I worked at EO...
Back in what seems like the mists of time, between 1991 and
1993, I worked for a
company called EO Computer Ltd.
We were based in the old Torch Computers building in Great
Shelford near Cambridge in the UK.
EO was an interesting place to work. It was genuinely high-tech - we were making the world's first practical tablet computers ! They weren't called tablets, because that term was not in general use yet. However, functionality was very similar to an iPad or Android Tablet, with control by gestures. However, there's more. While the EO machines had a tablet like size, they were intended as communication devices ("Personal Communicators") and they optionally included mobile phone hardware to allow communication over the mobile phone network.
While the EO machines had the size of tablets, they were really the first smartphones. They reached the market a year before IBM's "Simon" and three years before Nokia's 9000 Communicator. These days, many people give credit to Apple for the invention of the smartphone, but more than 14 years would pass before the first iPhone was available and more than 17 before the iPad. Apple's Newton was contemporaneous with the EO machines, but it was far more limited in scope: a pocket organizer with no communication ability.
The EO 440 and 880 (the difference was in the size of the screen and the whole tablet) included functionality of many different normal office devices:
It's like everything else. You can fax with it. You can phone with it. Send an electronic letter, or receive one. Exchange information with distant computers. Communicate over the airwaves, without wires. You can do these things in ways and places that were never possible with the traditional tools--fax machines, phones, computers. And instead of many devices, you can use only one.
We call it the EO Personal Communicator. Others have called it the future of communications. It may just be the biggest breakthrough since the telephone.
Shortly before I started work at EO, the Active Book Company was at the same Great Shelford address. However, that company was bought out, along with its engineers, to by AT&T, Matsushita and Olivetti who were funding EO. Hermann Hauser of Acorn and ARM fame had been behind the Active Book Company and would become the MD of EO Europe.
I saw the job advertised in "The Guardian" and applied. The letter that I received inviting me for an interview was actually still on Active Book Company letterhead, but my interview was with EO. The atmosphere seemed relaxed. Jamie interviewed me while wearing bermuda shorts. A few weeks later I started work at the company.
The Active Book Company had been designing their product around the ARM processor and an in-house OS based around smalltalk. Unfortunately, they were running out of money. Meanwhile, GO had their own 286 and 386 processor machines running Penpoint (written in Objective C) but wanted a big backer. AT&T had been trying to sell their Hobbit processor to Apple for the Newton. However, Apple switched to the ARM and AT&T were left looking for someone to sell their chip to, so they ended up funding EO to use the Hobbit.
The ARM of course went on to be wildly successful. Today it is the microprocessor in nearly all mobile phones, as well as devices which followed the ideas seen in the EO, such as Samsung's and HTC's telephones and tablets as well as Apple's iPhone and iPad.
GO stopped producing hardware, leaving it to other companies including EO (formed by hardware people from GO) and other hardware developers (NCR, for instance) to make the machines which would run their OS. EO took over ABC to get
The product was amazing for the time. It used analogue mobile phone technology to send email and faxes and allowed you to mark up faxes on the screen before sending them on to someone else. Fax was important at the time. Most people didn't have email yet.
The operating system was GO's Penpoint. This allowed control by "gestures" instead of traditional pull down menus. It was a very nice user interface. Rather than a touch screen, this interface was pen-based. This was necessary because the technology of the time required an active device at the tip of the pen. The pen technology came from Wacom.
I was employed to do test work on the software. This involved writing test code to exercise the operating system and such things as a Hobbit inverse assembler (real time disassembler) for an HP logic analyser. It also involved quite a lot of rather boring repetitive going through things in the UI that had been shown to cause problems before. We had a good system for logging bugs, and found thousands of them. The major ones were fixed before product release. Flash memory was too expensive at the time, so masked ROMs were used (these took weeks to come back from the factory). There was a jump table loaded into RAM so that problems could be corrected with patches without having to make new ROMs. However, it was vital that no really catastrophic bugs were in the ROM code.
The power supply seemed to give endless trouble. Either interference from it was affecting the accuracy of the wacom tablet, or it wasn't charging its NiCad batteries correctly. The external switched mode power supplies themselves which were supposed to support 100 - 240 V turned out to catch fire at more than 220 V, so we had to use 240->220 V transformers until improved power supplies turned up.
Industrial designers at Frog convinced the company that "wings" were required to make the product distinctive and they earnt a good fee for their design. However, most of us who worked on the product thought the design was a bit silly. The wings were awkward as you pushed the device into or pulled it out of a bag and they made the product larger for no obvious benefit. At the time, there was an advert on British TV for a sanitary towel "with wings" and this of course was compared with the shape of the EO. A flat rectangle with rounded corners was the obvious shape for a device like this. It was the shape of prototype machines and I still believe this is what EO should have gone for.
