[Radio Wrangler]

If you're trying to get the narrowest skirt selectivity for a maximum acceptable amount of insertion loss, then the conventional planned responses like Butterworth ('maximally flat' but really means monotonic accelerating slope) ot Chebyshev/Cauer shapes are off the menu anyway. The best paper on this for filters with a finite resonator Q is by Seymour B Cohn. I can only describe the shapes his studies resulted in as rather erratic looking with a main peak, and some moderately prominent ripples (almost sub-peaks) into the transition region.

My last receiver design was an airband job for actual aircraft use, so it had to meet full DO-185/ED-23 requirements. The specs for avoiding intermod and overload from FM broadcasters got toughened up not that many years ago. I opted for a coupled pair into an RF amplifier into another coupled pair into the mixer. Each resonator of each pair being varactor tuned. With a large-ish tuning range and fixed coupliing components, it was an interesting balancing act.

Fortunately I'd gone for a fairly high IF in order to wangle a nice trick with digitisation and higher order Nyquist zone folding later on (29.25MHz) and for which crystal filters were available. So the image was the best part of 60MHz away.

Designing a receiver is an interesting puzzle.

David