I think you are misunderstanding what “charging the clicker” means. It means that you create an association between the sound of the click and the treat, and you only ever need to do this once when you first introduce the clicker to the dog. I usually recommend my students do this 10 times. It helps them get acquainted with clicking, and helps them, as well as the dogs understand the concept.
Following this set of 10 click/treats (click and then treat, repeated 10 times), then you can start training. The puppy understands that click predicts a food treat, so that when he hears the click, he comes to expect a treat and starts to work for the treats. The click takes on secondary reinforcement value.
The click should be made at the time a behavior occurs. For example, if you want to mark a sit, click at the moment the puppy sits his rear end on the floor or ground. If you want to mark a recall, it gets a little more complicated, but basically you click for movement toward you. As Caroline said, every click must be followed by a treat in order for the click to remain reliable as a predictor of reinforcement.
Don’t be too quick to wean off the treats. Once the dog has learned the behavior and is performing reliably, I stop the clicking, but I continue to deliver treats on a one to one reinforcement schedule (a treat for each time the dog performs the behavior) for a while, depending on the dog, the behavior and the situation before changing the schedule to an intermittent reinforcement schedule (a treat given sometimes after a behavior is performed, but not every time). I will often add in secondary reinforcers such as praise and play before reducing the food. Food given on an intermittent schedule is very reinforcing and will keep the behavior strong.
Please see the following videos for more help. The first is a video introducing clicker training concepts, and the second is one in which I am teaching a dog how to sit using clicker training, so you can see the principles put to use. The third video may also be helpful; it shows how to correct behaviors using clicker training.