Doing my version of k6bez's antenna analyzer here (warning, pdf).
First, please notice that there are far more resistors on this board than there should be. I could not find 50 ohm resistors anywhere on my workbench. I ended up putting four 200 ohm in parallel. (Also used two 10k in parallel for the 5k resistors, and used a 680 ohm instead of the 648(?!?) ohm he calls for.)
Testing again, I get happier results. However, they still aren't what I'm expecting. I would expect a 50 ohm load to have a 1:1 reading, but is either 1.3 or 1.1 depending on what method I use to calculate (it should be (fwd + rev)/(fwd - rev), but I've seen people use the square of the individual values as well).
I saw some discussion on the Yahoo Group where someone was asking about the accuracy of the numbers. They talk about the nonlinearity of the diode in there, and I also noticed in his slides that he talks about needing more power for accurate swr readings. Maybe I'll make an amplifier next.
As a side note, something in my setup seems to be sucking down power. Nothing is getting hot, but my setup browned out while I was testing. Found the 9v battery I was powering this all off of was down to 7-ish volts. I put in a new one, and in a few minutes it was down a volt, too.