I spent a few hours last night doing experiments. The challenge is a self powered radio transceiver with significant range, and one critical component is the power supply. Constraints are that their needs to be a regulated 5V supply if the device is going to do any sort of sensing (temperature etc), and there are advantages to using a single cell as it doesn't do any harm if it goes flat. The next constraint is the current draw and I'm waiting on the high sensitivity modules from china that hopefully can boost the range I already have of 400 metres. These draw 5mA and with a picaxe and maybe an op amp, I think the total current may be 8mA at 5V regulated which is 7-8V unregulated which is about 40mA from the battery.
So the experiments last night were to see how much current a simple step up converter could produce. I used a 1000uF cap and a 1k resistor as a load and the test was to get the maximum volts by tweaking components and pulse widths. The pleak current means the resistance of the coil needs to be very low which means a coil with thick (1mm) wire. I used a toroid about 3cm in diameter and it has a measured inductance of 100uH. The max efficiency peaks at 1Khz (which is a bit strange as I would have predicted 10Khz) with a square wave. The peak volts were 23V so with a 1k load that is 23mA at 23V = 529mW. The components were very simple - a BUK555 power mosfet, a 914 diode and the toroid inductor. Nothing was even remotely warm. It even managed to produce 10V output under load with the battery volts down to 0.55V.
Next I changed the driver circuit from a 555 to a 14M, and measured the output volts using a voltage divider (100k and 10k in series) into an ADC. By overclocking the picaxe at 8Mhz a 1Khz square wave can be produced with "pause 1" (there probably is another way to do this, eg sound, or underclocking and using pwm - any suggestions?). The output volts are sampled, then the picaxe runs for 20 cycles and then the output volts are sampled again. With a few lines of code the volts can be kept in a fairly tight range between 7 and 7.5V, and the stepup converter only works intermittently as needed. Given the picaxe is already needed for doing the radio comms, I think this might end up being cheaper than using commercial step up chips eg the coil is needed for both so is a BUK555 and a 914 cheaper than a TL499?
The next component is a bootstrap stepup converter to get the whole thing running in the first place. I'm thinking of using a solar light and even using the white led as the regulator. This will self start and run to about 3V which ought to be enough to get a picaxe going. The picaxe can then disable the solar light circuit (not quite sure how - probably short one bit to ground using a 547).
If that all works, the picaxe driven regulator has another advantage in that it can change the output volts. So it could run at 7V most of the time, but go up to 12V to charge up a cap to run a data packet from a higher power RF transmitter.
Then the next bit will be to build a better nicad overcharge circuit so the cells can be charged with big (1A x4) solar cells, but not overcharge. I'm thinking of two temperature sensors, one strapped to the cell and one in free air and measure the temp rise when a cell is charged. Picaxe can do that sort of thing very easily.
If this all works, maybe the 400m distance can be improved!