I finally managed to implement renormalization for unequal port reference impedances. The images show unequal resistances for a 10.7 MHz Murata ceramic filter. I optimized them by hand for minimum group delay variation over 10.6 to 10.8 MHz, the passband of an FM broadcast signal. The optimization also flattened the passband and lowered the spurs.
I'm not quite ready to release this. The new renormalization algorithm I'm using can also handle a reactance shunting either port. This would let you include the standard 10 pF Murata test load. I think it might be useful for other kinds of filters and transformers as well. The problem is that I'm running out of screen real estate, at least on the bottom line. I'll figure out something. In the meantime, would there be any use for a shunt inductance? I don't recall seeing this, but it would be easy to implement.
Brian
