I received this clock as a gift in 2002, and in 2015, I was noticing weaker output from the transmitter and then eventually the outside temperature was not updating.  The WWV radio update (atomic operation) and the projection of time and temperature is a very nice feature we like.

The transmitter is sealed, and I started by sawing with a thin serrated blade along the case seam.  There were no inner attachment points, and the case opened up at that point.  The front and back view of the board are shown below.

Referring to the back of the board, it is easy to see that the large trace on the bottom of the image and the square at the top is the ground reference plane.  The processor is clearly encapsulated in the black blob, and the RF transmitter is at the top.  This latter module is controlled probably by the crystal in the can on the front of the board.  There are fortunately only two traces that cross into the square ground plane at the top, which is power and serial output.  There are three adjacent round holes in the middle of the board, and these are power (left on back), ground, and serial output (right on back).  

Putting the signal on the serial output on the scope shows that the following data protocol (at room temperature).

• There are three groups of data transmissions.  Each last about 35 msec with a 20 msec pause between them.
• Zooming into each group, it is clear that all the 'low' periods are the same duration, and there are two 'high' durations.  This is similar to the Sony SIRCS protocol, and so we can characterize a '1' bit is a long high and a low, while the '0' bit is a short high and a low.
• The timing of the '1' is  

Cleaning the whole assembly (especially the battery contacts) and then reassembling showed that the transmitter worked fine.  As a mod, I soldered a few inches of wire to the top terminal on the back as an external antenna.  This mod appears to make the signal strength much better.