Lessons from building the Faraday Cage at the Interop Labs

With Interop over, it’s worth jotting down a few lessons about the Interop Labs Faraday cage. First, the results were better than I expected, given the difficulty of building a complete RF shield. Based on measurements we took using both test tools and laptop-based tools, it appears that our cage provided about 40-45 dB of shield, which is a good amount given our abbreviated time and limited materials budget. For comparison, the VeriWave test chambers are guaranteed to provide at least 80 dB of shielding (though the number appears to be much higher in practice).

  • 802.11 shielding: 40 dB of shielding is good enough to provide a substantial reduction in the number of visible APs. Even before the show started, it was possible to see 300 to 400 APs immediately outside the door of the cage. With the door closed, we could reduce that number to 20-30, which includes the four APs that were powered on inside the cage.
  • Mobile phone shielding: The cage did cause some signal loss for cell phone signals, but the service came through loud and clear. Occasionally, some phones would exhibit the loss of one bar, but I don’t believe there is a standard indication for what a bar means between vendors, or even between different phones from the same vendor.
  • Related to the first point, the team from DiVitas found the cage extremely useful. Their demonstration of an fixed/mobile convergence application depended on having “reasonable” 802.11 service to set up a call on 802.11, and then degraded service to hand the call over to the voice network. Our cage provided that without any difficulty, and even allowed people to hear that voice quality on 802.11 can be better than voice quality on cellular networks.

After going through the process, we did learn a few lessons:

  • Buy, don’t build. We had to spend lots of time fiddling with the cage, and there were complicated entry/exit procedures for making sure that the door was adequately screened. Various flaps had to be secured in particular ways to get maximum shielding.
  • Related to the previous point: Don’t buy from us. We had fun building it, but this is not a vocation for us.
  • Materials matter. The cage was made of aluminum screen because it is inexpensive material, but aluminium has properties that are suboptimal for a project like this. Aluminium naturally forms a protective oxide layer. In aerospace, that’s cool. In Faraday cage manufacture, it’s not. The aluminium oxide is an insulator, so we saw resistances across the outer mesh screen approach an ohm if the current had to cross sheet boundaries. Aluminium is a difficult material to work with, since it melts at 660 °C, but its oxide melts at 2054 °C. In practice, we were unable to join the metal sheets together directly because we didn’t want to try brazing (and we lacked equipment which could generate the requisite heat anyway); we ultimately settled for “sewing” the sheets together with copper wire as shown in the photos included with original post.
  • Complete screening is hard. Our first attempt was to keep the cage completely isolated from everything, with only a power drop. However, putting Ethernet cables in to the cage did not seem to change the screening effect at all. Ethernet is tightly twisted to resist carrying interference, but it can still act like an antenna. There are two possible conclusions: we did a perfect job putting the Ethernet in, or the cage leaked enough that there was no incremental penalty from the Ethernet cables. We’ll go with the latter choice.
  • As a sign that the cage leaked even without the Ethernet, we did not notice any difference between (1) no Ethernet cables, (2) one Ethernet cable, (3) one Ethernet cable with a snap-on ferrite core, or (4) two Ethernet cables, one of which had a snap-on core. For demonstration purposes, we put two cables in to the cage, but a “real” cage should take penetrations much more seriously.
  • Grounding didn’t matter. The electrical contractor set up a ground wire for us from the show electrical system, but connecting it did not make a difference in the effectiveness of the cage. Our suspicion is that the ground wire was good enough to act as an electrical safety, but that it had high impedance to the ground itself. In the end, we decided to leave it connected because it “looked cool.”

Leave a Reply