I have this circuit matching.
The antenna is ANT3505B002TWPENS and the matching recommended is:
but, in my design I put the matching components with a PI configuration because I think is the best choice to match the 2 frecuencies (900 and 1800).
In first time we match 900Mhz with series C and parallel L (because this configuration have low influence at 1800 Mhz):
In second time we match 1800Mhz with parallel C and series L (because the this configuration have low influence at 900 Mhz):
And I remember that we can move across Smith´s chart with the next image:
Ok. We begin with our matching. In first time we calibrate our VNA (I have miniVNA Tiny).
I calibrate the short with 0 Ohm resistance at L8 and open L6:
I calibrate open with L6 open and L8 open:
And I calibrate 50 Ohm with a 50 Ohm resistor at L8 and L6 open:
Ok. Now we measure how is my circuit (shorting L6 and opening L8 and L9) and we have:
Point  is 900 Mhz and point  is 1800 Mhz.
My first impression is very bad because the 2 freceuncies are far of the center.
From now on I will put the exact value of the component to correct match although I solder and desolder components hundred of times to find the correct.
Now We begin with 900Mhz:
We have the 900Mhz frecuency at center, now we are going to match the 1800 Mhz.
Remember we can only use:
Because another configuration move the 900 Mhz frecuency and unmatch it.
And now I need to add an L in series to move the 1800Mhz at center of the circle, but, I don´t have any space to solder it with PI configuration.
I need to change my circuit matching configuration by this?
Why this data of the correct configuration by different frecuencies don´t tell at the datasheets of the antennas? I feel stupid because I design a complete circuit and now I discover that all is wrong.
More people make this mistake? With a PI configuration we can obtain a good matching, but, never a perfect matching?
Now I´m going to redesign the circuit and I need your advice J
Best regards (sorry about my english).