Re: build a hack box of choice?

Long posts get long responses   :-).

> > Scott Serr wrote:
> > > I might be narrow minded, but... hardware specifically for WiFi Linux
> > > projects.
>
> Watch out!! Long posting!! Sorry about that but I have stuff to say! :-)
>
> Let's spec this thing.
>
> Really you'd want a PC card slot to allow any WiFi radio to be used
> but beyond that isn't the rest of the hardware pretty generic?  Don't
> get carried away adding sound and video and USB. (I'd love them but
> let's keep it as cheap as possible.) How does this sound?
>
> Geode processor (133 or 233?) (x86 per Keith's suggestion)
> 16 MB ram (or 32 if cost effective - is it enough?)
> 8 MB flash (how much?)
> 10/100 ether w/POE
> 9 pin serial port
> A few LED's under software control.
> WiFi card
> Power supply

For an access point, adding support for a parallel port for a print server
wouldn't be much of  a stretch... especially since typically serial and
parallel port can be handled by a single winbond chip.  Also, adding
something like a compact flash card port to store the firmware, instead of
on board, and boot off that... to simplify upgrades  (ie if one could
compile a new firmware, and just load it onto a CF card on their pc, then
just plug the CF card into the unit, and viola, instant firmware upgrade.
cheap and geek friendly.... may even be less limiting, no worries about not
enough space for what you want to add, just get a larger capacity CF card if
you need it... for whatever each individual wants to compile into their
linux for the unit.  again, from an individual component cost, these only
add a tiny amount to the cost, yet greatly expand capabilities.

>
> plus COMPLETE DOCUMENTATION on CDROM,
> including a Linux-based development tool chain
> and source code and flashable image(s).
>
> What do you use for development with Geode? JTAG?
>
> (and of course additional boards available separately)
>
> How much is this worth to you? My experience says that
> the tool cdrom is the most valuable, missing part. If it's done so it's
useful!
>
> With a single PC slot and sources you could pull the WiFi card and
> replace it with other cards to put the box to use in other formats so
> it's still got flexibility. I think it might even be worth the extra
> to put TWO PC slots in, so that a bad decision on memory space does
> not make it obsolete for example. You could add flash or a modem or
> USB or another ether port to your own box without the base price going
> up for the rest of us.
>
> Would it be worth it to put the flash / ram in sockets to allow
> changing it out later? At Coactive we had two boards: development with
> lots of LED's and sockets for memory devices and production with
> everything soldered in and just the LED's you see on the front panel.
>
> Last night I spent some time looking at Openbrick and about 3 dozen
> other SBC's and can't find anything I really want to shell out my own
> money for. The disclaimer at the top of the Soekris price list says it
> all.  It warns that docs are bad and that you have to know what you
> are doing. Sort of the oppposite of documentation. And marketing. :-)
>
> Openbrick is more like a full PC at a full PC price in a small box.
> Like the Soekris, the other SBC's generally seem to be sold by
> hardware guys and look like they'd be painful for a mortal like me to
> learn to use.  I fear I'd spend the bucks and end up using it as a
> nice wall hanging. <BTW Soekris really is a pretty good deal on
> price.)
>
> Back in the real world we'd hire an engineer and get him to learn the
> board inside out and the vendor would offer training to us because
> we'd be planning on buying 30,000 units. But here I am all alone and
> those are not options... I suspect that I am all alone with several
> thousand other people.

I suspect you're right..... I seriously do see a huge market for such a
thing.... another idea might be to approach a company such as D-Link or USR
about building it.... since they've already got a lot of the work done, and
it eliminates the crushing competition factor... show them how they would
save on R&D costs, since this single design offers upgradeability via
ability to change out the radio, and easily modify the firmware to support
it by simply adding the appropriate linux driver (something no AP on the
market has that I know of)... show them how if they design this type AP, and
make its firmware open source, they'll actually have a product that will
carry them well into the future, which will save them retooling their line,
and will therefore bring the cost per unit down to a minimum, while at the
same time increasing value for the consumer, due to its upgradeability and
versatility and therefore allowing them to charge a slightly higher price
than the average soho AP.  Lets face it, from a marketing perspective,
printing the Linux penguin on a product these days is like printing money!
I'm sure this fact wouldn't be lost on the marketing execs.  Of course, they
would have to understand that this benefit is completely lost if they don't
adhere to a policy of completely open source.  This would mean increased
profit on the unit for a long time into the future.  They could even offer
"upgrade kits" with a new radio, and appropriate firmware for a much lower
price than a whole new router as upgrades became available.  It would
basically eliminate that 6 month life cycle of products, so they're not tied
to developing a whole new product every 6 months to keep up with the
technology.  They could commit to this core basic design for years to
come... They could even build different products themselves, based on the
same design... want a wifi bridge, take the same unit, and sell it with
Linux configured to handle bridging... want a repeater, again, take the same
unit, and configure linux to serve as a repeater.... And they wouldn't even
have to sell the entire unit for these capabilities, simply sell "patch
upgrade kits" that will provide whatever feature one wants to add..... for
the layperson, yet offer access to these advanced capabilities for the geek
who wants to do his own customizing....  versatile, upgradeable, good
value.... what more could you want?  The only real problem I can see for a
company in offering such a unit would be FCC certification.... I'm not 100%
certain that certification could be obtained if it were easy for the end
user to tweak the unit to say increase power output of the radio beyond
specs... or for that matter, to change the radio to a higher power unit than
the initial specs called for.  However, on the other side, it would be the
individual radio that would be subject to its own FCC certification, and so
wouldn't be looked upon any differently than a wifi radio marketed for a
full PC linux box.