A metal plate in LEO will cycle from –170°C to 123°C depending on its Sun face and its time in sunlight. If your picosatellite is spinning, this will even out the heat distribution a bit, but that’s the range to assume. An orbit has approximately half its time in sunlight and the other half in Earth shade, so the temperature behavior is worth modeling.
Since the picosatellite is spinning, this range is fortunately smaller (as heat has time to distribute and dissipate), and with a 90-minute orbit, you should cycle through three ranges: too cold to register; transition regions where the sensor returns valid, slowly changing data; and possibly oversaturating at the high end. You can add a heater if necessary—satellites have used heaters and coolers depending on the instrument and facing.
Therefore, a thermal sensor (like a microDig Hot brand sensor) that covers –40°C up to 100°C will suffice. The range of –40°C to 100°C is a feasible area to measure. In any event, past that range, the rest of the satellite electronics may have trouble.
Similarly, a light-detecting sensor, for a spinning picosatellite, is likely to return just a binary signal: super-bright Sun in view and Sun not in view. So all that it will measure is the timing of when the Sun is in view. The function of the light sensors will be largely binary, to catch Sun-dark cycles as it spins, as well as the overall day/night cycle of the orbit. If there is a slight tumble to the satellite, all the better. These light sensors will provide a basic measure of the satellite’s position and tumbling. If you want to measure actual light levels, your design will have to ensure the Sun doesn’t saturate your detector.
The ionosphere has a field strength on the order of 0.3–0.6 gauss, with fluctuations of 5%. For a polar orbit, you’ll have higher variability and higher magnetic fields than an equatorial orbit (as the Earth’s magnetic field lines dip near the poles, hence the auroras). If you want to measure fluctuation, not the field strength, you need to capture 0.06–0.1 gauss signals.
A default Hall effect sensor tends to be designed for Earth work and measures tens of gauss, so you need to ensure the sensor is calibrated for the space environment, not Earth’s surface. You can’t measure below 0.63 gauss on the Earth’s surface, because its intrinsic field will create too much background noise. However, that’s not a limit of the sensor. A $10 Hall effect sensor plus an op-amp could measure variations down to as low as 0.06 gauss if there’s no large external magnetic field. Below that, the noise from your sensor’s circuits, not the sensor, will likely be the limiting factor.
Hopefully this overview gets you thinking about some things you can do with your picosatellite.