dual band antenna impedance matching with lumped elements help

Hi all,

I need some help about antenna impedance matching. I have a design with a dual band antenna which is fed by the output of a power amplifier. and from amplifier output
to antenna there are pads for 3 series elements and for two shunt elements. Also path from the power amplifer output to antenna is a little bit long. I want to match this antenna
using lumped elements. So, first of all i measured S11 parameter of the antenna itself by soldering a coaxial cable just in the entrance of the antenna.(Also i disconnect the rest of the circuit that lies behind this coaxila cable)
Then by considering the output of PA as 50 ohm, i tried to design a matching network in AWR Microwave office. I modeled the antenna as a subcircuit in AWR with data file that is exported from our network analyzer. But the problem here is for example i am adding an inductor in AWR to achieve a better S11 result but i cannot see the same effect of this componenet
in reality. Although i am taking into account SRF and tolerance of components, i cannot have a correlation between simulation results and real measurements. I am not sure but i think length of path and presence of discontinuties in components pads effect the matching drastically.
In this point, I am just wondering that if I am performing matching right? If not what is the correct methodology to perform a dual band matching with lumped elements? If I run a EM simulation in AWR can i obtain more realistic results?

Thanks.

Javier.

Seems relative correct done but when calculating a matching component should antenna S11 measurement be done at the place for the matching component due to the electrical delay. Else will electrical delay be wrong.
If it is hard to solder measurement cable at component location, is it still possible to measure correct, by doing in place calibration where the component is intended to be, while measurement cable is soldered somewhere else along the trace.
To design a very precise network for impedance correction must a new reference be done for each component if traces between these components adds reactive or capacitive stray-losses. It can in most cases be seen in VNA if that is something to take in account. Additional phase delay between each component are in most cases low but it can be worth to recalibrate as it is easy to do.
At higher frequencies type 2GHz and upward can a medium bad PCB RF design add relative high losses in ground-plane which then must be added in serial with shunt components to get expected result for an inserted component value. Component value = reliable S21 data for actual component.

Thanks Kafeman,

I am gonna follow your suggestions and then share my experience here.

If you have a VNA with GPIB/LAN/USB interface and a PC available, is this kind of job real easy if you use a software type AnTune.
It adds semi-automatic electrical delay setting for VNA's that not have this function, and optimizes network impedance based on real components.