[Jon Stratis]

I think the key is to limit the people who can purchase items for the team. On our team, we have a few mentors who make almost all the purchases for the team (although occasionally students will want to do it, and they can buy some specific things from local sources). It's done this way for a few reasons. First, we can usually get stuff at discounted prices by ordering through our suppliers at work. Second, it ensures that we don't over-spend. All of the mentors on the team are pretty thrifty and know how to spend on a budget, while students can often have trouble with that.

At the beginning of the season, we prepare a rough budget - we know going in, for example, how much the bus is going to cost, how much the hotel is going to cost, etc. We can also look at our inventory and past experience and judge how many spikes, victors, jaguars, or motors we may need to buy. We may be right, we may be a little off, or we may be way off. But it's a rough budget that allows us to say (numbers are made up) "we'll spend around $400 on electrical components, $700 on mechanical components". That can all be done before the kickoff.

After the kickoff, we spend a few days developing our design, and a week or so prototyping with whatever material is on hand. During that time, you can really nail down the numbers - you can figure out exactly what motors you need to buy, how many motor controllers, approximately how much stock aluminum, etc. Normally, you're going to have a number at this point that's lower than your actual spend for the build season - to be safe, budget yourself and extra 25% for robot parts, and use whatever is left over for a team dinner at the end of the season.

One of the keys here, though - budgeting is just as important as everything else you do, and it goes hand in hand with your design/prototyping phase. You may come up with the best design in the world, but if you can't afford to build it, you should get started on the second best design that you can afford.

On an interesting note, this is how companies do it, too. Typically, you have a small budget for research/prototyping, and when something shows promise, you get a budget approved to develop it. Where I work, we call that going to contract. Once you go to contract, you have a due date, a fixed budget, etc... You can't decide (at least, not very easily) halfway through the project that you're going to need more money.