When performing emissions tests within a shielded enclosure (a.k.a. screen room), a common cause of test outages is ambient RF signals. The usual (and typically acceptable) practice is to turn off the unit under test (UUT), re-measure the ambient spectrum, then compare the results between the UUT powered and unpowered tests to identify which emissions are contributed by the UUT and which are contributed by the ambient environment. Not only does this cause additional test time, but sometimes the results are not so easy to interpret if both the UUT and the ambient environment have emissions at the same frequencies, particularly if they are near the same amplitude. This is often the case when the UUT exercising equipment is generating the troubling ambient signal. The culprit is not always the local radio or TV station.
But wait, the UUT is in a screen room; how can an ambient signal get inside the screen room and contaminate the test results? In a properly designed and maintained screen room, it is not difficult to achieve significant isolation between the inside and outside of the room, easily on the order of 100 dB. Rarely is the screen room the guilty party. Often your nicely shielded test cables are the culprit – the ones with the shields that properly terminate about the circumference of their associated connectors. Those carefully designed shielded cables that are fished through a pass-through or “stuffing tube” in the wall of the screen room. You remember that pass-through with a length-to-diameter ratio greater than five that ensures that only insignificant amounts of RF can pass. This is the “waveguide beyond cutoff” effect. Unfortunately, you have defeated this powerful shielding technique by converting the waveguide into a coaxial cable that connects to an antenna both inside and outside the screen room.
For this technique to work, the waveguide/pass-through, often made of bronze, must be filled with a good dielectric, usually air. Your test cable violates this condition when it passes through the pass-through into the screen room without special consideration. The shield on your test cable becomes the center conductor of a coaxial cable whose outer conductor is the screen room pass-through. Viola, you have converted a waveguide into a coaxial cable and 2 antennas! RF energy from numerous ambient sources is easily coupled onto the that shield (antenna 1) outside the screen room, passes through this short, makeshift coaxial cable, and radiates from the shield (antenna 1) on the inside of the screen room – right into your measurement antenna.
Fortunately, it is easy to fix this often-overlooked problem. Just bond your test cable shield to the screen room pass-through. At the bonding point, stuff the pass-through (that’s why it’s called a “stuffing tube”) with conductive, usually copper, wool to ensure the tube is RF tight. If this test setup is used often, a more repeatable and easier-to-install arrangement would be to mount a bulkhead connector in the screen room penetration plate. Your test cables would then need appropriate connectors inside and outside the screen room – each with properly terminated shields.
I’ve demonstrated this effect many times to “non believers.” A simple way to do this is to set up the emissions measurement system to sweep over the FM broadcast band (88 to 108 MHz) or greater. For the first test, ensure that no conductive cables pass through any of the screen room’s pass-throughs. Take a sweep and note the “amount of quiet.” Next, fish a length of hookup wire (or about any conductor) through a pass-through tube leaving a few feet on the inside and outside of the screen room as your antennas. (The effect will be more dramatic if you support the free ends of the wires above the floor.) Take another sweep and note the amount of Led Zepplin and elevator music! This usually elicits an “aha” from observers.
Stu Benner
E3 Design Engineer
EMConsulting, LLC
Well explained – question, what happens if the conductor passed through has it’s own signal eg. a Cat 5 data cable or a firewire cable? How much can that be eliminated by using bulkhead mount connectors and a shielded cable within the room?
Thanks, Kerryn.
An unshielded Cat 5 cable can have surprising low emissions if the signals, cables, and terminations are balanced thereby minimizing differential mode to common mode conversion. Contamination of the signal pairs with common mode noise from elsewhere in the circuit or system is still a potential problem, but is manageable with proper layout and design. While this is very often adequate to meet commercial (e.g., FCC Part 15) EMC standards, military emissions and susceptibility requirements are much more stringent and usually require a shield as you describe. When properly implemented, such a shielded cable can provide on the order of 30 to 60 dB of attenuation. The cable shield’s surface transfer impedance is one of most useful parameters in predicting the effectiveness of the shield. Just remember that the shield terminations are almost always the weakest link in the design.