Is there anyone monitoring these groups/reflector who knows the deep insides of filter design theory?

I've just synthesized a straightforward 50- 75MHz 5th order Top-C-coupled  bandpass filter to form part of a frequency converter that will feed an SDR-IQ digital receiver.   While playing with the filter design using a circuit analysis package, and recalling satellite microwave cavity filter designs from the dim-and-distant past, I remembered that by arranging cross-coupling between non adjacent resonators, it was possible to form pairs of real zeros in the upper and lower stop bands respectively.  These convert the filter to an elliptic type response, and considerably improve the cut-off.  I also recall it was a very critical thing to set up.  

So, I tried adding an additional cap between resonators 2 and 4, got a wonderful 10 - 20dB improvement in the low end cut-off response around 40MHz exactly where it was needed,  BUT ONLY THE ONE REAL-ZERO COULD BE FOUND !!!  There is absolutely no sign af any second null above the passband.

Where's it gone?  Shouldn't they always come in pairs?  This is really bugging me, I can't sleep at night, can anyone explain? 

Could it be because a top-coupled - C  BPF is non-symmetrical?

Full circuit details can be found at    http://www.scrbg.org/g4jnt/50-75bpf.pdf

Andy  G4JNT

www.scrbg.org/g4jnt