Loose Coupling for a coaxial resonator

I am trying to loosely couple a coaxial resonator using two copper ribbons. My line is shorted on one end as shown here: http://www.rfcafe.com/references/ele...-resonator.htm . The lines loops are fed through a BNC to and shunted to ground near the center conductor of the coax (my dielectric is air) using a 2 port network analyzer.

Also very similar to:
http://www.amalgamate2000.com/radio-...resonators.htm

I use a set of capacitors and inductors shunted on the open end to adjust the lines frequency and Q. Problem is when I change the length of the coupling loops and with the same components I get different frequency and Q measurements. I?m questioning the validity of f & Q since the coupling alone appears to change the lines behavior.

How do I know if my coupling is optimal (?real?)? For example, I can cut my ribbons lengths 1, 2, 5 and 10? long and they all change the Q of the line proportionally. 1? has the lowest Q where 10? has the highest Q. I?m assuming the shortest length is the loosest coupling where the 10? has the tightest coupling (much more flux is absorbed by the coupling hence lowing the Q).

I've seen some improved designs (and shapes) for the coupling loops which claims better performances. The loops are connected to the ground on the other side of the connector, near the walls. Worth to try.
../imgqa/eboard/Antenna/rf-2nw14h2slgu.jpg
https://www.robkalmeijer.nl/techniek...e12/fig02l.jpg

Hi vfone,
What metric do you think I should use to determine the 'better performance'?

Better performance means a good compromise between higher Q, greater rejection, less insertion loss.