Sunday, 28 August 2022

Dragon 32 (1982)

Introduced August 1982

By 1982 the home computer market in the UK was getting quite sophisticated with the BBC Micro, Sinclair ZX Spectrum and Commodore 64 all competing for attention. To compete with these three extremely capable systems you were going to need something very good indeed. The Dragon 32 was not that computer. Not by a long chalk. Yet somehow it managed to carve out a fairly respectable slice of the market for a couple of years, and it all started so promisingly.

Dragon 32

British toy firm Mettoy – manufacturer of Corgi Toys – had spotted that children were becoming increasingly interested in computers and decided to enter the market, creating a factory in Wales to build the Dragon. Mettoy knew a lot about marketing and distribution, and in particular it understood export markets. However, Mettoy got into technical difficulties and the Dragon Data business ended up under the control of the industrial giant GEC.

The Dragon 32 itself was based on a Motorola reference design and used their 6809E processor, rather than the more common Zilog Z80 or MOS 6502s that rivals used. The dragon wasn’t the only machine built to the same basic design – the TRS-80 Color Computer (CoCo) launched in the US two years earlier was very similar and was somewhat compatible when it came to software.

Making a sort-of-clone of a two-year old computer in 1982 – when technology was moving at a breath-taking rate – may not have been a great start, but the 6809E was a capable CPU, the machine was very well built and you could connect up joysticks, a printer and a decent monitor. RAM was 32KB, a so-so amount for the time (a later 64KB version, the Dragon 64 was launched not long after) and it had simple sound capabilities. The inbuilt Microsoft BASIC was pretty good to program, which was one of the main things people liked to do in those days. Software could be ported across from the CoCo with a few modifications.

Dragon 64 in use
Dragon 64 in use

On the more negative side – the graphics were terrible, especially when it came to the colour palettes. The Dragon was also incapable of displaying lowercase characters without modification, which limited its appeal as an educational or business computer, and you couldn’t easily mix text and graphics at the same time. Although the Dragon 32 was popular enough to have many best-selling games titles ported to it, the poor graphics meant that they didn’t look as good as games played on rival machines.

Overall it wasn’t a bad system, but it was up against more capable competition. It might have been a contender but by 1983 the home computer market was imploding, with an oversupply of systems, brutal price wars and a fragmented array of available systems that frankly needed shaking out. Dragon Data was one of the victims, going bust in 1984, but the assets being bought up by a Spanish company named Eurohard which sold the product line until 1987, when it too went bust.

Despite market failures, the Dragon 32 retains a following in the hobbyist market with many additional modifications including improved operating systems and peripherals, including modern add-ons such as memory card readers in lieu of tape or disk drives. Working systems can command prices in of a few hundred pounds, depending on condition and accessories.

Image credits:
Liftarn / Pixel8 via Wikimedia Commons – CC BY-SA 2.0
Rain Rabbit via Flickr - CC BY-NC 2.0

Wednesday, 24 August 2022

Commodore 64 (1982)

Released August 1982

This – ladies and gentlemen – is the big one as far as 8-bit computers go. The biggest-selling single model of computer of all time, and a system that had success worldwide and is still remembered fondly today. I give you… the Commodore 64.

Commodore 64 original "breadbin" case
Commodore 64 original "breadbin" case

At first glance, the C64 is difficult to tell apart from the previous year’s VIC-20 as it shipped in a near-identical case at first. Inside though this was a much more powerful machine, running on a MOS Technology 6510 CPU, essentially a custom version of the popular 6502. The “64” in the Commodore 64 name comes from the amount of available RAM. The C64 used clever paging techniques where the CPU can page between ROM and RAM and rearrange most of the computer’s internal memory map to maximise available memory. This sophisticated scheme gave programmers much more RAM to play with than the competition who mostly used a flat memory configuration where ROM and RAM had to share the same space.

Graphics were a huge improvement over the VIC-20, with 320 x 200 pixels in 16 colours plus sprites, controlled by the MOS VIC-II graphics processor. Another MOS chip, the 6581 sound generator, gave multichannel sound. There was a built-in joystick port. By default the C64 shipped with a tape drive, or you could add on an incredibly slow floppy disk or the IEEE 488 serial bus which also supported printing. The hardware was subject to constant revision which sometimes produced compatibility problems.

Software support was excellent, with around 10,000 titles produced during the lifetime of the machine. Initially some of this shipped on a ROM cartridge, but this had a limit of just 16Kb so eventually tape became more common for complex games. In terms of games, few platforms even game close to the C64.

Excellent software and hardware made it an attractive proposition, but Commodore were keen to make this as affordable as possible. The initial launch price of $595 continually dropped, reaching $300 by 1983 (with cheaper deals available if you shopped around), easily undercutting the Atari 400/800, Apple II and crucially the Texas Instruments TI-99/4A.

There was a lot of bad blood between Commodore and Texas Instruments... TI had nearly bankrupted Commodore in the 1970s during the pocket calculator wars. Commodore boss Jack Tramiel wanted revenge, firstly the low-end VIC-20 piled on the pressures and the price-cutting on the Commodore 64 forced Texas to sell their system at a huge loss in order to compete. Not only did this force Texas to crash out of the home computer market, but it also inadvertently started a huge shake-out in the home computer market too.

If you were a teenager in the UK at the time, you would probably have had endless playground arguments comparing the Commodore 64, Sinclair ZX Spectrum and BBC Microcomputer. The argument could never be won because – in retrospect – all three platforms were really good and had their own strengths and weaknesses… but try telling kids that.

Sales were strong throughout the 80s, but competition grew tougher. Commodore attempted to diversify the C64-based offerings, notably with the luggable Commodore SX-64 (the first colour portable computer), the wedgy Commodore 64C and Commodore 128 plus an unsuccessful attempt at a games console with the Commodore 64GS.

