3
\$\begingroup\$

I have a communications system that connects a self-powered sensor circuit to a wall-powered processing/analyzing circuit by coax cable

The coax reaches out about 60m, and it is matched/terminated at the far end. When I measure the noise/interference spectrum on the line at the near end, I see a significant amount of RF ingress from AM stations. When the cable's shield is grounded at the near end, the ingress seems to be worse than when there is no cable grounding.

If I uncouple the shield from earth-ground, the ingress is still notable, but it drops by 15 to 20dB.

There is no grounding possible at the far end, but at the near end I have the option of earth-grounding to the mains, or not. Of course, at the near end it is also matched/terminated.

  1. Is the shield acting as a 60m antenna, and is the signal possibly coupling into the core along the length of the coax?

My understanding is that it is best practice to ground at the receiving end, which in my case is the near end, but the above observation seems to be going against it.

In my measurement set-up the analyzer/scope is not grounded, but in the communication system where the coax is to be deployed, the circuit ground is (for now) earth grounded.

The chassis and power supply of the system on the near end must be grounded for safety reasons.

  1. Am I correct to conclude that the comms circuit ground should be decoupled from the earth/chassis ground?

Upon request, a crude drawing (sorry, won't use this drawing tool again)

A: the communications system

B: the measurement set up observing AM ingress with coax shield grounded

enter image description here

\$\endgroup\$
11
  • \$\begingroup\$ Are you using a dipole antenna? If so, is there a balun at the feed point of the antenna? \$\endgroup\$
    – vu2nan
    Jun 18 '20 at 18:40
  • \$\begingroup\$ @vu2nan Good point, no antenna. The far end goes into a self-powered sensor circuit. I have updated the question. \$\endgroup\$
    – P2000
    Jun 18 '20 at 19:03
  • \$\begingroup\$ And to be clear, is the coax braid is always connected to the chassis of the device, both ends? So you're wondering about connecting the chassis of the sensor, and of the receiver, to the earth below it? \$\endgroup\$
    – tomnexus
    Jun 19 '20 at 4:27
  • \$\begingroup\$ If thats all OK, unfortunately your AM pickup will depend on the geometry and length of the cable. 60 m might work best ungrounded, but in another installation grounding might be better. Have you tried a suitable ferrite bead, probably several turns on a powdered iron core, on the cable near the processing circuit? \$\endgroup\$
    – tomnexus
    Jun 19 '20 at 4:30
  • \$\begingroup\$ Yes I did, @tomnexus, it was tuned to the band of interest, and it reduced it almost entirely. But this would kill the signal too (in the same band) since the TEM cannot get past it. As for shield/braid grounding, the sensor cannot be earth-grounded, the receiver can be, and so should the Vdd be grounded to earth? It seems not, per the experiment. The receiver chassis is earth-grounded but not necessarily connected to Vdd. So what about my hypothesis that the coax is acting as an antenna if its shield (braid) is earth-grounded at the receiver? \$\endgroup\$
    – P2000
    Jun 19 '20 at 5:13
2
\$\begingroup\$

The problem is of common mode currents in the coaxial feed line, induced by strong local AM radio stations.

It could be addressed by using a balanced system with two coaxial cables serving as a twin line feeder. The braids would need to be interconnected at both the ends and earthed only at the receiver end.

The receiver and sensor would require balanced input and output matched to the feeder impedance that would be double the characteristic impedance of the coaxial cable used.

enter image description here

Attempts by others to resolve a similar problem, through the use of a common mode choke, isolation/matching transformers and selective grounding, have yielded favourable results.

The details are accessible at https://groups.google.com/forum/#!topic/sci.electronics.design/hKwMXrD9ksE and http://theradioboard.com/rb/viewtopic.php?f=10&t=9357&p=92911#p92880

It would be worth your while to explore the possibility of application of similar techniques to find a solution.

Should that be successful, the major changes required to convert your system to a balanced one, using a dual-coaxial cable feeder, could be avoided.

\$\endgroup\$
12
  • \$\begingroup\$ Interesting, so we get the better propagation from coax, but ingress gets rejected. Are the coaxs a twisted pair? If straight, are they snug? Where has this been done? \$\endgroup\$
    – P2000
    Jun 20 '20 at 22:32
  • \$\begingroup\$ Please refer 'Series-Connected Coaxial Cables' at the following link: on5au.be/content/a10/trans/spcoax.html \$\endgroup\$
    – vu2nan
    Jun 21 '20 at 10:08
  • \$\begingroup\$ Great info, @vu2nan, I think it's the parallel configuration that's of interest. \$\endgroup\$
    – P2000
    Jun 21 '20 at 15:23
  • \$\begingroup\$ Only Fig.1 is apt for your application. \$\endgroup\$
    – vu2nan
    Jun 21 '20 at 15:29
  • \$\begingroup\$ Oh yes, I see, because it will be a differential signal. Again, great info. Can you say something about the grounding in a single-coax arrangement? \$\endgroup\$
    – P2000
    Jun 21 '20 at 15:33

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.