint style, side by side. Underneath that, a cell can have a total, or a sub-total, based on tabular information in cells above, and which will change as you change the information from which it is derived. At this point, we pass into the realms of vagueness again: to oper- ate this powerful piece of software, you obviously need powerful com- mands. How readily people will adapt to the ones provided is a mat- ter of pure guesswork, and there wasn't sufficient time for me to try them out and become accustomed to them. In outline they can be do de- scribed, however. Normally a menu button will pro- vide the menu appropriate to what you are doing; then a selection but- ton scrolls through the items an the menu. And for complex functions, |
In perspectiveWith the machine working only in simulation form,there were, nonetheless, many things which could be assessed and about which I'm happy. But there are many, many question-marks left hanging over the machine until we can get hold of a final version (not until April or May, if things go according to plan). The question-marks are these. The display shows only eight lines. Is this enough for serious business use? Sinclair's plans suggest that he expects people to use this machine not just for note taking, but for all their data processing, so the answer is important. There is no disk drive, just plug-in RAM and ROM. Will directors sitting around boardroom tables really swap data by connecting cables between machines? Or will they demand some kind of universal medium? Or will they really be happy passing little chip modules around? There Is no way of predicting the answer to this. Not only is this an 8-bit system, and therefore incom- patible with IBM or Atari or Commodore or any other 16-bit system, but it is incompatible with any other 8-bit system. It will almost certainly be impossible to run CP/M on the Dozy. Is this machine really so fast and wonderful that we will all throw away our old software and switch standards? This machine is designed never to be switched off. It is also designed to be expanded by plugging items into |
it. Plugging hardware into a running system is always a gamble, and the electrical hiccups that will result can do strange things to software and data. Only lengthy testing with real machines and real software from real software houses can answer this one. Will it work? If so, when? Is Sir Clive Sinclair having sold Sinclair Research, including Pandora, to Alan Sugar, entitled to build a lapheld computer? All these question-marks have been considered by Sinclair's designers, and some attempt at a solution has been sought. Most of them are ingenious. The only solution which I'm not happy about is the display, and even that is better than you'd think because of its clever windowing. But the final point is one that no-one can answer. At the time of writing, in early January, the machine didn't exist except as a simulation on a Z80 second processor attached to a BBC Micro. The software was tested in simulation only. The hardware was still In final design stages, and not functioning at all. Apart from the appalling (but well-selling) Psion Organiser, no portable computer approaches this one in size, weight, price or function. The Psion approaches in price and beats it in size and weight, but I'm afraid I still can't take that seriously as a computer. At press time, there was no way of ensuring reliable operation of the Z88, nor of getting a real hands-on feel for the software. |
100 PCW MARCH 1987
sequences of characters have to be typed in; they're given on the menu. Well, we'll see. In the meantime, I'm delighted to be able to say that an awful lot of WordStar commands have survived, and WordStar-trained typists will be able to delete lines with control-Y in a delightfully famil- iar way. More important than all this is the fact that you can see, from the map on the right exactly where you are on the page, In principle I think the map is excellent, but I'm not happy about having only 80 columns. I know I can make notes on a screen of that size because the Tandy 102 has a screen of 8 lines add 40 char- actors. And people who use Word- Star with full Help and double spac- ing tend to have the same amount of text onscreen. But I can't easily read a document on that. As you adjust the text, the map changes. But in order to keep up with typing, the map doesn't try to redraw itself until you stop typing for at least a second. There's a simi- lar delay when reading. Experience will be the only proof of the theory that the map will make document reading - skimming through after writing something counts as reading - as easy as Sinc- lair's designers believe. The simula- tion I saw, on a BBC Micro, was un- fortunately too slow to give a real indication of what it might be like, and I'm going to reserve judgement, heavily. DiaryThe main reason why no-one hasever seriously used a computer diary is very simple: whenever you are in a situation where you are trying to set up an appointment, you tend not to have the computer with you. A key factor in the design of the Dozy Diary is the fact that the machine which contains it is so small and light, you might well take it with you, even on social visits. Tucked away in a briefcase, it will be there to inform you that you can't make the date for Thursday because you have to take the dog to the vet. The diary is nice, and doesn't do the daft things that lots of computer diaries do: you can make any appointment for any time, and set the alarm for any other time. And you can categorise appointments, and list all appointments for a certain category. The diary is actually just a free- form database, capable of holding a limited amount of ASCII text. The limits depend on how much RAM you have in the system. Coupled to it is a calendar which looks through the diary database for alarms, markers and times. If all it finds is gossip, it will ignore it. But if it finds that there is a time corresponding to the time held in its real-time clock, it will |
|
MARCH 1987 PCW 101
Back to Z88 Magazine Articles, Columns and Advertisements