Commodore 64C in the "wedge" case
Commodore 64C in the "wedge" case

At least 12 million Commodore 64 units were shipped up until 1994, only stopping when Commodore folded that same year. Over 12 years of production, the C64 was a massively influential machine – even today. Modern clones such as The C64 carry the torch, or used systems can typically be had for a few hundred pounds. Alternatively there are software emulators available. There's no doubt that even 40 years after launch, the C64 still has its fans.

Image credits:
Evan-Amos via Wikimedia Commons – CC0
Bill Bertram via Wikimedia Commons - CC BY-SA 2.5

Saturday, 16 July 2022

Grundy NewBrain (1982)

Launched July 1982

Largely forgotten today and not even very well remembered at the time, the Grundy NewBrain is one of those microcomputers that could have been a contender in the early 1980s personal computer market.

A compact Z80-based machine, the NewBrain featured exceptionally accurate floating point numbers and very high resolution monochrome graphics, which made it attractive to scientists and engineers. It could output to a monitor and TV, and interestingly most models sold had a 16 character display built into the case itself. Internal memory was split between 32KB of ROM and 32KB of RAM, a typical configuration. Additional paged memory could be added in 64KB blocks, theoretically giving a maximum of 2 megabytes. Expansion options included printers, disk drives and pretty much everything you’d expect for a microcomputer of this era. A portable version was also produced, utilising the inbuilt display plus a battery. The compact size of the NewBrain was due in part to a complex multi-layered motherboard that you tinkered with at your peril.

Grundy NewBrain
Grundy NewBrain

The hardware is pretty interesting, but the story of the development and eventual demise of the NewBrain is a slice of 1980s technology drama. Originally, the NewBrain was a project at Sinclair Radionics who were looking for a low-cost competitor to the Apple II. However, Sinclair Radionics were looking at a sub-£100 machine and the NewBrain was never going to be that cheap to build. Instead of going forward with the NewBrain, Clive Sinclair instead developed the ZX80 under his other company, Science of Cambridge.

Sinclair Radionics found itself in financial difficulties. This original Sinclair company had developed small radio sets and pocket calculators, but the money ran out and Radionics was rescued by the National Enterprise Board (NEB) who transferred the NewBrain to another NEB-owned company, Newbury Labs.

About this same time, the BBC was starting work on its computer literacy project, which would involve partnering with a manufacturer to create the BBC Microcomputer. The BBC was steered in the direction of the NEB-owned NewBrain which certainly ticked most of the boxes. It should have been a done deal, but when the BBC came calling the NewBrain wasn’t ready… and rival manufacturers had gotten wind of the BBC Micro and had insisted that they be allowed to tender. In the end, Acorn won the tender and their version of the BBC Microcomputer was born.

Grundy NewBrains with and without integrated displays
Grundy NewBrains with and without integrated displays

So, the NewBrain missed out on being both a Sinclair machine and a BBC Micro. In the end it ended up with a rather obscure company called Grundy Business Systems, who Newbury Labs sold the design to. It wasn’t an immediate market success, but it looked promising. So promising in fact that Grundy built a lot of them… but the hoped-for sales didn’t appear and by 1983 Grundy was in serious trouble. Essentially by 1983 it was all over, most remaining stocks were liquidated and the NewBrain ended up as a casualty of the early 1980s microcomputer crash.

Although it was a limited success in the UK, it was rather more successful in the Netherlands, Denmark, Greece and – for some reason – Angola. Had it been ready when the BBC were interested then it might have been the first of a series of machines, but in the end the NewBrain’s potential was never realised.

Today these are highly collectible machines, with working systems often commanding prices of £1000 or more. Alternatively, if you are a former NewBrain owner and want to rekindle old memories, then an emulator is available.

Image credits:
Rama & Musée Bolo via Wikimedia Commons - CC BY-SA 2.0 FR
Marcin Wichary via Flickr - CC BY 2.0

Thursday, 23 June 2022

Jaguar XJ220 vs McLaren F1 (1992)

Launched 1992

If you wanted a really fast British supercar in 1992 and you have about half a million pounds in your pocket, you had an intriguing choice between the Jaguar XJ220 and the McLaren F1. Thirty years later, one of these cars is considered to be a success and one a relative failure. But which is which?

Let’s start with the Jag. By 1992, Jaguar was owned by Ford but had spent the previous few years struggling with a range of increasingly elderly cars. However, a successful foray into racing (largely thanks to TWR) had resulted in a supercar project… not just any car, but a street-legal machine capable of hitting 200 miles per hour.

The concept version of the car caused a shockwave. An all-wheel-drive sports car powered by a mighty 6.2L V12 engine mounted in the back, this version of the XJ220 also had scissor doors and the slippery design made it look like no other Jag. The “220” part of the name was the top speed that Jaguar was hoping for and despite the then eye-watering price tag of £470,000 there were 1500 people who put down a deposit.

Jaguar XJ220 - not your grandfather's Jag
Jaguar XJ220 - not your grandfather's Jag

Between concept and product though there were several changes. Perhaps the most significant was the engine. The V12 that Jaguar had proposed was big and heavy and also had problems meeting emissions standards, but Jaguar had ended up with the rights to the engine in the short-lived but legendary MG Metro 6R4 rally car. The 6R4 had a relatively lightweight V6 unit somewhat inspired by the (also) legendary Rover V8. It was a promising engine for Rover, but Group B rallying was banned in 1986 after a series of accidents, and the 6R4 and its engine became essentially redundant. Jaguar took the 6R4 V6 and thoroughly reworked it, adding twin turbos in the process, giving about 542 horsepower. It was arguably a better engine than the V12, but people were expecting a V12 and not a V6. In addition, one other major change were the door – the scissor doors were dropped in favour of more conventional ones, and the car moved to a simpler rear-wheel-drive configuration. Many customers were very unhappy, and these changes plus a recession in the early 1990s led to many cancellations.

The XJ220 was extremely aerodynamic, including the underneath of the car. Body panels and the chassis were made from aluminium. Advanced technologies could be found everywhere from the braking system to the transmission… Jaguar were not short changing customers on kit. Being a Jag, the inside was a lovely place to be. Top speed was around 212 MPH, not quite as much as the 220 in the name, but nonetheless blisteringly fast.

