I need to convert a ±4.5V single ended signal to a differential signal for an ADC input (probably the LTC2107). Since the input Signal has a -3dB bandwidth of about 50MHz the ADA4945 in high power mode should be sufficient.
The Vocm is set to 1.2V and the gain to 0.267. The Diff-Amp Calculator gives me a decent solution using Rg=280R and Rf=75R. I am using 5V single supply (although a ±5V supply is also possible).
Now when I do simulate the circuit in LTspice the output is not what I would expect: the output should be symmetrical and centered around 1.2V. This is only the case if I do not GND-reference the input voltage source. Also it gets worse at frequencies above 4MHz. See images below.
-is this behaviour also to be expected in reality?
-should I use the ADA4930 instead or will it have the same issue (it does show a similar behaviour in the simulation, just starting at higher frequencies)?
-What is the correct way to simulate the SE-DIFF conversion to get around this issue ?
LTspice model with GND referenced input
Transient output at 20MHz ±1V input
AC analysis: outputs with GND referenced input
AC analysis: outputs when removing GND at the input