For reliability, the best is to either work in a controlled room or in a controlled tank. It will just take more time at the beginning for your aquarium than your room to reach the desired temperature (inertia), make sure you mesure the temp in the aquarium as well. For a controlled tank: place your "work" aquarium in another tank, bigger obviously, and you control the temperature of that tank. In both cases, your aquarium will be less sensitive to temperature fluctuations if something happens than if you control the temperature directly in the aquarium, especially if that aquarium is at another temp than the room. That's for the reliability part.
I don't know much about making your own reliable cooler: it depends on the temperature difference with ambiant temperature and on the length of your experiment. You can indeed work with colder water or ice, but that's not going to be reliable on a long experiment and you'll have to put time in it to control your temperature.
You could use the cold water from your household water supply. Just add a T to the cold water pluming, make a spiral to place in the aquarium, connect the spiral to the T with a solenoid valve and a thermostat. This should give you sufficient thermal control for this temp range. Just make sure that there is enough circulation around the spiral.