Psion: the last computer
Secrets of the Sony we never had
’Every microamp is sacred’ — David Tupman
Engineering for low power
I joined Alan Ferdman’s test department at Psion in November 1995. Most of the engineers came through there, and then went on to design. In testing, you had a great chance to look at the whole design and learn how it all worked. We benchmarked ourselves against other companies — against the new Windows CE devices, to see how much power they used for the same function. Psion handhelds used about a quarter of the power of the CE machines.
A lot of this was actually due to the software. With the hardware, you do your best, and use techniques such as clock gating. With process technology improvements we’d gain too, and we looked at offloading onto special hardware. While Windows machines took all the standard parts, Psion developed its own silicon. All credit to Mark Gretton for that — he was the big instigator for getting the Series 5 silicon going. He was just a genius, really — and we all worked very closely with the software team. But software played a big part — it was about how they used the hardware.
Power draw is all about how fast your clock runs — if you can slow that down as much as possible, that gives you the lowest power draw overall. Symbian OS back in those days had come from battery powered Organisers, and the Series 3, we had a ground-up understanding of making a lower power system. The Windows world was all about performance.
Why is software so important?
Remember that software isn’t doing anything 99 per cent of the time. If you’re entering something into Agenda, you only need the activity when you’re pressing the buttons; and you wake everything up. So it was all about that dynamic range — how low power can you go when you’re not doing anything, and how fast can you go when it’s active. We had a very wide dynamic range — we could get it down to nothing, but something happened, the full performance was there.
We had a saying in engineering that “every microamp was sacred”. We had to account for every microamp of power — every day that would be the mantra.
After the Series 5, I moved into the design team developing silicon for the 5MX and for all the follow-ons. As with Protea, we worked very closely with Martin Riddiford’s great company Therefore, to make sure everything would fit in the product — driving package side, components, to make the PCB as small and as thin as possible.
Martin always came up with these amazing mechanisms. Just look at the hinge on the Series 5 — or even the Series 3. Martin’s a genius at making these crazy mechanisms work. So Martin’s job was to figure out the mechanism — and our job was to figure out how to fit everything.
After the Symbian split, Ken McAlpine was just starting up his incubation team, his nursery team, back at Psion and a lot of ideas were coming out of that. Thinking back, I must have worked on 13 projects of which only three or four actually made it to production.
One of the problems we had was that every nine or 15 months we’d get a new general manager — there’d be a clean sweep and the old projects would get cancelled, and we’d move on to new projects. Sometimes we’d find out who the latest general manager was from the internet.
“We wanted to do SatNav.”
They were all were really interesting things to work on, it’s a pity they didn’t make it out of Psion. But they were all really interesting and every time you work on something like that you learn alot — and that makes the product you actually ship that much better.
There was the Bluetooth Revo; there was chip called Halla we worked with Parthus and Samsung to develop, and it was the first ARM9 system chip; Samsung had never done anything like that before. That was cancelled with the smartphone, Odin.
Back then here was a lot of technology going around that Psion was keeping an eye on. We’d meet all the component manufacturers and had a lot of ideas on what we wanted to do. We wanted to be in cellphones, we wanted to do Bluetooth stuff, we wanted to do SatNav. A lot of times in the end David Potter would say no, we’re not going to do that. And at the time, all the R&D money went to Symbian. We had so many great ideas, and so many of those industries have taken off: SatNav, with Mark at TomTom, and MP3 Players.
Ken and I spent a number of months just driving around the country trying to pull together a plan to make a hard disk drive MP3 player. Creative had their big CD-sized player with 2.5in drive, and there were lots of little Rio players; we thought we could make a play in that area because we had the technology. It was very much an investigation.
Unfortunately, there was no formal process for kicking projects off other than talking to David Potter.
Then the Psion and Motorola deal ended, we briefly tried to keep the phone thing going, but it really it wasn’t happening. In March  Psion laid off lots of hardware engineers [Psion laid off over 200 staff that day — ed].
I left shortly after that.
Our competition had gone from being the Windows CE people like Philips, to being Ericsson and Nokia, the cellphone guys.
One of my regrets is that Psion got out of the hardware business, pretty much. Actually, if you think about it though, a lot of the hardware business got killed by the cellphone. Look how much Palm suffered.
But it was an absolute blast. We shipped a million Revos, and to be part of that was great.
David Tupman is director of iPod hardware engineering at Apple
Sponsored: Benefits from the lessons learned in HPC