Placing the transmission line transformer on the input or output makes no difference to the balance. All it does is act as a common mode choke. However on the input side the differential mode impedance is fixed at 50 ohms. On the output side it varies, and contributes to the impedance curve.
?
The other issue is the low value of coupling capacitance (in a 50 ohm path) at the low frequency end of the frequency range. The amplifier measurements show this clearly, with a rapid decrease in ?S21. Real loops don't behave in this way.
?
I still think an air wound inductor, brought to resonance with a parallel capacitor, and loaded to present the correct coupling and overall Q, by means of a closely coupled overwinding, provides a more accurate impedance curve and S21 plot, plus it is already pretty well balanced, but an extra choke balun, on the 50 ohm side, could help at the higher frequencies.
?
The amount of coupling will "tame" the impedance excursion near resonance, so that is more like a actual loop. If not a parallel resistor could be added, to simulate other losses, such as radiation.
?
Just my thoughts...
?
Regards,
?
Martin
?
?
