Transmission line response on FR4 substrate

[aqeelqureshi543]

Dear all,
I have designed loaded line phase shifter and RF switch on FR4 substrate (frequency less than 2 GHz) using hsmp 386B pin diodes. But the measured results are coming at half the simulated/ designed frequency i.e. insertion loss simulations at 1.5 GHz yield optimum measured result at ~760 MHz.
Can anyone tell me the reason for this deviation????
How can I make the simulated and measured results close to each other???

 

[OldRFguy]

FR4 is a very poorly controlled substrate material that will show a lot of variation between manufacturers. I would have thought your best course of action would be to make some test boards up with various widths of line and made from the material that you intend to use for final production and then measure these on a network analyzer to see how closely they come out to your modelled results in terms of loss and impedance. From these measurements, try to define what the actual properties of the material are that you're using before creating you're next round of models and simulations.

I hope this helps...

 

[rfgeek]

Surprisingly FR4 can have an Er ranging from 3.8 to 4.9 and sometimes more. Some designers will use 4.4 or 4.6 but that may be nowhere close. That is why PCB manufacturers will ask if you are doing a controlled impedance design. They will figure out what you need for 50 ohm lines based on there manufacturing process and materials. Still FR4 is fine for lower frequency microstrip designs but if you are doing a sensitive design needing a good Q then I would not use FR4 much above 1 GHz.