Quote:> Hi all,
>I'm new to embedded scene, and totally lost here. I'm looking to design a
>totally stand-alone weather station and need to select a device for this.
>Requirements are:
>* low power consumption, ability to feed off single voltage (~12V solar
>power)
At what latitude are you ? That will determine the height of the sun
in the winter and thus the number of hours of daylight and the
atmospheric losses (airmass). What are the weather conditions in the
winter ? If there is a lot of clouds in the winter months, the solar
cell output will drop even further. Thus you may need backup power for
weeks or even a month. Thus, the average power consumption needs to be
extremely low.
Quote:>* 2x RS232 ports
>* Ethernet (10Mbit will be fine)
If this is a stand alone device, what do you need the Ethernet for ?
Such high speed devices consume a lot of power.
Quote:>* solid-state disk (regular HDDs will consume too much power, and probably
>won't work outdoors)
>* ability to run "conventional" Linux that I can manage using shell access.
>I'll also need to run Python and Perl on the box.
>Internet connectivity will
>be either IPv4 (DHCP, ethernet), or PPP via serial.
So this is not a stand alone device after all ?
Why not run the power over ethernet or take it from the RS-232
handshake lines (assuming your system runs on a few milliamps).
Quote:>Having a small MySQL
>server would be nice but I can probably live without it - it's probably
>going to require too much disk space anyway
>* CPU-wise requirements are very low - basically it needs to poll serial
>port every second, and send off aggregated data every minute or so
If those are the requirements, then the system could be powered off
for 99 % of the time to save power. Thus, you need a system which can
go into deep hibernation and wake up with an external interrupt, such
as a CMOS clock once a second or you need an OS that boots in much
less than 100 ms if you do a restart every second.
Quote:>* extremely stable - will need to run unattended for months
>* weather-proof(able)
This depends on your environment requirement.
If plenty of power would be available, it would be quite simple to
keep the component and cabinet indoor temperature 5-10 degrees above
the ambient temperature (even when a warm wet wind rapidly increases
the temperature). This will prevent condensation on the components and
you can use a box with holes in the bottom to ventilate the box and
get rid of humid air when the temperature drops. With sufficient power
you can even keep the indoor temperature above 0 C (since some
components are not specified for subzero temperatures).
Quote:>* cheap! This is my personal project!
>Any recommendations appreciated!
You must either do something about your power supply or rethink the
division of labour in the system, e.g. transmit the raw measurements
directly from a system with extremely low power consumption and do any
hard processing in a system with more resources.
Paul
by