Because the "wings" became part of the product they had to be used for something. In the end, the machine could make sound recordings (8 bit, 8 kHz sampling so not exactly hifi) through a microphone in one 'wing' and a tiny ZX Spectrum style speaker in the other 'wing' could play them back. This fairly simple piece of code seemed to take ages to get right and then there were problems of both mechanical noise from the disk drive and electrical interference from the screen being audible in the recordings.
I don't know if anyone has a working EO machine today (February 2008 when I finally got around to writing this down). I imagine the NiCad battery packs will all have failed a long time ago. I can remember one nasty Y2K bug that I found. If the year was set to 2000 or beyond then the fax software would completely crash the machine. I don't remember if you had to receive a fax for this to happen or whether it just happened if some part of it was called, perhaps to view a fax already received. If anyone's still using the machine, it's probably as well
that faxes themselves have pretty much faded out of existance !
The Hobbit processor came from AT&T's CRISP (C reduced instruction set computer) research, so was basically designed to run the output of the PCC portable C compiler. This is a very simple C compiler with no optimisation. The Hobbit had a stack based architecture with no registers, just a stack cache. So, the top few entries of the cache are more or less as fast as registers. For some reason we ended up using a C compiler from Metaware which didn't seem to understand this. It was a much more clever compiler, making optimizations such as register colouring (i.e. copying values back and forth between "registers" to try to make things go faster) but I'm not sure this actually helped performance in this instance because they were sometimes the wrong optimizations for the unusual architecture of the Hobbit. Due to working on the inverse assembler I ended up seeing a lot of the generated code and it never looked very optimal to me. I don't mean to suggest that Metaware's compiler is completely stupid in all situations. These optimizations were the right thing to do on a processor with traditional registers such as a 68K or 386, but wrong on the Hobbit. I think it was a quick compiler port, and perhaps a bad choice of compiler command line switches didn't help.
Eventually the product got out the door and got good reviews:
The spec sounds terrible now, but was good
for the time: 20MHz processor, 8Mb of RAM, 8Mb ROM (not flash!), 20Mb
hard disk. Serial and parallel ports, inbuilt 14.4kb/s modem. Also note
the mention of gateways to non-internet based email systems. A 20 MHz
Hobbit was competitive with a 25MHz 80486. At the time, that
was quite a fast chip.
We spent a bit of time worrying about the effect that Apple
might have on sales, but in reality I think Microsoft's "Windows
for Pens" which a bigger threat. This was pre-NT and pre-95. Their
product, based on MS-DOS and Windows 3, wasn't particularly
good. However, we had a lot of FUD to contend with. MS made
life very difficult for GO, EO and Penpoint. This is covered in
Jerry Kaplan's very readable book "Startup: A Silicon Valley Adventure
". Jerry Kaplan was one of the main people behind GO.
Unfortunately, the machine never seemed to recognize my handwriting,
no matter how much I trained it. I don't know if that applied to
The company itself started to go wrong in 1993. Some UK staff moved to the US office, but most found other jobs around Cambridge. After a week of three big events (Tuesday we find my wife is pregnant with our first child, Wednesday our offer to buy a house is accepted, Thursday we find out that my employer is shutting the doors so I won't have a job for much longer), I went to Harlequin to write device drivers for Microsoft's then very new and not yet publicly available NT operating system. This software drove film recorders and other large printing machines to support Harlequin's Scriptworks Postscript interpreter on NT.
Overall, EO was a lot of fun. Talented people working on something new, trips to the Wrestlers pub for Thai food at lunchtime, trip to San Francisco involving Alcatraz and quite a bit of beer as well as working quite hard. I'll add to this web page if I think of more to say.
Photo of EO Europe Staff
I'm not sure why this photo was taken when it was, and there are quite a few people missing from it. Unfortunately, I don't remember everyone's name, but here are the ones I do remember:
1: David Hembrow (myself)
2: Brian Knight
3: Shaun Pitman
4: Gwyn Jones
5: John Grogan
6: John Rickard
7: Peter Westlake
8: John Cowell
9: Pauline Diggins (Hermann's PA)
10: John Bowler
11: Paul Bond
12: Adrian Stephens
13: Chris Mayers
14: Hugh Barass
15: Liz Radley
16: Andrew Haley
17: Viv Dowton (?)
18: Chris Fox
19: Francis Blydenstein
20: Mick Taylor
If I have any errors in this list, if you are one of these people, or you know who someone is, please remind me.
Introducing PenPoint 1991:
PenPoint demonstration 1991:
Both videos show the GO prototype hardware, based on an 80386 processor.
There is also a wikipedia article about EO.
I never owned an EO myself, and didn't own any other tablet either until nearly 20 years after I left EO when I bought an Android table.
These days I sell bicycle components and lead cycle tours for a living.