RF PCB design in altium

[leaps]

hello everyone I am a newbie PCB design using altium, I am designing PCB for amplifier circuit at 80MHz for the circuit I am using surface mount components. I need some feedback with the fallowing questions as I am new to this

1- How does one determine the optimum height of board material

2- I have substrate height on 20 mils, Er=3.55 for Zo=50 ohms I get width of 44.8 mil and I have been told line widths more than 20mils is not recommended so what should I do my surface mount components are 0603m/0805in/2520mm.

3- I am aware RF signal path line has to have a width for a 50ohms line but the lines connecting for DC bias components can have different width??

4- what is the recommended spacing between smd inductors connected in series?

5- how do i determine the optimum width of signal trace on the bottom layer

 

[volker@muehlhaus]

3- I am aware RF signal path line has to have a width for a 50ohms line

Line width matters if the lines have a relevant length compared to the wavelength. At 80MHz, where your routing length is most likely < 1/20 wavelength, there is no need for controlled line impedance.
That said, you should keep your interconnects between SMD components short, and to avoid parasitic series L and parasitic shunt C. Make it small, that's the rule.

 

[leaps]

volker@muehlhaus thanks I am using bottom layer for dc supply distribution. the separation between Vdc is 20mil how do i determine its optimized any suggestions
a-

Also is it ok to route the RF signal under Q.
b-

 

[volker@muehlhaus]

Slots in the ground will cause return current to flow a much longer path, that's bad. You can solve this by placing a suitable capacitor (1nF or so) across the slot, so that RF current can flow on the ground layer underneath the RF trace.

 

[CataM]

routing length is most likely < 1/20 wavelength, there is no need for controlled line impedance.

That is too big a constraint. Controlled line impedance starts at more than 1/4 wavelenght.. Why do you use that number ?

 

[volker@muehlhaus]

That is too big a constraint.

I know ... used this number to indicate that it's really no issue here. The usual estimate for line effects (include or ignore) is lambda/10, calculated using the effective wavelength in that substrate.

Controlled line impedance starts at more than 1/4 wavelenght..

Wrong. At lambda/4 it doesn't start, at lambda/4 you are in massive trouble already because an open end converts into an RF short.

 

[leaps]

Hi I have a fallow up question I need to use conductor's as RFC's and how do i chose between shielded inductor vs unshielded one and layout considerations for both