Biquad antenna center frequency and bandwidth problem

I was hoping some one on here knew which parameters effect bandwidth and center frequency of a biquad antenna the most.

I created two biquad antennas using the many how-to guides on the internet most of which reference Trevor Marshal’s design. I also used the following site which provides a better analysis of the important parameters. Biquad WLAN antenna analysis. I measured the biquads with a network analyzer using the S11 parameter and found that they were terrible because all though they had extremely high gain, it was centered around 2.25ghz and had a 3db bandwidth only about 50Mhz wide or less.

Biquad #1 was not made as precisely and has driven element pattern with sides that are 30.5mm+/-2.5mm per side, reflector to driven element height is 16+/-1.5mm and spacing between center conductor and shield appeared to have no effect on center frequency/bandwidth. This one was made with a less than an ideal reflector about 105mm by 110mm. Center frequency is 2.15-2.2GHz

Biquad #2 is constructed better with 30.5+/-1mm max per side tolerance for the driven element, reflector to driven element height 16+/-1mm and reflector is +/-1.5mm from 123x123mm. Center frequency is 2.25GHz to 2.3Ghz

Wire used for construction is 18gauge which is a bit less than that recommended on the internet which is closer to 13 gauge. Both are constructed with lips that are within 2mm of the ideal 29mm. Driven elements are held in place with a nylon spacer at each corner glued to the FR4/copper reflector board with hysol ea934 (dielectric constant around 1k at 7)

1)Any ideas why the center frequency is so far off and/or how to increase the bandwidth?
2)Has any one else constructed this type of antenna using the dimensions above and measured bandwidth/center frequency and found a different result?
3)Is the ideal spacing between the driven element/center conductor feed point and shield/return determined by the size of the wire used to construct the driven element and should be adjusted to achieve an impedance match with the coax/transmitter(does it even matter assuming spacing is within reason)?

length of the driven element affects resonant frequency. shorter = higher.

diameter of the material used determines bandwidth. fatter = wider. copper pipe good; 22 gauge bad.

Thank you for the fast response!