CCS/CC1350: Miniature Helical PCB Antenna (DN038) impedance matching

I would assume that most of the difference is due to the fact that you use a smith chart calculator that uses ideal components. A physical component will not follow the ideal lines in a smith chart, hence measuring on a PCB will give different values than calculating with theoretical numbers.  

Hi TER, thank you for your answer.

I agree that the nonidealities will change a little our results. However, changing from (1.42 + j54.4) to around (50 + j0) seems not realistic since we should be working far from the component's resonant frequency. We also changed the values of the components and noticed that it didn't help to get closer to Z = (50 + j0) Ohms.  

On the other hand, we observed that just by inverting the matching circuit topology (C parallel and L series) we reach easily a Zin = 50 Ohm, as shown below. 

Which matching circuit would you suggest, in this case? Do you have any information about the DN038 impedance on CC1350?

Thank you,

We always recommend using a pi filter topology for the matching since by populating either the first or the second shunt you can cover the full smith chart. You will see this in a lot of our reference designs using an antenna. 

Yes, I understand that pi filter topology is very useful to match the antenna impedance. 

However, we should design the main circuit seeking to match the theoretical antenna impedance. 

I simulated the circuit with the values from "LAUNCHXL-CC1350EU_1_3_0_Schematics" and the matching got worst. So now I'm working with 3 possibilities:

1 - The point indicated with an arrow below is not 50 Ohms. This would imply the connection with P11 will not be matched if connected to a 50 Ohm coaxial cable, which is strange.

2 - The antenna impedance is far from (10 - j88) as indicated in the antenna Design Note http://www.ti.com/lit/an/swra416/swra416.pdf). 

3 - There is a problem with the matching circuit topology. This circuit doesn't help to perform impedance matching it is supposed to do.

Is there any other real possibility I'm not taking into account?

I would really appreciate if you could help me in this case. 

Thank you,

You write that you have simulated, does that mean you have done a simulation with component models from the vendor and the real PCB or are you still operating with ideal component values? 

DN038 uses a rather small PCB, the launchpad is a larger PCB. Note that the length of the antenna used in DN038 and on the launchpad is different (to adjust the resonant frequency) Board #6 from the antenna DK 2 has a size that is somewhere between DN038 and the launchpad.

As you can see from the measurements from DN038 and on #6 in http://www.ti.com/lit/an/swra496a/swra496a.pdf the impedance of the antenna is  different dependent on the board. Hence we don't have any theoretical matching values. The antenna impedance has to be measured on the board in question and the matching values found according to this. That is why you see placeholders for a pi filter before the antenna in our reference designs since it allows us to match to any impedance. 

Hi TER, thank you for your answer,

I'm still working with ideal components. I was expecting, however, to see some better results than with no matching component (as in the "New topology" mentioned in a comment above, where the matching circuitry got closer to 50 Ohms). 

Based on your answer, I understand that we don't have any approximation for the impedance of this antenna, even with a large margin of error (due to the fact that different boards with different characteristics can be used). Is this right?

If so, is the VNA the unique solution to perform a good tune of the matching circuit?

Thank you,