RF via's on circuit cards, general rules, reference info

[Higgler]

Maybe someone could point out a good reference book on RF via's and give some general guidance on RF plated thru holes (via's).

I plan on using Rogers Duroid 5880 or similar microwave dielectric. Thickness, 30, 60, 100 mils.

Questions;
1) what is the length/diameter ratio typically used as a maximum?
2) Can the hole be plated only on the inside, or does some copper need to be on both top and bottom surfaces of the circuit card surrounding the drilled hole?
3) When you wish to make a coax using a center contact via plated thru hole surrouded by outer via plated thru holes (as the outer shield of the coax), has anyone experience in doing that up to 18 GHz?
4) What circuit card company would you recommend to lay out and build an RF circuit card up to 18 GHz.

Thanks,
khiggins

http://www.toyon.com

 

[logbook]

The key thing about a via is that it should be small (in length) compared to the wavelength of operation. Lambda/4 would be a real problem, so we might say <lambda/20 so that we can consider the via as a lumped inductor.

When we are talking about the highest frequencies, the skin effect means that whilst we have conductor on the inside of a hole in the via, this is not a whole lot different to the cylinder of conductor on the outside of a solid wire. I would therefore suppose that the inductance of the via could be calculated using the inductance formula for an equal diameter solid wire. The inductance formula for a wire is a logarithmic function of length/diameter and therefore does not change very rapidly. Bigger diameter is better of course (less inductance). You should have pads top and bottom to support the via, even if the bottom pad is not used (due to the via's connection to an inner ground plane, for example).

Question 3 sounds non-standard. It seems as if you are trying to get an array of vias on the circumference of a circle (centred at a plated through hole) to act as the ground shield of a coaxial structure. That is very strange. Normally you would be working in microstrip or coplanar waveguide, both on the surface of the substrate. Now you are talking about going perpendicular to the substrate. This is worth discussing in its own right.

 

[Higgler]

Thanks logbook,

We are using CST Microwave Studio to analyze this, it's very accurate, well...it's getting more accurate as we do this, the engineers in Germany are writing code to improve their product as we continue.

On question 3, I know that certain connectors having four posts around the center conductor that are used to essentially make a coax. outer conductor. Hence four is probably the right number.

The pad that's added to the top and bottom edge should add about 10 mils to the diameter of the hole, hence 5 mils wider radius (per E-Fab in the bay area). Any other thoughts on this? i.e. is 2,4,6,8 mil added diameter done by any vendor, or is Efab using the typical value of pad size. This will affect our results at 16 GHz.

Ratio of thickness of board/hole diameter, Efab says 3:1 ratio for this, though for normal circuits I was told that 6:1 is common, 10:1 is difficult and 15:1 is the upper limit. I'd appreciate inputs on these ratio's from anyone, expecially if you have a favorite vendor who regularly does 6:1 or more.

Plating is typically 1 mil thick, so it closes off the hole diamter by 2 mils in size.

Material friendly to via's. Duroid 5880 has a huge Z axis expansion that can pop via's and I think should only be used for thin boards. Rogers 3003 and 6002 are alot better. Our design so far is 5880, I need to have coworkers change that. Any other thoughts on materials vs. via's. Our thickness is about 125 mils of circuit board to place a via inside.

Any thoughts on percentage of material that you can make via's on. If I have 5 holes (4 ground on the perimeter and one center conductor) and I remove alot of material, at some point there is a structural weakness. Anyone know of percentage holes or percentage material needed when you group alot of via's close together, i.e. 50% rule????

Any reference book suggestions,

thanks,

khiggins at

http://www.toyon.com