Background: a set of 110.51 MHz filters were acquired on eBay per usual, and I ended up with around 400 of them. Part number was 110E39.
No data sheet was available, though a potentially related data sheet for 90 MHz filters in the same package was available from NDK.
Here is the response when used in a 50 Ω system with no matching:
It can be seen that this filter is approximately 40 kHz wide, with a set of spurs on the high side of the filter. As with the SAW filter design, any image frequencies should be set up to be below the center. Using a 21.4 MHz IF from this design with a low-side LO would give excellent performance.
Measurements indicate this filter is around 400 Ω in/out, but the amount of ripple and insertion loss is low enough for IF use that the matching components aren’t really needed.
I built a test board using matching components and while it did reduce the amplitude ripple, the total performance was worse. The high side peaks were also still there.