I've been working on a design that requires me to pass an RF signal through a conductive panel. I'm using SMA jacks for that purpose as they're readily available and also available in an isolated configuration like these. The reason I'm using an isolated jack is to avoid a ground loop through the chassis and up the electronics assembly. Here's an image of what I'm doing:

enter image description here

This image shows the return current path or ground (blue line), as if I were using a standard SMA panel jack where the shield would be electrically connected to the case like this. So assuming I instead use an isolated jack, I'd remove the return path through the case. The problem now is that I've gotta get the assembly to be IP67 rated. I cannot seem to find any isolated RF jacks that also offer any kind of ingress protection.

The GPS receiver provides a DC signal to the antenna external to the assembly for powering a LNA, so breaking the ground at the connector is out as I cannot have the return LNA current, or RF return for that matter, returning through the case and up the electronics as that would surely wreak chaos on my signals.

Does anyone have any suggestions? I'm not an EMI/EMC expert but have been reading a bunch on the topic, though I still have a ways to go in developing an intuitive approach for nuanced situations like this one. Thanks.

  • \$\begingroup\$ I have used unisolated connectors in condutive casings for GPS signal without any problems. Have you tried using an unisolated connector? Is there a problem when you do? You might be trying to fix a problem that isn't there. \$\endgroup\$ – Peter Karlsen Aug 30 at 7:15
  • \$\begingroup\$ @Peter Karlson, yes in the former revision of this system, I originally used nonisolated bulkheads. There was significant issues which I solved by using the isolated jacks. I am willing to admit the problem may have manifested differently than how I assumed, but the isolation fixed, or at least obscured, the problem. This must be why EMC consultants are so well paid. Thanks! \$\endgroup\$ – AJbotic Aug 30 at 12:07
  • \$\begingroup\$ I've seen bakelite washers used to isolate a connector from the metal. \$\endgroup\$ – analogsystemsrf Aug 30 at 14:41

Does anyone have any suggestions?

  • Use an RF isolating transformer between your circuit board and the panel. Maybe even use one at your antenna.
  • Use two single-ended (obtainable) bulkhead connectors
  • Find a bulkhead connector that fits the bill (keep looking)
  • \$\begingroup\$ The RF transformer idea would work well for the RF portion, but the DC power to the LNA wouldn't get through. I like the idea of two single ended connectors. Are you suggesting passing both signal and return through on the center conductors? Would you anticipate issues due to impedance mismatches and reflections? Yeah I figured the keep looking option was likely. Probably time I get in touch with an FAE at Pasternack or Fairview. Thanks! \$\endgroup\$ – AJbotic Aug 30 at 12:14

The RF return will not significantly flow through the case, for two reasons.

  • First, even at moderate frequencies, RF flows primarily along the skin of a conductor. Because of this, the outer shield of a coaxial cable or connector essentially becomes two separate conductors - one on the inside surface, and one on the outside surface.

  • Secondly, at higher frequencies the return path for a signal will "stick" closer and closer to the outgoing path. This is very relevant for EMI in PCB design, for instance.

These effects kick in even at tens of MHz. At ~1.6GHz for GPS, they will absolutely dominate. The return current for a signal propagating down the centre conductor will return exclusively down the inner surface. This means that effectively, your RF return path is already isolated.

So I think your noise issue is with the DC bias used to supply the LNA at the active antenna. Armed with that knowledge, there might be other things you can do to resolve the issue - i.e. improve filtering/decoupling of the DC supply at the active antenna end.

Another thought - do you absolutely need a panel mount connector on the outside of the housing, or could you run insulated coax through an IP67 rated gland and then have an SMA connector on a short trailing lead?


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.