Deutsch English

Weather ballon

Under construction!


Since 2009 Finger from arranged an annual electronic festival. In 2012, a weather ballon was started by some members. Onboard there was an little CB-Transmitter speaking on the radio.

It was an crazy looking setup, but to the general amazement it worked very well. The collected data (temperature, batterie voltage) were converted to speech by an ATMega before transmittet to ground.



In 2013, Ludwig from the finger forum and I have started another run on the project and developed a new version. We have planned big: RTTY-raduiteletype, voice radio, GPS, barometrical altimetry, temperature sensing... but in the end, nothing worked. The software wasn't finished, The transmitter failed after 30 seconds of flight. Total flop.

2014 everything will be better.

2014 the mess we made shall not be repeated, so we've started early planing everything. At first, we've cut down the feature list to this:

  • voice output on CB- or AFU-radio
  • radioteletype
  • GPS location determination
  • Temperatur measurement with PT1000

The Hardware

Already in december I've started drawing the new schematics, after that I've started planning the PCB. Thanks to components in 0603 everything fits onto 38x50mm - soldering this was surprisingly easy.

On this images, the HAL-Surfice looks like the surface of the moon - in reality, the pads are sufficient smooth.

Unfornately, I've had run out of time when I've finished the PCBs because I wanted them before my first test phase at the University. Because of that, I've made some mistaktes. In this configuration, T1 is always conductive, the library for the LM358 has twisted pins for the supply voltage and to cap it all the differential amplifier inputs are twisted, too. Now only bodge wires can fix this. Apropos, the PCBs are only 0.8mm thick.

The absense of 20k-resistors and a fitting 100k digital potentiometer justifys the presence of another few bodge wires.


As mentioned above, the necessary hacks are documented here:

On the top layer, you have to scrape away one trace at the OP1. (marked blue).



On the bottom layer is much more to change. Also at the OP1 you have to scrape away two other traces before you can solder the two modwires marked in green. The yellow marked cap must also be soldered in, otherwise the filter will maybe oszillate. At the OP2 used for temperatur measurement you need to change much more. You have to scrape away nearly all traces. Be carefull at pin 3 and 5, the connection to the near via must stay.


Many GPS-Receiver are speed- and altitude limitated, as far as I know at 500m/s and 10km. The maximum speed isn't out problem because the ballon can't travel so fast, but the altitude limitation is a big problem as we plan to break the 30km mark. Because of this, Jan_tux spend an DFM06 professional dropsond. On this PCB is a Navman Jupiter 30XLP GPS receiver which should not have such a problem. Directly on the PCB I've soldered wire to the UART RX- and TX-Pins, the 3,3V supply voltage and the 3V backup power.

Fritzler consented to programm the gps-software, but he needed an datadump straigth out of the modul. Because of this, the modul had to move into an lunchbox, together wis batteries and an USB-UART-Bridge. With this construction on the dasboard and with my laptop on the passanger seat I drove 15minutes through the city.

The transmitter.

Ludwig had delieverd the schematics for the modulator and the oszillator, so it was his turn to test them. Build up it looks like pictured on the right, the biggest parts arte the two transformers and the crystal.

Everythink worked very well as documented by the oszillogram below

The current consumpten amounts to 8mA @ 12V.

The hardware 2

Because of the annoying bugs mentioned abouve I've made an Rev.2 PCB and orderd them by iTeadStudio. $11 PCBs with $25 delivery charges... but it's worth it.

Beside the error correction I've put an 16MBit flash memory onto the board to make it possible to store more than 15s sound at 8kHz sample rate. Also I've disconnected the digital potentiometer from the filter circuit because we figured out that lowering the NF-amplitude would not lower the current consume of the HF-amplifier. Instead, I've routet the wiper pins out to three SMD-Pads to connect it directly to the ampilfier.

The software

The main programm

A small part of the software like the RTTY-Encoder is recycled fromt he 2013 version, but most of it I've rewritten. First I've thought out an basic concept shown in the program flow chart on the right. Every function must noch block the program, otherwise it's possible for the GPS to loose characters.

RTTY Operation

When all sensores worked well, I finaly could assembly the first RTTY-Telegramm. My PC is running TrueTTY that decodes the telegram.

Here is an example telegram:

DM2LCT TORTRONIK.DE DATA:4740.0067N:00926.6211E:

In this approximately 30 seconds long telegram are the following data coded:

  • Ludwigs callsign
  • internet reference for more informations
  • position(4740.0067N zu 00926.6211E)
  • altitude (455,8m)
  • spped (0,1km/h)
  • direction(164,16°)
  • sattelite count(6)
  • inside temperatur (-44,07°C)
  • outside temperatur(-44,28°C)
  • batterie voltage(10,375V)

For this test, the temperatur sensors weren't connected.


What has to be done until the launch?



Programming RTTY-generator finished
Programming sound output finished
Programming batterie measurement finished
Programming speak generation started
Programming GPS finished
Programming temperatur sensors finished
join all modules        started


Testen µC-Umgebung   finished
Testen R2R-DAC & Filter finished
Testen Temperaturmessung finished
Kalibrieren Temperaturmessung finished
Testen Oszillator finished
Testen Modulator finished
Entwickeln Endstufe                          started






1545 Download
1305 Download
1259 Download
1782 Download
2296 Download
1556 Download
2260 Download