I guess we all have this one project in the back of our mind. A must-do projects we've been collecting facts
and parts for over the last decade. Knowingly or not. In my case, that's the weatherstation. OK, I can buy a
ready built station for just a few euro's, but that's not what I want. I want to MAKE one, not just BUY one. I
want to take up the engineering challenge.
Also: my weather station needs to be on the net. It must be able to collect data, process the data in tables
and graphs and present them to the world on demand. Possibly it should also collect data from fellow weather
stations from all over the world and create a free overview of the current weather. Keeping track of old
weatherdata for Guatemala since 1967 might be too much. We must not spoil people too much.
Now, there are several ways to skin this cat. Cheap ways and expensive ways. Easy and difficult. high and low
level. Microcontroller or microprocessor.
-
One method could be to use a microcontroller like a PIC or an AVR. Some of these already have a TCP/IP stack
available, supplied by the manufacturer or by a third party. Which still does not relieve you from having to
design and produce the ethernet hardware (circuits plus signal conditioning).
But it might as well be better to add a hardware network card from a third party. The WebTiger (produced by
www.achatz.nl)
is just about the best you can buy for little money. You would only need to write your own network software,
or use the examples in Bascom.
Going this route will set you back a few hundred euro's and many months of programming to get the network
running stable.
-
Another method is to use a microprocessor, like a Z80 or an ARM, but then things get even more difficult.
You will need to build just about everything from the control logic upto all interfaces. Unless, of course,
you take the easier route and buy a WebTiger. Still, this is the least attractive road to go, if you want to
arrive someplace in the near future.
-
A much better approach is to get yourself a Linux based SBC. I've seen many on the web. All commercial
products. Without any charm. They use Linux to circumvent the cost of a Windows CE or VxWorks licence. Not
my kind of people, in general.
By far the most attractive Linux SBC is the Fox Board (as can be seen at
http://www.acmesystems.it/?id=4).
This is a real neat system, affordable (at EUR 125) and the makers have made a fun out of running Linux.
Take a look at this gadget: the Tux case! See it NOW on
http://www.acmesystems.it/?id=21.
You won't find this level of dedication on the sites of the big SBC makers. As I told before: they are in it
for the money. Acme-systems seems to be in for the fun and spirit of Linux.
For this money, ca EUR 130, plus EUR 30 for the magnificent case. In total: EUR 200 for a fully functional
system. Not bad at all.
-
So far the good solutions. Time for a bad one, be it a cheap one. Instead of searching the web for a compact
yet good solution, there's also the junkbox solution. Our kind of people has at least one old mainboard
(MoBo) laying around somewhere in a corner. Now, if this mainboard happens to be a 486 board, you're ready.
The old MoBo has all that's needed and if it lacks a few thingies: that's what the ISA slots are for!
Just install a small Linux like Slax
www.slax.org,
or DSL (Damn Small Linux
www.damnsmalllinux.org)
and you're done. These small distro's occupy less than 200 MB so you can even run it off a Compact Flash
drive, if required.
The benefits of using this kind of computer for a Linux SBC are:
- Free hardware (in most cases)
- Runs YOUR flavour of Linux
- Is fully 80386 compatible
- Will run YOUR compiler
- As much I/O as you can afford
- Keyboard, mouse, VGA are easy to connect
- Free case and power supply
The disadvantage of course is that the 486 board will be rather power hungry, compared to (e.g.) a Fox
Board. Still, at the current energy prices (1 WattYear = EUR 2) you can run it 2 - 3 years for the price of
a Fox Board.
Page created on 20 August 2006 and
Page equipped with FroogleBuster technology