The XJ220 concept had a massive V12 engine, the production car a more compact V6
The XJ220 concept had a massive V12 engine, the production car a more compact V6

The bad points? Well, it was more than two metres wide and lacked power steering and ABS, so it wasn’t much fun as a daily driver. There was also limited luggage space (despite the huge size) making it impractical as a grand tourer as well.

It wasn’t a sales success – Jaguar had never intended it to be a high-volume car, but with just 281 built they fell short of their targets. It did help to raise Jaguar’s profile as a sports car manufacturer, but ultimately the XJ220 was a little too flawed and compromised. The XJ220 was in production for just two years – and Jaguar never made another production car that was anything like it afterwards.

At around the same time, McLaren were developing their first road car around similar themes. A bit more expensive than the XJ220 at £540,000 (in 1992 money), the McLaren F1 wasn’t saddled with the compromises that the Jaguar possessed. The F1 was powered with a normally aspirated (i.e. not turbocharged or supercharged) V12 like the XJ220 concept. McLaren chose the normally aspirated route for reasons of control and predictability – early 1990s turbochargers gave uneven power and suffered from turbo lag, so a normally aspirated engine was much smoother. McLaren shopped around for a suitable V12 eventually settling on a power plant made by BMW.

The McLaren F1 would look fast parked up in Sainsbury's
The McLaren F1 would look fast parked up in Sainsbury's

The body of the F1 made extensive use of carbon fibre, except for the engine bay which has gold foil acting as a heat shield. A combination of other lightweight and strong materials are found throughout the car, including magnesium, Kevlar and titanium. The whole body shape produces downforce rather than having a fat spoiler, but one clever trick was the introduction of two fans in the base of the car with both produced extra downforce and cooling at the same time. The top speed? The McLaren F1 was (and still is) the world’s fasted normally aspirated car with a top speed of 240 MPH.


The McLaren F1's three seats practically make it an MPV
The McLaren F1's three seats practically make it an MPV

Inside the F1 is highly unusual, featuring three seats – the driver sits in the centre and slightly in front of the two passengers either side in the rear. Entrance to the cabin was through the dihedral (scissor-like) doors, something the XJ220 sorely lacked. Luggage compartments are hidden around the car, although best used with the proprietary matching bags. The F1 also included air conditioning and a number of other aids to make it usable on the roads, transforming the F1 into an almost practical grand tourer as well as a sports car. It wasn’t designed as a track car, but it was pretty good at that two with race variants such as the F1 GTR being made. 


The McLaren F1 looks purposeful from the back as it disappears over the horizon
The McLaren F1 looks purposeful from the back as it disappears over the horizon

However, despite the advanced engineering, only 106 cars were produced (including prototypes). McLaren did turn a modest profit on the F1 during the six years of production, ending in 1998. McLaren didn’t build another road car until 2011. Nonetheless, the F1 was an engineering success and it didn’t make the compromises that the XJ220 did.

Today a McLaren F1 is worth around £16 million, but an XJ220 is only worth about £450,000 – about the same as it was new in actual pounds, but adjusted for inflation the XJ220 cost around £1 million when new. According to collectors at least, the F1 is a far more desirable car.

Image credits:
Vauxford via Wikimedia Commons - CC BY-SA 4.0
Morio via Wikimedia Commons - CC BY-SA 3.0
Jaguar Cars MENA via Flickr - CC BY 2.0
Craig James via Wikimedia Commons - CC BY-SA 4.0
Ank Kumar via Wikimedia Commons - CC BY-SA 4.0
Neilhooting via Flickr - CC BY 2.0

Sunday, 12 June 2022

Columbia Data Products MPC 1600

Introduced June 1982

No wait. Don’t go. The MPC 1600 is a hugely important milestone in computing, just one you may not have heard of. Let me explain.

Columbia Data Products MPC 1600
Columbia Data Products MPC 1600

August 1981 saw the launch of the IBM PC into the fast-growing microcomputer marketplace. It wasn’t the most advanced microcomputer on the market, but it did have the magic letters “IBM” on it which made it attractive to corporate buyers.

Unlike other IBM products, the PC was made largely of off-the-shelf components that anyone could buy. IBM had also documented everything in painstaking detail in order to attract third-party developers to create hardware and software for the new platform. Theoretically anyone could build a machine like the IBM PC except for one major component… the BIOS.

The BIOS is an oft-forgotten part of the PC. Lying somewhere between hardware and software in the layer known as “firmware”, the BIOS provides the most basic software functions that a PC relies on. Unlike most of the rest of the IBM PC, the BIOS was strictly proprietary. However, developers needed to understand how that BIOS worked, so IBM provided full specification of the functionality. Not enough to clone the BIOS… or so they thought.

So when Columbia Data Products (or CDP) wanted to make a machine just like the IBM PC but better value, the BIOS was an obstacle. However, IBM had published the full BIOS specifications (but not the code) to help developers, CDP took the specifications and created a clean room design of the BIOS which replicated the functionality but used none of the code.

1982 ad for the MPC 1600
1982 ad for the MPC 1600 with funky Lear Siegler terminals

When launched in 1982, the Columbia Data Products MPC 1600 was about half the price of the IBM, but had more memory, more built-in features and more expansion. It was a quality machine in both terms of hardware and the 100% compatilibity with the genuine IBM PC, usually measured in those days by being able to run Microsoft Flight Simulator. For people who wanted an IBM PC but didn’t want to pay IBM prices, it was an attractive deal.

CDP’s sales grew quickly and expanded their range, but the problem was that they weren’t the only players in the market. Other firms joined the fray, usually competing on price and squeezing the very thin margins the clone makers had even further. Initial success gave way to red ink, and by 1985 CDP was bankrupt. However, that wasn’t the end for CDP and subsequent rescue led to a change of emphasis, and Columbia Data Products still exists today making data backup products.

