The Double Crystal Lattice Filter in the Swan 240 -- Smoothing it out with a NanoVNA

In SolderSmoke Podcast #234, I said that I was scrutinizing the filter from the Swan 240 that I had picked up around 1994 in the Dominican Republic.  I cannibalized it out in the Azores in the early 2000s and used the parts to build -- among other things -- my first SSB transmitter.  I never really focused much attention on the filter that I pulled out of that old rig -- I was just happy that it seemed to work. But I am now older and wiser, and I have some test gear that lets me look at the passband of that filter. 

First, take a look at what it is supposed to look like.  This is from the manual.  Yikes!  That passband looks far from flat.  I can almost hear homebrewers around the world shrieking in horror and disgust.  

Above is a description of the filter, and the schematic, again from the manual. 

Here is what my extracted and somewhat re-built filter looked like in my NanoVNA (more shrieking!).  The dip in the passband is a lot worse here -- it looks like 10 db vs. 3 db in the manual.  This is probably because I'm not even attempting any impedance matching on the filter -- it is just seeing the 50 ohms in and out of the NanoVNA. 

Here is my 2002 attempt to rebuild the filter and put it in my SSB transmitter, along with my more recent attempt to flatten the passband.  I no longer had the adjustable coil L8, so I made my own coil based on guidance from Ben Vester W3TLN's January 1959 QST article on "Surplus-Crystal High-Frequency Filters." (Ben had an early influence on Pete Juliano's tube-rig  designs.)  In the picture above I have 1k pots between the filter and the input and the output of the NanoVNA, as described by Nick M0NTV. 

Adjusting the 1 k pots, I could smooth out the passband quite a bit.  Measuring the pots and adding the 50 ohms of the NanoVNA, it looks to me like this filter is smoother with about 280 ohms at the input and output.  I may build two matching networks or some transformers. Some TIAs may also be needed.