Impedance matching of ADL5391 output

Guys from AD engineer team, please, help me with this problem!

It looks like the S-parameters on the product webpage has S22 included.  Please see the Readme.txt file.

The output circuit should load the output with 50 Ohms differential between 1 & 2 GHz. 

I think the evaluation board is matched better at the lower frequencies; you cannot have a perfect match at all frequencies when the Input/Output Z is changing over frequency.

Thanks a lot for your responce!

I've found a zip file with S-parameters on the ADL5391 webpage.

What does it mean  "the output with 50 Ohms differential"? Should I replace R4 resistor from 100 Ohms to 50 Ohms and then output network suits more for 1-2 GHz bandwith? Are there any principles how to adjust output resistors R4, R5, R6 for users application?

Please, correct me if I'm wrong: evaluation board is matched better at the lower frequencies only because of coupling capacitors C7, C18, C4, C20, C14, C2. The evaluation board will become matched better at the higher frequencies if I replace them with more suitable for my bandwith (1-2 GHz).

Thanks a lot for your responce!

I've found a zip file with S-parameters on the ADL5391 webpage.

What does it mean  "the output with 50 Ohms differential"? Should I replace R4 resistor from 100 Ohms to 50 Ohms and then output network suits more for 1-2 GHz bandwith? Are there any principles how to adjust output resistors R4, R5, R6 for users application?

Please, correct me if I'm wrong: evaluation board is matched better at the lower frequencies only because of coupling capacitors C7, C18, C4, C20, C14, C2. The evaluation board will become matched better at the higher frequencies if I replace them with more suitable for my bandwith (1-2 GHz).

The output of the ADL5391 will "see" 50 Ohms differential when looking into the R4, R5, R6 and the TC1-1-13M (balun) network.  The balun is a 1:1 impedance ratio so it presents 50 Ohm to R4, R5, R6, when J1 is loaded with 50 Ohms.  The whole network (R4, R5, R6 and the TC1-1-13M) makes 50 Ohms differential across WPLS and WMNS.  Note that this does not match the output to 50 Ohms, but rather loads the output with 50 Ohms.

The capacitors that you mention are for coupling into and out of the part, and decoupling/AC grounds for the baluns.  They don't really do anything to help the matching; their impedance just decreases as frequency increases. 

Your answer clerify many things, thanks! So, as I see, evaluation board is "matched for lower frequency operation" because of internal structure and properties of ADL5391. And if I'm right it is impossible to achieve better gain on high frequencies (Figure 6) with another input and output matching networks.

Correct me, please, if I'm wrong.