LMH5401: LMH5401

Morning Anton, 

Kind of an odd question? Your input balun quarantees no input CM signal to worry about. Using only one side of the FDA output will show a signal, but you have probably lost your HD2 performance - need both sides to hold 2nd harmonic down. If you need a single ended version of your differential output consider something like this - your input Balun will be the bandlimiting point probably, while this approach is very broadband,

You would adjust the resistors to what you want in this topology, 

Hi, Michael!

Sorry for misrepresentation. Г can't use Balun, because i use signals started with very low frequency (about 100 Hz). I know about HD2, there was an answer about reducing HD2 by using only one output.

It's interesting to me what will happen with CMRR if there's for example A*Sinwt and A*Sinwt on the both of inputs of fully differential amplifier and using only one output of it as on new sheme.

Well your drawing shows an input balun, so perhaps you mean to say what if I drive a centertap on the input balun secondary? Well regardless of how you use the output signal, the internal CM loop will reject an common mode input from DC to >1Ghz. So if you imagine a centertap generated input signal, you can run a TINA sim with that as input signal and trace out the rejection to the output. 

Well yes, if you drive both inputs together, the common mode loop will do its best to reject that at the output pins. So there should be a very small residue of this input increasing with frequency as the CM Loop gain rolls off. 

And it doesn't matter, if i use only one output? 

Well the sims are not promising, measuring one output gives only 17.9dB rejection. 

If I test for the output CM, I actually get less rejection? I can't find any specs or plots that are helpful for this - might be a modeling error? 

I tried a different FDA with a different modeling flow, this is what I expected to see - here is single ended output 

And then CM output 

Not much change going CM output, I have to believe the LMH5401 model is not right here, but the part should have this kind of behavior, 

I got next results:

Thank you, Michael.

I simulate this process in Tina-Ti and LTspice too.

There is just one question. Should we believe simulation result or really there will be much more less CMRR when only one output is used.

That looks better, wonder why I could not get that result - too low of R's? Not sure of match to model, I think it would be easy to try on the EVM board. 

I tried your scheme. Got next results. Don't know either, why your simulation gave those results.