SMA RF end-launch transitions

[LukeW]

If you look at some MMIC reference boards, like this one for example:

http://media.digikey.com/Photos/Hittite%20Microwave%20Photos/121923-HMC669LP3.jpg

There is a fairly wide flap, a wide capacitive pad, that is just after the SMA connector, where the SMA transitions onto the 50 ohm transmission line, which is substantially wider than the CPWG line.
What is its function? It looks like it's adding a bit of capacitance... to balance the inductance in the SMA transition?

But not all ref boards from RF vendors include such a thing. Some Hittite ones do, for example, and some don't. What factors go into determining whether it's appropriate or not? (Other than an empirical measurement of the return loss and other characteristics?)

[T3sl4co1l]

Curious what it looks like on the bottom.  Should think the ground plane is continuous, in which case, yes, it would be quite a bit of capacitance (effective).

The transition itself is probably inductive, so it serves to level things out.  But it might be they haven't done a good analysis of it either.  Which could be empirical, or badly simulated, or painstakingly simulated..

Tim

[ejeffrey]

Remember the return current has to flow outward to the ground pins on the edge launch connector, which adds some inductance.  It looks like that pad would both reduce that inductance (by reducing the loop area) as well as add a bit of capacitance to compensate.  To what level this was carefully designed/simulated vs. cut-and-try and they found something that works I don't know.

[mathias]

Quite strange, I thought that the transition always added capacity due to the grounded metallic fingers protruding onto the board.

Anyway, you can find an "interesting" article about optimized transitions here: http://mpd.southwestmicrowave.com/showImage.php?image=287] [url]http://mpd.southwestmicrowave.com/showImage.php?image=287[/url].

[G0HZU]

If you look closely, this is a multilayer board. The reference ground (for most of the PCB) is the first metal layer down from the top layer where top layer = layer 1.

So this presents a problem for a simple/cheap SMA end launcher. It's not easy to get a good connection to this inner layer with this type of end launcher.

My guess is that they have used defective grounding in the reference layer to try and offset this problem. So the microstrip is thicker over the defected ground area. So it 'launches' against the very bottom layer? You could try peeling the PCB apart to see if the inner reference ground (layer 2) has a cutaway pattern near the SMA launchers?

But all this is just a guess.

[G0HZU]

I tried a quick simulation on Sonnet and I think they might be doing something like the 3D view below.

There are three metal layers in this simple simulation and you can see where I have defected the inner reference ground layer (shown in red) under the fatter microstrip section on the top metal (green) layer.

The grey layer is the bottom metal layer of the whole PCB stack.

Note: I haven't bothered to model/simulate it as grounded coplanar as it would take a lot longer to simulate and my PC runs very hot when running Sonnet on a hot summer night like tonight. But the basic idea is still the same anyway.