9
\$\begingroup\$

Disclaimer: I'm not a trained electrical engineer...I'm a recent Computer Science grad with about ten years experience as a RF test tech. I don't know everything, and to me, experience reigns supreme for practical applications like this. I'm also the only RF person in my organization. I'm reaching out for an experienced RF, EMI, or EMC specialist to chime in on this issue.

I have a wireless shielded room which doubles as an office constructed of 16 gauge welded steel used for isolating emissions within it from the outside world. It hasn't been maintained well and there have been several improper electrical penetrations of the shield in the last couple of decades. As the new RF engineer, I'm trying to restore the shielded effectiveness (SE) for electrical field and plane wave attenuation of at least 60dBm up to 10GHz. We'll have an SE test done after the repairs are complete.

I've made several repairs, including designing and installing waveguide-below-cutoff (WBC) penetrations for fiber and getting rid of ethernet and coax cables. I've properly filtered all of the POTS phone lines, made some repairs to the finger stock & knife edge doors, patched holes, checked the honeycomb filters, and treated several areas for corrosion/rust.

Here's the last remaining issue...

The room has digital control lines for HVAC thermostats (Metasys N2, RS485, shielded twisted pair) going through the "dirty side" of the filter enclosures and penetrating through a small 3/8" hole in 16 gauge steel to the "clean side" of the filter enclosure. I'm not allowed to remove the lines. I have two options that I can see: convert to fiber and pass through a WBC or bond the cable shield to the steel enclosure. Electrically filtering the digital cables, from what I think I understand, would be problematic because it would filter out the high-frequency components in the digital signals. Please correct me if I'm wrong there. Converting to fiber is difficult due to a lack of AC outlets for the fiber converters on both sides of the shielded room.

I think I want to bond the cable shield to the steel enclosure and stuff the 3/8" hole around the cable with steel or copper wool. Do you experienced folks think that this will be "good enough" to achieve the shielded effectiveness of >60dBm for electrical field and plane wave attenuation up to 10GHz?

\$\endgroup\$
4
  • \$\begingroup\$ Aren't you going to have a lot of noise from the RS485 itself? \$\endgroup\$
    – Voltage Spike
    Feb 7, 2018 at 22:45
  • \$\begingroup\$ Probably. Not much I can do about it. It has to stay. At least it's shielded twisted pair, though. \$\endgroup\$
    – GroundRat
    Feb 9, 2018 at 0:17
  • \$\begingroup\$ How about building a screen cage, preferably copper, around the thermostat, bonding it to the wall? You'll need to engineer a removable cover for adjustment, but the feed-through hole will now be completely shielded. \$\endgroup\$ Jul 12, 2018 at 4:02
  • \$\begingroup\$ keep in mind standing wave resonances of internal noise will peak at 1/4 wavelength and cascade CLCLC common mode chokes to span the range of ingress spectrum with about 2 f decades per balanced CM choke selected for each range with cap current to shields \$\endgroup\$ Oct 7, 2019 at 7:24

3 Answers 3

1
\$\begingroup\$

Everything you describe sounds reasonable so far. Your 10GHz SE requirement is pretty high and typical filters for audio will never come close. Having a thermostat within the shield room is not typical however and the honeycomb ventilation (inlet & exhaust) is usually sufficient for most facilities managers (best to not mention sprinkler heads). Will you be doing RF radiated immunity testing? If so you may damage the thermostat itself.

If necessary you could enclose the entire length of cable within steel conduit, run along one of the corners the room (horizontal & vertical). While not perfect, the EM fields tend to be somewhat attenuated in these areas. I would be tempted to then enclose the thermostat within a metallic box that has fine mesh copper screen and bond this to the shield room walls. Finally, it can't hurt to add Low-Q ferrite beads to the cable on both ends. You will have to select the ferrite for effectivity around the frequency(ies) of concern. Lastly, if you are concerned about the metal conduit reflecting RF signals, you can obtain ferrite tiles about 4" square and glue them in place over the conduit on a 45 degree angle in the corners. Same goes for the thermostat box on the wall, just cover the face of the box with ferrite and leave the sides open to allow airflow.

Included with your shielding effectiveness tests, maybe you can set up an antenna and to a background ambient screen with analyzer or receiver. This will also give the emissions profile of the digital signals on the thermostat. If in the end these cannot be eliminated, you will simply need to list these as part of the background noise every time you do any testing.

\$\endgroup\$
0
\$\begingroup\$

what about putting the 485 cable in a steel/iron conduit. It's probably magnetic rather than static radiation. I am not convinced about using wool - screening basically requires density and mass of a ferrous metal. The only time I ever got into this was screening audio gear from a retro-fitted power supply. We bought expensive mu-metal but in the end the most effective was some scrap pig-iron from the workshop floor. Go figure. You are gonna have to experiment a bit, but go for iron/steel and mass.

\$\endgroup\$
0
0
\$\begingroup\$

Ferrite CM filter? When I did EMI work that required a signal run to the DUT, the test house had a large box of ferrites though which we'd run the digital lines. This would suppress common-mode noise but didn't affect differential. Worked fine with Ethernet for example, so I think it would do the job with RS-485 too.

Convert the thermostat to use optical coupling? Assuming you could get a bias supply to it in the room somehow. This would give galvanic isolation from the HVAC mainframe.

Surround the box with Faraday screen? You probably need to do that anyway regardless of how you deal with its cable, assuming it has some kind of microcontroller and display in it.

\$\endgroup\$
2
  • \$\begingroup\$ The cable is a huge ingress antenna transferring outside signals to the inside. A dual stage CM CLC pi filter for VHF and UHF will solve this problem with a shield around the circuit. SHorting the cables while measuring ingress will tell you if you found the ingress path. THe other factor is the condition of all the berillium copper door springs and corner brace tension for 1/10 wave between screws. \$\endgroup\$ Jul 23, 2019 at 20:07
  • \$\begingroup\$ With double layer copper gauze walls, one should be able to get -120 dB in the UHF band at 1GHz \$\endgroup\$ Jul 23, 2019 at 20:22

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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