# Best way to connect coax along a microstrip

I have an open-ended (stripped, if needed) coaxial cable that I would like to use to insert a signal into a microstrip. The circuit (operating at ca. 18 GHz) however, is made in a way where I can't connect the coax to the end of the microstrip like one would usually do but instead the signal will have to be inserted somewhere along the line.

What is the best (=best coupling to the circuit) way I can make the connection in this situation? I know that one way would be to just solder the stripped end to the top of the microstrip but is that really the most efficient way of doing it or just an simple way?

• The phrase non-invasive just doesn't "fit" at those frequencies. Commented Apr 18, 2021 at 12:53
• Flex coax is highly variable at 18 GHz so one might be better off designing the termination and path with semi-ridged coax with open lead at ~ 1nH/mm may be undesirable, thus using right angle connector if possible. Commented Apr 18, 2021 at 12:54
• BTW 1nH or 1mm approx is > j 100 ohms ! Entire design Return Loss specs need to be defined, Commented Apr 18, 2021 at 13:02
• This challenge may be exceeding bandwidth of designer. Commented Apr 18, 2021 at 13:03
• Perhaps a microstrip Directional coupler design is better with controlled impedances on Ceramic. Do you have scattering parameter and signal level specs? Commented Apr 18, 2021 at 13:29

What is the best (=best coupling to the circuit) way I can make the connection in this situation?

The best = 'lowest loss coupling' you can make to the microstrip is to cut a small section out of the microstrip track with a scalpel. Then connect the end of your coax to the half of the strip that you want to inject the signal into. Connect the ground of the coax to the local ground of the microstrip (sometimes easier said than done, but just as, if not more, important than connecting the signal). When you're done with your signal injection experiment, remove the coax, and bridge the gap with a capacitor of few pFs or a piece of foil.

The best = 'most predictable coupling without cutting the microstrip' is to connect a 500 Ω resistor between the midpoint of the line and the end of your injection coax. Terminate the end of the coax with 50 Ω to ground, and connect the coax ground to the local microstrip ground. This will give you a significant signal attenuation, but it will be flat with frequency, and will disturb the microstrip only minimally. Used the other way round, this is an excellent way to monitor signals on a microstrip without disturbing the conditions there too much.

Just soldering the coax onto the microstrip will get your signal on there. However the mismatch will mean that the loss could vary from nothing to lots as the frequency changes. Anything feeding or driven from the microstrip line may be upset by the mismatch.

This likely won't be of much help to you right now, but for any future designs you do, it's worth noting that there are dedicated coaxial connectors with an integrated switch, to allow you to simultaneously create a break in a controlled-impedance PCB trace and attach a connector to that trace.

I've never used one myself, so I can't give any advice or caveats for their use, other than that they seem to have a pretty low cycle life in general.

• Those connectors are generally limited to 6 GHz. I think there are some that can go up to 12 GHz, but couldn't find them lately. Anyway, none of those will help at 18 GHz. Commented Apr 18, 2021 at 17:01
• @AntennaGuy I'm not an RF person, so I'm unfamiliar with the frequency limitations of these switches. Just knew they existed and thought it would be helpful to metnion. Commented Apr 18, 2021 at 21:34
• Rightfully so, it's good you mentioned those. Just thought I could also add some useful information on top of your suggestion. Commented Apr 27, 2021 at 20:36