WiFi (2,4G) Antenna Isolation. Why 3 capcitors version does not work

Question

I need to isolate the antenna from the WiFi transmitter. There are some solutions on the market but none suitable for my application.

So I have started some experiments.

1. 2x PCBs creating DIY capacitor (separated by the layer of Kapton tape)

Which effectively creates 9 capacitors (three on the signal line six on the shield. The shield ones are parallel connected in pairs)

^{simulate this circuit – Schematic created using CircuitLab}

And it works pretty well.

But I would like to use stock capacitors instead. The second board has 3 1pF capacitors in series on both sides (values on schematic are not correct)

But this version efficiently is a short circuit for this frequency signal.

When I reduce the number of capacitors to one it is much better.

Two capacitors behave much worse.

The length of the boars is 1/4 of the wave length in the copper.

The question. Why is the three (or two) capacitor version is almost blocking the signal?

Answer 1

Your trace impedances are not controlled or defined for each segment of path.

Ballpark figures

e.g. 1pF and 1nH (~2mm) wire can resonate @ 5 GHz, 3pF @ 2.85 GHz This will affect your antenna S21 curve and S11 differently.

• unless you sweep over a wider range e.g. from 1 to 5 GHz you cannot see the magnitude of the problems with resonance. You may not be seeing high Q problems.

Using small tools or your fingertip with a neoprene glove to act as a small shield capacitance in pF over traces, so you can locate the sensitivity of interactive reactances.

• But a proper design would require better design tools to analyze the LC and loss tangent of substrate. (Maxwell, etc)

• GETEK Polyamide or equiv. or Teflon is required for PCB substrate for much lower loss tangent of S21.

• for floating antenna , you have not defined the insulation breakdown voltage or impedance reason for isolation at low frequency.

• 1pF at 2.4GHz is about -j4 ohms, but the path length adds about 0.3nH/mm for this aspect ratio and about 0.8 nH/mm for a skinny wire. (ballpark) so replacing a cap with a jumper won't work well.

• there is no ground plane under cap