Gregory Pfeil writes: >Anyway, I've been thinking about the minimal computer needed to do what I >needed--more of a size and battery life constraint than a price thing. [...] >I started designing a computer which was basically nothing. It has ports >for a monitor, keyboard, audio I/O, and PCMCIA. It's powered by a 100MHz >StrongARM and has a linux kernel in flash. The shell is minimal, having >only very basic commands, such as telnet and shutdown. It should run on >two AAA batteries. You're making a few leaps here. First, a StrongARM, especially at 100MHz, is massive overkill if all you want to do is telnet; even in low power mode, it has about 3x the power consumption of a few of the alternatives (for instance, the 68k processor used in the Palm Pilot). It won't run on two AAA batteries; the Itsy fellows only did that for form factor purposes, not because they anticipate it working out that way long-term. Second, if all you want to do is connect to a remote system using, e.g, PPP, then you probably don't want to use Linux. Linux is great, but without storage (your note didn't include ROM, RAM, Flash or disk; I assume you're putting a bunch of flash in), it's not very workable. Consider instead systems designed for small spaces, such as QNX, VxWorks, DOS. Third, you mention a bunch of different interfaces offhandedly. If you're designing for low power and low space, you need to make sure that you completely understand and require the appropriate systems. If the system will let you telnet, are you sure you need a 'monitor' out? What would you put in the PCMCIA slot, and why wouldn't you build it onto the board? (PCMCIA is fairly nasty to design into portables because of its power, space, and cost problems). Audio I/O means that you intend to speak to the unit? To what purpose? >Add a cellular modem PCMCIA and you have a system that will let you >harness the power of the damn fast computer sitting on your desktop at home. (at 9600 baud max, ~$9.00 an hour) >I'll have sketches and preliminary plans on my site soon (hopefully by >this time next week), and I'll put a link to them here. I expect that it >will be roughly twice the size of a type III PCMCIA card. Your expectations are too high; the RAM, circuit board, PCMCIA slot, chips, batteries, interfaces and packaging would make this a little bigger than that. >So, what am I forgetting? Someone tell me this is crazy before I go too >far and lose a lot of money trying to build it. If no one yells at me, >I'll assume this is practical and throw my life savings into it--so >please speak up. I wouldn't throw any money at this until you've figured out exactly what it does and, if you intend to market this as a product, exactly what the size of the market is, the expected portion of the market that would buy your product would be, the cost of making the unit in quantity, and the margin (cost to consumer or, more likely, channel, minus your cost) you could expect to get. My immediate guess is that for a single garage entrepeneur, this is not an economical play. If you intend to produce one nonrepeatable unit for your own pleasure, I'd expect to spend about 25-30k (before considering opportunity costs, etc.) For that kind of money, I would survive with a $10k PC/104 stack and a downpayment on a nice house. Not to dash your hopes, but this sort of thing is non-trivial. Disclaimer: I work for a company that does a great deal of design and implementation work on tiny/handheld/next-generation consumer and industrial electronics, including work on StrongARM and small-form-factor units. So I could be trying to discourage the competition. :) Felix
From Wear-Hard Mailing list Archive (WH)
Maintained by R. Paul McCarty
Archive created with babymail