Thanks for your reply. Looking at the photos in my original post, I looks like the wheels were not in the same position.
I do not have any unpainted five spoke wheels, so I did the following experiment with a three spoke wheel.
Look at the photo below, from left to right and top to botton.
The first photo is with the wheel as far away from the encoder without falling off the shaft.
The second photo, it is a little closer.
The third photo, it is just about right. The forth photo is with the wheel removed.
The point I am trying to make is that with the wheel on the shaft in any position, I cannot achieve a 4v high output, yet with a painted wheel that has narrower gaps between spokes, I do achieve it. There must be some surfaces on the wheel that are providing some low ambient light form the IR source. That's what the paint is eliminating.
What do you think?