Inside the FSS box, the FAST path has a ~10 Hz pole made up from the 15k resistor and the 1 uF cap before the output connector.
This should be moved over to the output of the driver to make the driver happy - without yet measuring the high frequency response,
it looks like to me that its becoming unhappy with the purely capacitive load of the NPRO's PZT. This will require a little surgery inside
the FSS box, but its probably justified now that we know the Thorlabs box isn't completely horrible.