Welcome to the Yargy Bot Wearable PC page!
These pages document my project to create my version of the wearable PC (hereafter WPC for short). The project's home page is here. This page is written in the style of a blog, but the entries are in proper chronological order with the oldest entries first. Click here for the latest entry. - - -- --- -- - - 2007-03-20 - First (overdue) update I have a working (but not quite complete) main unit. It is based on the
Commell LS-371 3.5" single board computer. I have
equipped this board with an Intel Core Duo T2500 (2 GHz, 2 MB L2
cache, 667 MHz FSB, 31W power), 1GB of 667MHz DDR2 RAM, and an Intel PRO/Wireless 2915ABG mini PCI WLAN card. Apart
from cabling and basic front-panel components, the only other thing in my
little black box is a Seagate ST980825AS 80GB, 7200rpm, 2.5" serial ATA hard
drive. While this machine is far more powerful than anything I had
envisaged at the outset of this project, it is also bulkier than I
originally intended. As for "how I got here from there", I suppose it started when I finally got some pricing information on the PFU Systems Plug-N-Run modules I had been looking at. I don't remember the exact numbers and I've since thrown out the scrap of paper I noted them down on while I was on the phone (another reason why I prefer email), but they ran from around $1100 to well over $1500 (all dollars are Australian unless stated otherwise). The carrier board would have added another few hundred dollars (or more), bringing the total to about $2000 for a higher-end unit. This seemed a bit on the pricey side to me given that the OQO model 02 (for example) also goes for about $2000, but that's for a complete system with a hard drive, battery, display, keyboard, and pointing device. Admittedly the high-end Plug-N-Runs would outperform an OQO model 02, but the OQO is also more compact than any complete system based on a Plug-N-Run would be. I was once again motivated to go searching for alternatives. I briefly looked at Via's Nano-ITX boards, but while 12cm would be small enough in the long dimension, 12x12 is a bit big overall and I was a little dubious about the performance I could expect from these processors. I then heard that Via is apparently about to release a line of "pico-ITX" boards (story here). The size of these boards is much more suitable, but no-one seems to know when they will be available and I got no response when I emailed Via (except an auto-reponse saying that it was Chinese New Year and they would get back to me when the holiday was over). Similar performance questions apply to the pico boards as to the nano boards. After a bit more poking around, I finally made some real break throughs into the world of single board computers. There are a LOT of these out there and I could devote a whole page to my researches into various EPIC- and 3.5"-sized boards. The major factors that eventually decided it were
I was "this close" to buying Commell's LE-370 board for $440 from Synrax. They did not (and as of this writing still do not) list the superior LS-371 on their website, but at the eleventh hour I thought "no harm in asking" and it turned out that they had a single one in stock at only $511.50. Even though I knew it was going to give me power consumption problems, I just couldn't pass up this newer, superior board at such a small premium to the older board. The T2500 CPU is overkill, but I got it fairly cheaply at $239. I started out looking at 400-series Celeron M processors, but it turned out that they're pretty hard to come by here, and I eventually realised that the Core Duo's are not all that much more power-hungry even at full speed, and they can be throttled back to 1GHz to reduce their power consumption to 15W or less. For seriously low power consumption you need an ultra-low voltage CPU, but they only come in the surface-mounting BGA package so you'd have to buy a board that has one soldered to it at the factory. I prefer the flexibility (and repairability) of being able to swap out (or upgrade) the CPU. The hard drive cost me $120 (I know I said earlier that I could get an 80GB drive for under $100, but that was a 5400rpm PATA drive). The RAM was $90 and I spent $100 on thirty-two 2800mAH NiMH AA batteries. The LS-371 will run off anything from 8 to 24VDC, so my plan is to make four 9.6V battery packs. For portability I will try running it off two packs in parallel (while the other two packs are charging), but I don't know how long this 5600mAH will last and I suspect I'll end up using three packs in parallel for 8400mAH (buying 16 more batteries so I can have a total of six packs). I am still waiting on one last item to complete the main unit: an 8GB CompactFlash card. I eventually want to make this the primary boot device (for short boot times), and I also want to see if I can set up hibernation under Linux such that it stores the system state on the CF card, but without having an active swap partition on the CF card during normal use because I suspect that would wear out the flash memory too quickly. I paid $114 for the CF card, so that's about $1200 all-up so far. I also bought a Multicomp AB78 box to put it in, a key-operated switch to turn it on and off (key operated to prevent it getting turned on accidentally when I'm carrying it around), a couple of LEDs with bezels for the front panel, a 40mm fan, bits and pieces for the battery packs, etc. I also used a few spare parts that I had lying around, and then there's wear and tear on my tools and so on. Even so, if you reckon it to be about $1500, it seems pretty reasonable compared to the Plug-N-Run or OQO pricing (remembering that my black box has the fast 80GB hard drive, 1GB fast RAM, batteries, dual-core CPU, etc). My box is also more user-servicable than a Plug-N-Run or especially an OQO system would be. Heat is still a bit of an issue and this box may (or may not -- I don't know yet) need a bit more air than an OQO does, but it's not meant to be hand-held, so the heat is not so much of a direct problem for the user. On the chordite keyboard front, I have some half-baked ideas that have been on the back burner while I built the black box. I made an alginate mold and I now have a plaster copy of my hand. I made a latex mold from this plaster model and I'm waiting on some polyester resin to make a plastic version of the plaster model (since plaster is porous and a bit too delicate to work with). I'm not entirely sure where I'll go from there, but I'm hoping that I'll eventually be able to make a new and improved, better-fitting chordite. Well that's all for now. Hopefully it will take me less than six weeks to post the next update. - - -- --- -- - - 2007-04-18 - Update I have pretty much given up on the chordite because it seems to me that it will just take too much time, effort, and money to get it to a stage where I'm happy with it, and because I discovered the AlphaGrip AG-5. The chordite has the advantage of being a potentially smaller one-handed device, but the AG-5 is a finished commercial product (although sometimes still characterised as a prototype by its inventor), has better pointing device support, and would seem to be capable of faster typing speeds. I only ordered my AG-5 yesterday (directly from AlphaGrip), so it will be a while before I have it in my hot little hands. I don't have time to write any more now, but will try to update further once the AG-5 arrives, at which time I should be pretty much there. |