Today, the chances are that the computer you use is a PC clone. It was always likely that IBM would create a beast that it couldn’t control and that clones would take over, so even if Columbia Data hadn’t been the first it would likely be someone else. But the fact remains that they were the first…

Image credits:
Ben Franske via Wikimedia Commons - CC BY-SA 4.0
PC Magazine, November 1982

Sunday, 22 May 2022

Sun-1 (1982)

Available May 1982

If you wanted to put a computer on your desk in 1982, there were a wide variety of choices. Businesses might go a system like the IBM PC or Victor 9000, home users might go for something like a VIC 20 or TI-99/4A. If you had more exotic requirements there were high-end devices such as the GRiD Compass or Xerox Star. The Sun-1 workstation – first shipping in May 1982 – fell firmly into the last category, putting minicomputer power in the hands of the individual.

Sun-1 Workstation
Sun-1 Workstation

The Sun-1 was the first commercial product of Sun Microsystems, which had grown out of a workstation project started at Stanford University – the name “SUN” was derived from “Stanford University Network”. The original series of Sun workstations were built for on-campus use only, but the Sun-1 took that experience and turned it into a commercial product.

Designed to be powerful enough to run UNIX or other multitasking OSes, the CPU was the surprisingly modest Motorola 68000 coupled with 256KB of RAM out of the box, which was upgradeable to 2MB. Custom Sun silicon enabled the CPU to reliably support multitasking, the 1024 x 800 pixel graphics also had hardware acceleration. The standard display was a 17” CRT although other options were available. Although it was designed as a single-user computer, you could hook up to two text terminals to the back to use it as a small-scale minicomputer.

Expansion options were comprehensive including Ethernet, mass storage and other peripherals. In a rackmount version the Sun-1 made a capable server, but its real home was sitting on a desk or in a lab where all the power could be used by just one person. It wasn’t cheap of course, starting at $8900 at 1982 prices (around $25,000 today) so it was limited to those organisations that had the budget and the need for that much computing power.

Sun-1 Workstation
Another Sun-1 Workstation

As a product it was still a little rough around the edges, but a year and a half later the Sun-2 came out with both improved internal hardware and a more professional external design. The Sun-2 and the Sun-3 (launched in 1985) established Sun Microsystems as the player to beat in the workstation market.

Sun itself thrived until 2001 when it was badly hit by the collapse of the dot-com bubble, and the following years were dominated by red ink in the balance books, caused in part by more powerful Intel-based machines running Windows and Linux which could outperform and undercut Sun's products at the same time. In 2009 Sun were bought out by Oracle, and although Oracle still sells servers based on Sun architecture you probably wouldn't know it. Oracle - after all - has a reputation of where good products go to die.

Image credits:
Richard Masoner / Cyclelicious via Flickr - CC BY-SA 2.0
Carlo Nardone via Flickr - CC BY-SA 2.0

Saturday, 7 May 2022

Orbitel TPU 901 (1992)

Launched May 1992

Early mobile phones were terrible things. Not only were they big and clunky, but the old analogue networks that they ran on had terrible call quality, poor reliability and were very insecure. These early technologies such as AMPS, TACS and NMT became retrospectively known as “1G” – these days often forgotten and unloved.

By 1992 these 1G networks had been around for a decade or so and their weaknesses were becoming obvious. The market was ripe for something better, and in 1992 the world’s first 2G GSM networks came online. These digital networks had better call quality, security and required a smaller slice of the radio spectrum, and the first certified GSM phone to be available was the Orbitel TPU 901.

Orbitel TPU 901

A bulky device even by the standards of the time, the 901 had a handset connected to the base station via a curly cord and it weighed a whopping 2.1 kilos. It wasn’t a big seller – smaller and cheaper GSM phones were not far off – but the Orbitel TPU 901 does have the distinction of receiving the world’s first SMS text message with the words “Merry Christmas” sent in December the same year.

Orbitel was a British-based joint venture between Racal (who owned Vodafone) and Plessey which eventually ended up in the hands of Ericsson and effectively vanished in the noughties. Today the TPU 901 (and the car-mounted TPU 900) should still work on 900MHz GSM networks, if you ever managed to get your hands on one.

Orbitel TPU 901
Orbitel TPU 901

Of course, the 901 was the first of many GSM phones on the market, more memorably the Motorola International 3200 launched later in 1992 with a memorable brick-like design that summed up the era perfectly. About a million others followed, but the Orbitel TPU 901 – largely forgotten today – was the very first.

Image credits:
Science Museum Group - CC BY-NC-SA 4.0
[1] [2]

Thursday, 28 April 2022

Raspberry Pi (2012)

Available April 2012

Single board computers were common in the early days of microcomputers, with the KIM-1 offering a relatively low-cost way of playing with the then-new 6502 CPU and later devices such as the Acorn System 1 made it cheaper still. But single board computers appealed most to hobbyists, and as technology developed so did microcomputers, eventually evolving into complete systems that were easier for novices to use.

Original Raspberry Pi Model B
Original Raspberry Pi Model B

As the decades rolled on, the amount of computing power that could be squeezed into a board computer grew. First came Arduino, a series of open source board computers that could be used for microcontrollers. A few years later, TI came up with the BeagleBoard which was a general purpose computer on a single board. But perhaps the best know modern single board computer is the Raspberry Pi, shipping to customers in April 2012.

Unlike some other designs, the Pi was a complete system on a compact board. With built-in USB, video and networking ports all that was required was a memory card with an operating system and a monitor, keyboard, mouse and power supply. These are all pretty common peripherals, and in most cases Pi users could just re-purpose old equipment used elsewhere. The Pi didn’t come with a case so a cottage industry started up making them, all of this echoing the rather do-it-yourself approach of the original Apple I.

The first Raspberry Pi models were announced in February 2012, coming to market in April the same year. Like the BBC Micro, there were two launch models of the Pi – A and B. B was the most popular, based around a Broadcom chipset that included an ARM CPU, RAM and all of the other silicon needed on a single chip. But perhaps the biggest breakthrough was the price – this complete computer system cost just $25 or the local equivalent for the simplest model.

