I do several $50 (US $ or equivalent Indian currency) experiments. Mostly it is asking some student to do a small task, build a prototype, learn something and testing to see how fast they learned. In every experiment, I start with one or more hypotheses and get some data that goes towards verifying these hypotheses. The $50 does not include the time I spend on the experiment.
When the experiment is successful, I continue with several others and start increasing the amount and time, but always make sure that I go through this ask a question, postulate a few hypotheses, test them and learn some thing.
This is one of the costs of my education of learning by doing something. The other major cost is the books I buy and the time I spend reading and thinking about them.