Acorn Electron

~ the last words ~

Acorn had the schools. The BBC Micro was in most British classrooms by 1983 and it cost too much for the families whose children used it, so Acorn built a cheaper version aimed straight at the ZX Spectrum. The Electron was announced in 1982 for release the same year, slipped, and finally went on sale on 25 august 1983 at £199. It had 32K of RAM and BBC BASIC II in ROM, which meant the programs a child typed in at school mostly ran at home.

Getting the price down meant getting the chip count down. Where the BBC Micro used a shelf of separate parts for video timing and sound, the Electron folded almost all of it into one 2,400-gate uncommitted logic array built with Ferranti. Sinclair had done the same trick on the ZX81 and it was the correct engineering answer. The cost showed up in what got dropped, one sound channel instead of the Beeb's three and no teletext mode at all. It showed up worse in the memory bus, which the ULA monopolized during the visible part of every scan line, so in the heavier screen modes the processor ran at a quarter of the speed of the machine it was imitating.

None of that decided anything, because the machines were not in the shops. Acorn had orders for around 300,000 units for delivery before Christmas 1983 and the Malaysian line produced about 30,000, since only one ULA in ten came out of Ferranti working. People fought over the few that reached Boots and WH Smith. After Christmas the yield problem was solved and the Electrons arrived in volume into a market that had spent its money. Acorn's own marketing manager put it plainly years later: deliveries went up to meet demand that was no longer there.

The part of this that stays interesting to anyone who has tried to put hardware into production is that the Electron was a finished design. It worked, it was cheap, it did what the brief said. What killed it happened at a fab, in a step nobody writing software ever sees, and it turned a correct product into an absent one for the six weeks that mattered. I was not born when this happened, and I have still spent years watching the same gap between a board that works on a bench and a board somebody can actually build a thousand of. Yield is the least romantic thing in computing and it decides more of it than anyone admits.

~ the record ~

~ what it cost acorn ~

The unsold inventory landed on Acorn in early 1985 alongside an abandoned attempt to sell into the United States, and the company that had been reporting profits up 156 percent in april 1984 was suddenly unable to fund itself. Olivetti rescued it by taking a 49.3 percent stake, which is the moment Acorn stopped being an independent British computer company even though the name stayed on the case for another decade.

Total Electron sales over the whole life of the machine were somewhere between 200,000 and 250,000. For a few weeks it was the best selling micro in Britain. Retailers discounted the leftovers into bundles with games and expansion units, which is how most of the ones that exist got sold.

~ the long tail ~

Acorn walked away and the Electron kept going without it. Cheap discounted machines built a user base after the fact, and third parties supplied the expansions Acorn had shipped late or never, along with a steady flow of games through the second half of the eighties. Because the Electron ran BBC BASIC, that market propped up software production for the BBC Micro too, which is a strange afterlife for a machine that failed as a product.

Acorn's next serious idea was to design its own processor rather than buy one, and the team that did it had come off the BBC Micro and the Electron. That chip is in the phone you are reading this on.

@}-->--

              )
             (_)
            .-'-.
            |   |
            |   |
            |   |
            |   |
          __|   |__   .-.
       .-'  |   |  `-:   :
      :     `---'     :-'
       `-._       _.-'
           '""""""

« previous  ·  random  ·  next »