The following only really matters for active satellites with electronics on board, because if they get too hot or too cold, they can break.
Everyone knows that space is cold (~3K), but it can be hard to dump heat because you can only radiate it away. At the same time you also need to keep your heat while in shade. With the satellite being so small, you're pretty much screwed in terms of relative surface area, so storing heat is quite tricky.
So I conducted an analysis for which I will only post the assumptions and conclusions. There are lots of assumptions but they are all reasonable so I believe the conclusions to be fairly accurate.
Assumptions:
1) The satellite is 20 grams, extreme upper end of the range.
2) The specific heat capacity of the satellite is 1.5 kJ/kg/K.
3) The satellite is a cuboid with dimensions 1*4*5 cm (20 cm^3 volume, 58 cm^2 surface area).
4) The average frontal area is (20-4)/sqrt(2) + 4 ~= 15 cm^2.
5) The intensity of sunlight near earth is 1360 W/m^2 and the intensity of radiation from the Earth while in orbit is 220 + 100sin(theta) W/m^2 on the day side and 220 W/m^2 on the night side.
6) The satellite is in a circular orbit at ~ 275 km (90 min period, 40.8% of time in shadow of earth).
7) The satellite has a battery power output of 0.05 W (on the high end I expect).
8) The satellite is deployed at midday at a temperature of 290 K.
I'll spare the equations in this post, but if anyone would like them let me know.
I concentrated on the emissivity of the satellite, it's ability to emit (or absorb) radiation, as a passive method of controlling temperature. I have attached a graph from simulations of a satellite with ε = 0.2, 0.4, 0.6, 0.8, 1.0. Note that the emissivity of a material varies over the electromagnetic spectrum, however for simplicity I assumed a constant value over the whole range. The simulation was run for the first 9 orbits only, but as you can see, the temperatures got into a fairly steady pattern quite quickly. The vertical axis is temperature in Kelvin (but does not go all the way down to 0) and the horizontal axis is time in seconds.
As is clear from the graph, the lower the emissivity, the better, as it results in a far lower fluctuation in temperature. However, going extremely low would similarly be bad, as then not enough energy can escape and everything inside is slowly roasted thanks to the battery. Depending on the power of the battery, an emissivity of 0.05 is probably about as low as you'd want to go.
Handily, Aluminium foil has a low emissivity and is very light. Covering the above satellite (1x4x5 cm) with a single layer of Al foil would add less than 0.25 g, just 1.25% of the weight limit.
If anyone would like to suggest any improvements to the list of assumptions, I'm open to that. Additionally, if anyone would like me to run some simulations for their specific satellite, I can do that too.