I moved to Australia early this year (Jan 2008), so not much progress has happened on this project other than a few new ideas!
I have recently been monitoring the activities of the QRSS Knights (link) who operate mW level transmitters, mostly on 30m (10.14MHz). I will possibly move the transmitter to this band, and look at some other modulation schemes such as Hell or WSPR. I was recently sent some 30m band crystals from QRSS Knight Paolo in Italy (thanks Paolo!), so hope to put these to use soon.
The expectation is that the hardware will fly away, never to be seen again. A solar balloon can fly all day, and hence it is likely to end up somewhere in the pacific. Hence one of the goals is to keep the hardware cost very low and to keep the hardware very light. Current target is for less than $100 (NZ) and less than 200g.
The transmitter concept is simple. Use a CPU as a transmitter. The CPU has the ability to adjust its own clock frequency using a DAC output (generated from a PWM output) to control a varicap diode (in fact any diode works). Add an amplifier and a LP filter, and you have a frequency modulated transmitter!
The modulation system selected is DominoEX. This is a realitively new system developed by Murray (ZL1BPU) and Con (ZL2AFP). See http://www.qsl.net/zl1bpu/DOMINO/Index.htm. Software (using a PC soundcard) is available through this site.
The transmitter uses DominoEX 4. This operates at about 4 baud, with 1 nibble per symbol, giving 16bps throughput. Tones are spaced about 8Hz apart. At present just a fixed message is transmitted. This will soon be enhanced with GPS location information. Transmission is on 3.58 MHz (or thereabouts), this being 1/4 of the CPU clock frequency of 14.31818MHz.
The prototype transmitted:
This is assembled from PCBs on hand. It's obviously not a flight version! The CPU used is an AVR 'ATmega8'. For now the RF amp and LPF use kits from http://partsandkits.com/ . It is all housed in an old AWA trunk radio module.
To calibrate the transmitter, a local reference oscillator (running at 1.842MHz - since that is what I had) is used to help measure the CPUs local clock. The software self calibrates the appropriate PWM setting for each of the 18 tones. The final unit will use the 1PPS (pulse-per-second) output from a GPS module as the calibration reference.
Some of the tasks to be done:
last update - 9 July 2008
www.isnotcrazy.com - the site for mad scientists