Coincidentally, the ARM CPU in the Pi was originally designed by Acorn, whose experience with the 6502 (starting with the Acorn System 1 board computer) inspired them to create an inexpensive, simple but very fast processor based on similar principles.

The target market was initially education – instead of expensive laptops, students could simply plug their own Pi into a PSU, monitor, network socket, mouse and keyboard and do whatever they wanted with it. The easily swappable memory card meant that different configurations could be experimented with easily. But the appeal turned out to be far greater, everyone from hobbyists to engineers wanted to play with one and the Pi became a significant success. Raspberry Pi devices can be seen in almost any application from controllers to servers, often performing tasks as well as machines costing hundreds of times as much.

A decade on, the Raspberry Pi is still going strong. Later models offered more ports, a faster processor and more memory and even cheaper models such as the Pi Zero and Pi Pico slotted into the range below the fully-featured Pi. A wide range of peripherals are available for almost any application, and OS support has grown from Linux-only to include Windows 10 IoT and even a version of RISC OS (originally designed for the very first ARM-based computer, the Archimedes).

Raspberry Pi emulating a DEC PDP-8 and PDP-11
Raspberry Pi emulating a DEC PDP-8 and PDP-11

Millions of devices and a decade later, the Pi has proved to be an antidote to the anodyne world of modern personal computing. The Pi helped to re-ignite some of the early hacker ethic of early micros and taught a new generation that what they could do with a computer was only limited by their imagination. Not too shabby for just $25.

Image credits:
osde8info via Flickr – CC BY-SA 2.0
Wolfgang Stief via Flickr – CC0

Saturday, 23 April 2022

GRiD Compass (1982)

Released April 1982

Even though practical microcomputers had only been around for a few years by 1982, there was a growing market for portable devices such as the Kaypro II which offered all the computing power you probably needed in a luggable package.

Back then people accepted that a portable computer would weigh something like 13 kg and come in a huge case. Practically speaking you’d typically carry it between a desk and car. Unlike modern “laptop” computers, most portables of the early 1980s would possibly break your knees if you tried to use them on the sofa.

GRiD Compass
GRiD Compass

The first practical laptop computer is widely considered to be the GRiD Compass. A clamshell on the front of the device held a 320 x 240 pixel electroluminescent display and a keyboard in a format instantly recognisable today. Although the display was relatively small, it was sharp and clear compared to early LCD panels and the limited resolution was actually pretty competitive with most computers of the time.

Inside was an Intel 8086 CPU with an 8087 maths coprocessor, but this was no DOS-compatible computer. Instead the Compass ran a proprietary OS called GRID-OS which was menu-driven and quite friendly. One novelty was storage – the Compass used magnetic bubble memory giving 340Kb of non-volatile storage. Most production systems also included a modem, and an IEEE interface bus was standard. The lightweight but strong magnesium alloy case contributed to the relatively light weight of around 5 kg.

This was a highly advanced machine, and it came with a substantial price tag starting at $8500 in 1982 money which is around $25,000 today. OK, it is possible to spend more than that on a computer today (a high-end Mac Pro can cost $60,000 or more) but that was nearly six times the price of the Kaypro and to be honest it couldn’t do as much for a typical end user.

Where it did find a niche was in government sales. The tough but lightweight design lent itself well to military applications, and the Compass was also certified for use on board the Space Shuttle. Large corporations were drawn to it as a practical and highly portable device, but few found their way to private users due to the high price.

GRiD Compass running a spreadsheet
GRiD Compass running a spreadsheet

This was the first in line of several GRiD systems, and on top of healthy sales they also owned a patent for several of the elements of the clamshell design, meaning that other laptop manufacturers had to pay GRiD a fee for each system built. GRiD was taken over by Tandy in 1988 followed by a management buyout in 1993 which moved the company from California to the UK. The company – now called GRiD Defence Systems – still makes ruggedized laptops and other hardware.

The Compass set the pattern for all modern laptop designs, years before they became commonplace. Today first-generation GRiD Compass systems are very rare and you can expect to pay between £5000 to £10000 for a working system.

Image credits:
Cooper Hewitt, Smithsonian Design Museum
Niall Kennedy via Flickr - CC BY-NC 2.0

Thursday, 14 April 2022

Sinclair ZX Spectrum (1982)

Introduced April 1982

If you were a British child of the 1980s, the chances were that you possessed one of the holy trinity of the BBC Micro, Commodore 64 or the Sinclair ZX Spectrum. A rivalry leading to many playground arguments, these three machines duked it out for years with no clear winner.

Sinclair ZX Spectrum

Out of the three, the cheapest and most popular (for a while) was the Sinclair ZX Spectrum. Sinclair’s follow-on to the ultra-low-cost ZX81 launched the year before, the Spectrum added rudimentary but usable colour, graphics and sound in a package with either 16kB or more desirably 48Kb of RAM in a stylish package – all at a very attractive price.

Like the ZX81, the Spectrum was based on a Z80 processor. But where the ZX81 struggled to do anything due to its clever-but-simple design, the Spectrum was highly competitive with the new generation of early 1980s home computers.

It wasn’t a big machine – roughly the size of a sheet of A5 paper and weighing around 550 grams – but Rick Dickinson’s industrial design consisting of a black case, grey keys and the 1980s-on-a-stick rainbow flash on the corner looked far more impressive than the competition. Those keys were something else though – each one performed up to six functions in the Spectrum’s capable BASIC environment, but the strange rubberiness of the keys felt like touching dead flesh.

The multifunction keys bear some examination. All the BASIC keywords were assigned to a key which would activate depending on context, or with the CAPS SHIFT and SYMBOL SHIFT keys. This layout was first seen on the ZX80 and while it reduced errors and made programming more accessible, it was becoming more fiddly as the version of BASIC evolved. The Spectrum’s version of BASIC was pretty sophisticated – not as good as the one in the BBC but better than the Commodore 64. Budding programmers took to the Spectrum and coded furiously from their bedrooms.

