I have the latest version of my PCB's ready to build. I added a trimpot for the bias on the prescaler chip to do away with the need for selecting a resistor, which is accessible through a convenient spare hole in the side of the chassis.
This way you can easily dial in the bias to optimise the tradeoff between stopping random counts with no input signal and sensitivity of the prescaler.
Here is the sensitivity of my previous version (pdf), once I have my latest version constructed, I'll make a better measurement of its performance.
It exceeds the HP spec by a bit up the top of the frequency range, so I'll have to figure out how to fix that.
And a 3D render of the design:
Jared Cabot