Re: VNA calibration on board calibration kit
I have been making some test. Here the results for actual pcb, actual with a cooper tail connected to gnd and TI evaluation board:
The evaluation board has a little more gnd. It is not all the surface full of gnd. Can i add something to simulate a larger gnd in the same size?
It seems that pcb thickness also has some influence but i don′t understand why.
Good weekend!
Variations of PCB thickness and antenna distance to ground are in this case just causing minor variations on a much bigger problem.
Reduce as much as possible of any error sources. Some of the thing I had done in a similar situation:
Replace whole coax-cable. They all get broken sooner or later but half broken can be hard to detect.
Same for connectors. Even SMA connectors do not last for ever.
Use a empty blank PCB in similar size as current ground plane, no pattern, no components.
Solder coaxial-cable shield directly on PCB.
Solder antenna directly on coax center wire.
Antenna will be hanging in air but that is ok for testing.
If needed can iron wire be used between antenna and coax.
Soldering should be "wet" and not to small contact points to ensure good RF connection.
All this will cause different antenna effects such as center frequency and impedance will change some but what you are seeking is to find which of above actions that makes RL changes form almost straight line to a clear dip around -20 - -40 dB at center frequency, whatever the center frequency is.
The S11 values achieved with an extended ground plane are poor compared to expectable values, or e.g. the λ/4 rod you previously tested. It still looks like a bad antenna or some other defect.