Some design decision have been taken, and here a picture of the box the experiment will be in:
The black, lightproof, inner box will house the temperature controlled tritium light, photo diodes, and pre-amplifier. The PCB will hold a PIC 18F14K22 and some other electronics, a power MOSFET to control the heater resistors glued to the tritium light in the black box. The photo diodes will be 2 Osram BPW21, visible light range, as I use a white tritium light (those come in many colors, depends on the phosphor used). One photo diode will be blacked with paint, both will be glued against the temperature stabilized tritium light. A LM335 sensor will be glued against the tritium light too. The tritium light will be held at a constant temperature of 45 C, plus or minus 1/2 C, heated by resistors.
I have made 2 designs for the preamp, and will try both to see what is the most stable, one a differential current amplifier with transistors, and one using CMOS opamps, both designs are truly 100% differential The PIC will use its internal reference of 1.024 V, if this was to prove to drift in temperature tests, then the PIC will be moved into the back box too. The rest of the box will be filled with Duracell AA batteries, and there will be a connection to the outside for a 12V AC/DC adaptor. There will be a RS232 connector so the PIC can be read out at any time to check temperature stability and data acquisition. Data will be stored to FLASH memory, with the option of an external to the PIC FLASH chip for longer runs. The box will be filled with more thermo padding material. All inputs and output will be RF decoupled, and the RF screening is why the larger alu box.
The thing will be placed in a place away from direct sunlight and extreme temperature variations.
Thermal stabilization requires some power, if I can make 100 C/W on the tritium light and preamp box, and have an average of 45 - 21 = 24 C over-temperature, than that means I need to heat with 24 /100 W, or say 250 mW. This is where the AC adaptor comes in, as about 1 Ah / 4 V batteries means
4 Wh, or 16 hours for the experiment to keep running without AC power. I do not expect a total of 16 hours mains down time in one year. Decisions, decisions. Parts are on order, sailing on.Things may still change at this point as the project unfolds.