If you wouldn't mind, I would like to draw your attention to my latest post on the Etherkit App Notes blog. In it, I detail how to create a 10 milliwatt WSPR beacon using nothing more than the Etherkit Si5351A Breakout Board, an Internet-connected PC, and a low-pass filter. A simple project, but one that gives quite a bit of fun testing the ionosphere given the cost and complexity.
I don't want to take away from the post, so I will advise you to go there to read it, but the bottom line is that with about 10 mW, I was able to get a signal decoded over 2000 miles away. I remember reading the old exploits of the QRPp gang in books like QRP Power, where you had to be really dedicated, organized, and good at decoding CW in the worst conditions. Now, we have the luxury of a mode like WSPR, which lets us do milliwatt propagation experiments without breaking a sweat.
One idle thought I had about this is whether it would be feasible to put this transmitter on the 13 MHz HiFER band (check out Dave AA7EE's excellent treatment on the matter) and whether that would be something that would be fun and useful for schoolkids to experiment with. Of course, it's technically feasible, but I would want to be sure that 1) it's legal and 2) there would be interest in doing it. A single PCB could be made with one Si5351A output attenuated to around 4.6 mW and low-pass filtered for transmit, while another output could be used to drive a simple fixed-frequency receiver based on the SA612. Let me know what you think about this in the comments.