Transistorized - an electronics engineer's idyllic spot.



Embedded Automatic Weather Station - My Bachelor Thesis

After three years of regular studies, finally my time for writing a final thesis came. I was hesitating quite a lot about what topic to choose, and what to start designing. My initial idea was more connected with analog IC design, or to be more specific a design of a low-noise operational amplifier. Finally due to the lack of access to the very specific software and process design kits needed, I decided to switch the topic and dive deeper into microcontroller programming, which is also one of my electronics specialities.
Since I was long ago interested in measurements and data collecting, one day the idea of building an embedded automatic weather station came to my mind. The topic happend to be quite interesting and inspiring for future improvements of the developed prototype.
Because this topic is quite huge for a bachelor thesis and moreover I find it interesting to work with people, I proposed to one of my colleagues (Tsvetomir) the idea of building a weather station together as a thesis and luckily he accepted. Further we got an idea of expanding it with one more measurement parameter (Gamma background radioactivity), there another colleague (Kamen) got interested and started working on the Geiger counter part.

As a summary of the idea I can say that it was a great pleasure and fun working on that topic. Sometimes frustrating, even disappointing (when we were encountering issues) sometimes energy-recharging and fully-inspiring for further work. The total time spent on the project was around 6 months and it was totally worth it.

Here is the time I stop writing mere talk and I should say some more details about the weather station itself.

Firstly - the weater station's web server IP address is:

*Note1 - Dilema Ltd. stands there for my father's company since he was the main money sponsor for the station
*Note2 - The Weather Station is located in the back-yard of our house in Ruse - Bulgaria
*Note3 - Since that the Internet Service Provider in our neighbourhood is quite poor, sometimes the station may not be accessible, although an improvised server is always running in my house, so all the station data is saved for statistics on it - no data is lost.


I'll start by listing the weather elements that the weather station measures:

Parameter Range Accuracy
Air Temperature -55 ÷ 100 °C ± 0.5 °C at -10 ÷ 85 °C
Soil Temperature -55 ÷ 100 °C ± 0.5 °C at -10 ÷ 85 °C
Relative Humidity 0 ÷ 100 % ± 3.5 % at 30 ÷ 80 %
Atmospheric Pressure 750 ÷ 1250 hPa ± 1 hPa
Wind Speed 1 ÷ 30 m/s ± 2 m/s
Wind Direction 8 directions -
Solar Radiation 0 ÷ 1500 W/m2 ± 15% at -10 ÷ 85 °C
Gamma Background Radioactivity 0.01 ÷ 1000 uSv/h 5% at -10 ÷ 85 °C

The temperature sensors used are Dallas/Maxim DS18B20 OneWire digital sensors directly connected to the I/O of the micrcocontrollers.
For a relative humidity sensor a Humirel HS1100 capacitive sensor was used initially, after 7 months of work I decided to change it with a Honeywell HIH-4031 which gives an almost linear analogue output.
The atmospheric pressure sensor used is a MEMS integrated sensor manufactured my Freescale/Motorola MPL115A1 - SPI version.

Since that the budget for the thesis was highly limited I could not afford it to buy ready made calibrated anemometer and wind vane as the ones Met One Instruments and Vaisala sell. Therefore I had to think of a way of building them on my own. This happened to be a very tricky task, but finally I think the overall accuracy of the anemometer got quite OK, for the tools that I used for building it.
Luckily I had access to a lathe and with the help of some people we constructed the anemometer. The basic principle of operation is the following: a metal disc with 36 holes (360 deg or 36holes@10deg) was attached to the anemometer shaft and a simple opto-coupler was used for generating impulses. A frequency counter algorithm was implemented in the microcontroller system, which I will describe later.

Pictures of the home-made anemometer and wind vane:

Anemometer and Wind Vane bases

Two Nr 403 ball bearings were used.

2 Ball Bearings

I call it a bell-based construction.


Here's the 36 hole @ 10 degrees disc. in the bottom-left corner you can see the black opto-coupler used for pulse generation (36 impulses per revolution).

Optical Sensing

The second build anemometer (looks alot better) - the main problem was finding proper machines (a fine lathe, drilling machine, etc.etc...) I am an electrical engineer, so I'm ~excused.

The second build wind vane (looks worse than the first one), this time there was a lack of finding proper materials for it.

The assambled anemometer (a photo just before attaching it to the mast)

Assambled anemometer

The wind vane has a sort of the same construction. 8 Reed ampules and a disc with an attached magnet is used for sensing. This design prooved to work pretty decent, moreover I have an idea of implementing a software algorithm for sensing 16 directions by using only 8 reed ampules. It is pretty simple though.

Wind vane reed sensing ampules


I move-on to the electronics part. Three 8-bit microcontrollers were used for the whole system. This is not the optimal solution at all, but due to some university-resource restrictions a PIC 18F4525 and two ATMega32 microcontrollers were used. Two of them with fully utilized resouces (pins) and one of them almost free.

An article about the weather station was written and published in the Polish journal Elektronika, you can find out more details about the weather station from it. For now I do not have time synthesizing a brief introduction to the weather station electronics.

You can download a copy of the article (Automatic Embedded Weather Station) from here.

For the more interested ones, here's a pdf copy of my thesis
[sorry only bulgarian language, but I think it looks quite nice - LaTeX rulez] - there you will find full information about the station.

A friend of ours (Ivo Raychev) drew a 3D model of the station on SolidWorks, here's a 3D pdf of it
*Note - Adobe Acrobat with 3D extention is needed to view the 3D pdf

A 2D version is available here.

A small data extract collected by the server in csv format


First Prototype exibited at university's main hall:

First Tests of the first prototype:

The very first developement prototype - just a bunch of cables and boards:


Now more pictures of the station and the design process will follow:

It would be good to mention that two prototypes were built. The one below is the first prototype now installed at University.

Geiger counter measurement unit

Note that the video intends to show only the electronics part, the measurement sensors although may be outside the window, they do not measure properly, later they were installed on a mast on top of the building, unfortunately I do not have pictures nor videos.



Home Page