As standard the Spectrum loaded and save programs to a cassette, which was quite slow. Video output was to a domestic TV set, so the Spectrum could easily plug into what you probably already had in the house. The desirable 48Kb version cost just £175 at the time (equivalent to around £650 today) but you really didn’t need anything else if you had a TV and cassette recorder.

Like the BBC, the Spectrum could address only 64Kb of memory. The ROM was simpler than the BBC, taking up just 16Kb which left up to 48Kb of RAM available. The Spectrum’s curious colour graphics mode didn’t eat up much memory either, meaning that there was quite a decent amount of RAM available for programs, something that the BBC struggled with.

The colour graphics were rather strange. The 256 x 192 pixel resolution could display up to 15 colours, but you could only have one foreground (INK) and one background (PAPER) could in each 32x24 pixel character grid. This made it tricky to code colour games (for example) but it was very memory efficient. Sound output was fairly simple with a one channel output, but it was good enough for most purposes.

Like the ZX81 and ZX80, and edge connector on the back of the machine allowed access to pretty much all hardware functions. Sinclair’s official accessories on launch included a tiny thermal printer and the ZX Microdrive, which was a high-speed tape cartridge which was plagued with delays. Popular third-party addons included the Kempston Micro Electronics joystick interface but also various adapters for disk drives, speech, serial and parallel ports and perhaps most important a variety of aftermarket keyboards that improved on the Spectrum’s unpleasant chicklet affair.

Spectrum with daisy-chained ZX Microdrives and sound enhancements
Spectrum with daisy-chained ZX Microdrives and sound enhancements

The Spectrum was an enormous success - the combination of pricing, features and the brand recognition of the “Sinclair” name were key factors. Success bred success with huge variety of games and other applications along with hardware enhancements coming to market. Few competitors had a fraction of the third-party support that the Spectrum did.

1982 and 1983 were probably the peak years for the home computer market in the UK. Sinclair found itself up against increasing competition from less well-known machines which were often better (though rarely cheaper). In 1984 the Spectrum+ was launched, essentially a 48K Spectrum in a Sinclair QL-style case. A 128Kb version dubbed the Spectrum 128 was launched the year after, using memory paging to break the 64Kb limit. In 1986 Sinclair found itself in difficulties and was bought by Amstrad who styled new models after their popular CPC range leading to the Spectrum +2 with an integrated cassette recorder in 1986 and the Spectrum +3 which included a built-in 3” floppy disk drive, launched in 1987. This +3 was the ultimate development of the Spectrum platform, capable of running CP/M but it wasn’t 100% hardware compatible with the original which caused problems. The last Spectrum models in production were the +2B and +3B which were basically hardware fixes of previous versions, production ended in 1992 giving the Spectrum platform an impressive ten year lifespan.

ZX Spectrum +3 with 128Kb RAM and a 3" floppy drive
ZX Spectrum +3 with 128Kb RAM and a 3" floppy drive

In addition to the official Sinclair version, licensed and unlicensed clones proliferated – notably licensed variants made Timex in the US and Europe, and a huge number of bootleg clones in Eastern Europe and South America into the 1990s. In the 2010s there were several attempts to recreate the Spectrum with modern technology, perhaps most significantly with the ZX Spectrum Next.

Despite the success of the Spectrum in the market, ultimately it was something of a dead end – even though fondness for the platform lingers on four decades later. However, the significance of the Spectrum was profound in the markets it succeeded in: this low-cost, easy-to-use and versatile device inspired a generation of programmers and computer enthusiasts, many of whom went on to carve careers out in the IT industry. This simple but effective machine not only help to shape lives, but also whole economies. Not bad for a cheap computer with a nasty rubber keyboard.

Image credits:
Bill Bertram via Wikimedia Commons – CC BY-SA 2.5
ccwoodcock via Wikimedia Commons – CC BY 2.0
ccwoodcock via Wikimedia Commons – CC BY 2.0

Thursday, 24 March 2022

Nokia 3510 (2002)

Introduced March 2002

It is sometimes said that there are only two things that would survive a nuclear war: cockroaches and old Nokia phones. The Nokia 3510 – launched in 2002 – has the potential to be something that a future cockroach civilisation would unearth and worship as some type of cockroach god.

Maybe in this future roach utopia the 3510 might find itself in an epic Godzilla vs Kong battle with the legendary Nokia 3310, but the 3510 takes the indestructible design of the 3310 including the funky changeable Xpress-on covers and adds polyphonic ringtones, bigger screen, GPRS and a WAP browser. Indeed, the 3510 was once of the very first popular consumer phones to offer GPRS.

Sober or funky, the Nokia 3510 had interchangeable covers
Sober or funky, the Nokia 3510 had interchangeable covers

Like the 3310, the 3510 could survive a direct strike from a Tsar Bomba – but it couldn’t quite wrestle the roach god crown from the 3310. But a few months later the 3510i arrived, adding not only a colour screen but also support for Java games. The bugs raised temples in its honour.

Although the 3510 (and 3510i) are primitive by modern standards, they form part of the golden age of mobile phone design where every new handset looked different and new technologies were being introduced at a rapid rate, here with colour screens and Java but also including cameras, Bluetooth, media players, expandable memory and so much more.

XpressOn covers allowed a high degree of personalisation with the 3510i
XpressOn covers allowed a high degree of personalisation with the 3510i

If you fancy a retro Nokia like this then you’ll be pleased to know that they are as cheap as chips, buy enough of them and you could even make a Nokia Stonehenge for future cockroach archaeologists to puzzle over.

Image credits: Nokia

Wednesday, 23 March 2022

Kaypro II (1982)

Introduced March 1982

A few years after the introduction of the first generation of practical microcomputers, manufacturers started to look at the possibility of having a computer that you could take anywhere rather than have tied to a desk. Perhaps the best known of this first generation of “luggable” microcomputers was the Osborne 1, launched in 1981.

