I would like to use LTSpice to model an ADC drive circuit with a center-tap transformer. However, my attempts to use the recommended coupled-inductors method have left me confused. The circuit takes a single-ended 50-ohm input. The load is a differential port on my ADC with 100 ohm input impedance. The goal was to use a center-tap transformer with a 2:1 impedance ratio to match the load to the source impedance (I need the center-tap for a bias voltage, which is grounded for simplicity).
Stripping things down to the core circuit, I have the following model:
Transformer_2to1_verify_2to1.asc
I assumed this would produce an effective load impedance at Vt of 50 ohms (due to the 2:1 ratio of the transformer), which means I expected the voltage at Vt to be ½ the input. But I got a ratio of 15% (which means the effect impedance is about 8 ohms(?)). I’m not sure why.
To ensure I’m not totally off base, I created this simplified model without the center tap. This produced the expected result with the voltage at Vt is ½ the input voltage (after the startup). 
Transformer_2to1_verify_simp.asc
So, what is wrong with my center tap model? I can get something close by fiddling with the L2 and L3 inductance values or the coupling coefficient. But this is just stabbing in the dark, and the required values are odd. I assume it’s my model, but I’m unsure how to fix it.
