I tested QPL9504 on evaluation board, and it seems that it does not perform as expected. I checked suggested value for 5-6 GHz, and calculated my own matching network, and components values are very different. Maybe, because of different matching network topology. You will find copy of my calculation below:
To match output of the device at 5.GHz, I used an L-section matching network consisting of a 1.37 pF shunt capacitor and a 1.02 nH series inductor (or alternatively, a 0.71 nH shunt inductor and a 6.30 pF series capacitor) - for low-pass (or high pass) configuration.
I also noticed that you suggested to use 18 nH 0603 inductor L1 on evaluation board, optimized for 5-6 GHz. It is Coilcraft p/n 0603CS-18NXJRU. However, datasheet for it specifies min self-resonance frequency 3.3 GHz. This means that on 5–6 GHz this inductor may have capacitive character, and it will not work as an inductor anymore. Did I miss something?
As about an input matching network on the evaluation board, optimized for 5-6 GHz - did it optimized for the minimum loss or for the minimum NF?