Hello Robert and Merry Christmas to you and yours,

The first thing I would check is if D1 is installed correctly.  Of course if it is in there backwards, I would expect you to read a somewhat normal diode voltage drop from the junction of D1 and L4 (easily measured on the exposed lead of L4) instead of the varying voltage needed in order to change its capacitance.

So if you measure the voltage on the junction of D1 and L4 and it does in fact vary at that point from the amount of your supply voltage down to the .6 volt diode junction drop of D2, then it must be installed correctly.  You can check for cold solder joints by measuring the resistance from the actual lead of the part to the pad of the next component it connects to.

You can also remove D1 and connect any old air variable capacitor you may have in its place and see what difference that makes

The frequency swing of only about 2 KHz suggests the behavior one would expect if there were only one crystal connected.  So if D1 is installed correctly and replacing it with an air variable doesn’t change things, then I would look carefully at L4 to be absolutely certain you don’t have the 1.2 uH coil installed there instead of the needed 18 uH. 

If that is OK (and you could remove it and measure its inductance if you have that ability), then I would check to be sure that the leads of each crystal are actually making good contact with their traces so that they are in fact connected properly in parallel.  Again, it is behaving like what I would expect if only one crystal was connected.

When paralleled crystals are used in a VXO, it would seem that the motional capacitances are multiplied instead of simply adding and that is why we are able to pull the frequency of the paralleled crystals so far away from their stated frequency.

The above should give you some starting places at least.

73,

Wayne  NB6M