The Osborne had plenty of limitations, not least the tiny screen. The market was ready for something better, and something better certainly turned up with the Kaypro II. It was designed by a company called (at the time) Non-Linear Systems (or just “NLS”), who up until this point had made lab equipment. NLS knew how to make reliable, rugged and portable electronics and they turned this expertise to a microcomputer.

A Kaypro II with several other vintage portables
A Kaypro II with several other vintage portables

Running the de-facto standard CP/M operating system and running on a 2.5MHz Z80 CPU with 64Kb of RAM and dual floppy disks, the Kaypro II had highly competitive specs for the era. Built into the high-quality painted aluminium case was a pin sharp 9 inch display which you could comfortably use all the time. The keyboard was also a high-quality design and it clipped firmly into place on the case making a practical if somewhat hefty 13 kilogram computer.

It wasn’t just the hardware that made the computer competitive. The Kaypro II was bundled with an office suite from Perfect Software that included a wordprocessor, spreadsheet and database, plus a version of BASIC that could be compiled into CP/M .com programs. Eventually the bundled programs included the class-leading WordStar wordprocessor and SuperCalc spreadsheet. The bundled software theoretically cost many hundreds of dollars, making the Kaypro II seem more of a bargain. If what you wanted wasn't in the box, then there was a wide range of business and home software available, including the legendary Zork.

Please give this Kaypro II a home
Please give this Kaypro II a home

This was a high-quality, reliable, well-designed and inexpensive product and it became quite a success. If you took one home from the store there really wasn’t much you needed to do except plug a few cables in. Everything else was in the box, making this an extremely consumer friendly-product.

It ended up as a popular tool with writers. Arthur C Clarke had one, as did Jerry Pournelle. Although it was always a bit of a niche system, the Kaypro II was enough of a success to make NLS a major player in the early 1980s market.

Curiously, there was never a Kaypro I as such. Aping the rival Apple II, NLS decided on making their first commercial machine another “II”. In 1983 it was followed by the improved Kaypro IV and Kaypro 10 (with a 10Mb hard disk), then in 1984 followed the Kaypro 4 and Kaypro 2X which were confusingly named, and if you thought that was confusing another Kaypro 2 followed in 1985 and the Kaypro 1 in 1986 (because why the heck not call it the “1”). All those Kaypro machines ran CP/M, it took until 1985 for Kaypro (as it changed its name to) to come up with a PC-compatible system with the Kaypro PC and the Kaypro 286i, which was the world’s first AT-compatible system. Kaypro also adapted their luggable boxes with the MS-DOS capable Kaypro 4 Plus 88 and Kaypro 16. Kaypro even managed an early MS-DOS laptop with the slightly peculiar Kaypro 2000 and the qurkly and deeply unreliable CP/M Kaypro Robie. Somewhere along the line NLS changed its name to “Kaypro” as well.

Kaypro went from boom to bust quite quickly. The II was launched after the IBM PC but before it became really popular. It took four years for Kaypro to come up with their own PC, by which time the rapidly-evolving market had moved on. Despite soldiering on for a few years, Kaypro went bankrupt in 1992.

As with almost all microcomputer companies of the era, the success of Kaypro was relatively short-lived. However, they were the first company to come up with a practical and affordable portable computer, furthermore one that required very little work to get it fully operational, with a high-quality build and smart looks. The Kaypro II undoubtedly influence other computer designs of the same and next generation.

Today the Kaypro CP/M portables are somewhat collectable, although many have become separated from the floppy disks that held the software. Most available systems are in the US but some can be found in Europe, but expect to pay several hundred pounds

Image credits:
Diaper via Flickr - CC BY 2.0
Daniel Boulet via Flickr - CC BY-NC 2.0

Sunday, 13 March 2022

LINC (1962)

First delivered in March 1962

The LINC – short for Laboratory INstrument Computer - was one of the world's very first minicomputers, helping to break the computer out of the corporate machine room and making it available to individual departments, labs and (at a stretch) homes.

Designed at MIT for academic work, most LINC machines were built by Digital Equipment Corporation, who were based in Massachusetts as is MIT. By 1962, DEC had already launched the PDP-1 (another candidate for the world’s first minicomputer) but the LINC was considerably cheaper and more compact.

Based on a 12-bit architecture, one innovation with the LINC was the tape drive (the LINCtape) which could store up to 400Kb and allowed a slow but reliable form of random access storage (somewhat like a very slow disk drive). The LINCtape evolved into DECtape, a common feature on DEC’s PDP line.

Digibarn's LINC system
Digibarn's LINC system

A small CRT could be used for output, and a rather clunky keyboard from Soroban Engineering (responsible for many computer keyboard of the same period) allowed input. Additional output could be made to a teletype, and the LINC could also be controlled by a set of rotary knobs which were essentially a precursor (pun intended) to the mouse.

The key application for the LINC was interfacing with lab equipment through the inbuilt A-to-D (analogue to digital) and D-to-A interfaces. This made the LINC a successful lab machine, although only 50 were built so it didn’t exactly change the world.

LINC exhibit at the Computer History Museum, California
LINC exhibit at the Computer History Museum, California

Significantly though, the LINC may be the world’s first home computer. Programmer Mary Allen Wilkes had a LINC system installed in her home, something that would be unfeasible with the 730kg PDP-1. It would take another 15 years or so before home computers became something that you could just go to the local electronics store to buy..

The LINC architecture grew into the PDP-5, PDP-8, PDP-12 and DECmate well into the 1970s. In the end though, the 16-bit PDP-11 and 32-bit VAX architecture (which were unrelated) moved things forward from there.

Image credits:
Jonathan Assink via Flickr - CC BY-ND 2.0
Don DeBold via Flickr – CC BY 2.0

Tuesday, 22 February 2022

Intel 80286 (1982)

Launched February 1982

By 1982, Intel was on a roll. Their 8086 processor series (launched in 1978) was gaining widespread acceptance and in particular had found itself in the IBM PC launched the previous summer.

