Matching two ports on CPW (Coplanar Waveguid) printed circuit board

[HS]

Matching two ports on CPW (Coplanar Waveguide) printed circuit board

Hello,
There is a nightmare called CPW pcb design.
I have an FR4 board (1mm thickness) and I need to match two ports.
One port is 50 ohms and the other (antenna) is 23+j38 on 2.4GHz ISM band.
Of course I need to introduce some attenuation on harmonics too: 4.8GHz and 7.3GHz.
Apparently simple for someone using Ansoft Designer and optimizer, but this is a CPW pcb and
I am using two small semirigid coaxial cables to measure this board build with 0402 components.
Help me !
How to calibrate the entire setup ?
Option 1: I have an E5070B VNA which has the option of port extension and of course electronic calibration box too.
But using this option is not straight forward as it was using it on two layers pcb boards with ground plane.
Placing an inductor or a capacitor doesn't perform on the Smith Chart as expected.

Option 2: I did the calibration manually at the end of the semirigids using open, short and a 50ohms resistor to gnd as a load.
Still, the same behaviour. Unrealistic insertion loss, strange behaviour on added components.

Please tell me how to wake up and get smart on tuning this CPW pcb.
Any link on Internet or an application note somwhere which tells the steps to follow ?
Thank you in advance !
HS

 

[jiripolivka]

The problem I can see in that all signal sources and detectors used for calibration are matched to 50 OHms.
Therefore the best way to calibrate other structures like yours is to get good transitions from CPW to coaxial 50 OHms, then you can calibrate and tune your CPW devices.
Developing signal sources, detectors and couplers in CPW is possible but ,much harder way to go.

 

[E Kafeman]

Most cal.kit are of the type "offset calibration" which add internal delay and capacitance that then is compensated for with coefficients stored in VNA. Doing in place SOL calibration at PCB, using wrong coefficients and you can get any reactive readings, even outside of Smith chart.
It is a bit tricky as SOL calibration selecting wrong cal.kit coefficients still can result in perfect VNA readings when calibration is checked, but reactive loads gives wrong results.

Select or define the correct cal.kit coefficients in VNA.
Solder coax at PCB.
Cut traces in both directions, leaving you with a short trace and a component pad as the only connection at coax center pin.
SOL calibrate with a 50 Ohm SMD resistor. Verify after calibration that electrical delay is correct.
Verify calibration with reactive components around 0+/-50j. Check in a S11 file for the actual type of inductor or cap what to be expected at actual frequency. Data can be found at most cap/ind manufacturers homepage.
If it gives incorrect reading, check VNA manual how to setup user defined cal.coeff.
www.antune.net/calibration.html can maybe give some practical tips.