Four years is a very long time in the microprocessor market, and by 1982 things had moved on. While the 8086 was good enough for a single-user business computer, more sophisticated systems needed a more sophisticated CPU. Intel wanted to capitalise on the success of the 8086, but come up with a processor that could be use in more powerful multiuser systems while maintaining a good deal of compatibility with the older CPU.

The resulting Intel 80286 processor was launched in February 1982, and it became commonly referred to as the “Intel 286” quite quickly. A 16-bit CPU clocked at 5 to 8 MHz at launch, the 286 could either run in “real mode” which was directly compatible with the 8086, or a “protected mode” which was more suitable for multitasking operating systems. Up to 16MB of RAM could be addressed, although few systems would come equipped with that much memory due to the price.

Intel 80286
Intel 80286

Although you could multitask with the 286, you could not run virtual “real mode” sessions. So it was impossible (for example) to run multiple DOS sessions on the computer. Switching between modes required either a reboot or some clever hardware and firmware trickery.

Some operating systems did use the full power of the 80286, including Microsoft’s long-forgotten version of UNIX called XENIX and a couple of other Unix-like OSes. These were niche markets, instead probably the best-known use of the 80286 was 1984’s IBM PC/AT which was a redesign of the original PC with the 286 dropped in, along with a new bus and various other improvements. Although the PC/AT still ran DOS and didn’t use the full features of the 286, it was much faster than the original PC which was frankly a bit of a tortoise.

Because IBM and other suppliers insisted on a second source for the 80286, Intel licenced the processor to other manufacturers. These included IBM, AMD, Harris, Siemens and Fujitsu. Harris pushed the speed of the processor up to 25MHz, twice as fast as the top-of-the-line Intel 80286 which ran at just 12.5MHz. For AMD, the second sourcing of the 286 gave them an entry into the Intel-compatible CPU market where they are still the only real competition to Intel today.

The 286 certainly moved things forward, but in terms of the PC it wasn’t the technological leap forward that it needed. In 1985, Intel launched the 80386 which could run multiple virtual 8086 real modes… this meant that a PC could run several DOS applications at once which was a key factor in the uptake of Windows rather than MS-DOS in the PC marketplace.

Although Intel officially dropped the 286 in 1991, Harris and AMD continued to develop it and provided some serious competition to Intel’s new 386 for quite a while. Today these 286 CPUs are still used in some embedded systems, and the Renesas CS80C286 is still available today for these applications.

Second source AMD 286 variant
Second source AMD 286 variant

The 80286 established that the PC architecture wasn’t a one-off design, and marked the beginning of a continual evolution of the platform which of course is still with us today. Because it couldn’t multitask DOS programs, its impact was more limited than the 80386 that followed. However, Windows supported the 286 until version 3.1 (launched in 1992). Many 286-based PCs soldiered on well into the 1990s, some even making it to the Internet age with applications such as Netscape Navigator – far outliving the usefulness of the previous generation.

Image credits:
Thomas Nguyen via Wikimedia Commons
- CC BY-SA 4.0
Pascal via Flickr – CC0

Sunday, 23 January 2022

Apple iMac G4 (2002)

Launched January 2002

Apple is a company with more ups and downs than most. By 2002, Apple had already had success with the original Apple II in 1977, wobbled a bit with the Apple III in 1980 but then moved on to more success with the original Macintosh in 1984. Successes followed, but during the 1990s the company’s fortunes declined significantly and by 1997 it was a whisker away from bankruptcy. But in 1997 Steve Jobs returned to Apple and injected some new ideas, and aided by now legendary design Jony Ive the company came up with the iconic iMac G3 – launched in 1998 – which fundamentally changed Apple’s fortunes.

The iMac G3 is one of Apple’s best-loved designs. Beautifully designed around the shape of the cathode ray tube (CRT) contained inside in translucent candy-coloured cases, the G3 caused a monumental stir in the market. Here was a computer than both looked beautiful and just worked out of the box. Sure, the basic design idea had been around for more than 20 years (the ADM3A being a notable example)… but Apple did it better.

Apple iMac G4
Apple iMac G4

By 2002 though technology had changed. Just four years previously the CRT was the standard display for almost all desktop computers, but by the early 2000s there was a shift towards LCD panels which were rapidly dropping in price and increasing in capabilities. Although CRTs still had a bit more life in them, the obvious choice for a forward-looking company such as Apple was to go with LCDs for their next-generation iMac.

In design terms though, the LCD panel necessitated a completely different design from the CRT in the G3. It wouldn’t make any sense to replicate the G3’s design when most of the box would be empty space with an LCD, but the G3 had set an incredibly high bar in design terms. So where could Apple go next?

So the Apple design team – led by Jony Ive – came up with something completely different. Instead of tucking all the system components in close to the display, the Apple iMac G4 featured the LCD mounted in an adjustable arm with the actual computer tucked into a large dome at the base. The base contained everything, including an optical drive, and it gave the G4 a distinctive look in the manner of an oversized desk lamp.

Inside was a PowerPC processor running at between 700 MHz and 1.2 GHz. Maximum memory was 1 or 2GB depending on model, but a rather more modest 128 or 256MB was supplied in the box. Internal expansion was limited as this was a compact all-in-one unit like its predecessor, but peripherals could be added via USB or FireWire. A modem and wired Ethernet port were built in, a wireless network adapter was available as an option.

It was an elegant – if odd-looking solution – with everything you needed in the box which required minimal effort to get working. However, the LCD panels were still pricey compared to CRTs so a few months later the eMac was launched, the last CRT-based Mac which was a fair bit cheaper than the iMac and sold well to educational markets.

Alternatively you could have bought a PC in a beige or grey box
Alternatively you could have bought a PC in a beige or grey box

The G4 stayed in production for just two and a half years before being replaced by the rather more sober G5. Today collectors can pick up a G4 for less than the price of the G3 with typical prices for a working system being just £120 or so.

Image credits:
Maxime Bober via Flickr - CC BY 2.0
Carl Berkley via Flickr - CC BY-ND 2.0