Test & Measurement World, July/August 2012

Issue link:

Contents of this Issue


Page 24 of 35

16-in. chassis seam. I need a way to better understand near-field measurements from chassis seams or around connectors and how they relate to the far-field amplitude. I've had some people tell me to fix every leak I see, but this can be expen- sive and a waste of money. I've read ar- ticles that say near-field measurements may or may not be related to the far field, and that sometimes problems seen in the far field aren't seen in the near field and vice versa. Answer: Correlating near-field with far-field measurements is a common area of concern. Before correlating the measurements, engineers need to be sure they understand the physics of the situation. Consider a "point source" (electri- cally small) radiating E-field dipole an- tenna (Hertzian dipole). In the near field, the radiating electromagnetic waves will be spherical, and they will gradually become planar (plane waves) in the far field. As you move away from this radiating point source, the E-field (measured in volts/meter) drops off with distance. In the near field, the equations for the electromagnetic (electric and magnetic) fields are complex and include factors of 1/r2 and 1/r3 . In other words, the field strength tends to drop off quickly. In the far field, the measured E-field drops off by a factor of 1/r, where r is the distance away from the point source radiating antenna. Because this conver- sion from spherical to planar waves is gradual, it's difficult to definitively say where the exact transition point is, but most EMC engineers assume the point is at: distance = wavelength/(2*π), or about 1/6 of a wavelength. If you were to plot the wave imped- ance (Zω = E/H, E-field and H-field plots) versus frequency for a dipole an- tenna, you would find that in the near field, the E-field is high (over 5 kΩ) and the H-field is minimal (about 20 Ω). If the emission source has a high current and low voltage (E/H < 377), the near field is predominantly magnetic (H- field). If the source has a low current and high voltage (E/H > 377), the near field is predominantly electric (E-field). As you move out to the far field, the E- field and H-field wave impedances con- verge to approximately 377 Ω in free space. Henry Ott has a plot of this, along with a good explanation, in his latest book, Electromagnetic Compatibility Engi- neering (Ref. 1). So, what does this mean when mea- suring near-field emissions on a prod- uct? You'll need to recognize that the near-field to far-field correlation is very complex and not a one-to-one relation- ship. You'll also need to consider wave- length as you measure circuit traces, ca- bles, seams, and apertures. Jeremy is right with his guess that the 6-ft. cable will tend to be more of a radiation risk than the 2-in. slot. The rule of thumb of 1/20 of a wavelength is good to keep in mind. Think of the seams, gaps, traces, holes, and cables as antennas. Small, or electrically short (that is, less than 1/20 of a wavelength), antennas are ineffi- cient radiators. As the electrical length approaches 1/4 or, especially, 1/2 wave- length, the antenna becomes an efficient radiator and can cause trouble. When I'm measuring enclosure seams, for example, I identify the domi- nant harmonics and use a marking pen to record the length of the particular seam, as well as the frequencies of con- cern. If the seam is electrically short, then it's probably not the dominant emission source. If I'm measuring com- mon-mode currents on a cable, I iden- tify the dominant harmonics and, be- cause most cables are electrically long, I can be relatively sure emissions on that cable will be a problem. The nice thing about a current probe is that you can maximize their readings by sliding it back and forth along the cable, then fix- ing it in place, before you begin trou- bleshooting. The one thing to be wary of, though, is that a 10-dB drop in near-field emis- sions does not usually equate to a 10-dB drop in far-field emissions. Remember, you're dealing with distance factors of 1/r3 and 1/r2 in the near field (tends to drop off quickly) versus 1/r factors as you move into the far field. You can be assured, however, that a significant drop in the near field will also help to some degree in the far field. If you find several potential seam leaks in your enclosure, do what Ott suggests and "kill it dead" by throwing everything you've got at the problem. Then, once you've achieved compliance, start removing fixes that aren't required. Simulating a far field Question: To simulate a far field, I've tried setting up a small digital TV an- tenna in my lab that is about 1 m away, but my ambient environment is too polluted. The 30-MHz to 60-MHz am- bient is higher than my own emissions, so it doesn't really do any good. Answer: Setting up a nearby antenna where you're troubleshooting is a great idea! I use that technique all the time, because it shows a truer picture of rela- tive increases or decreases in field strength. Of course, if you could locate the antenna at 3 m or 10 m away, you'd be able to predict much more accu- rately whether the product would meet the required radiated emission limits. The problem is that all the other ambi- ent signals from radio, TV, and mobile communications tend to interfere with observing the harmonic signals from a product. Hence, EMC labs use shielded measurement chambers for formal qualification testing. I've set up temporary 3-m measure- ment sites in the basement of a building, which tends to help. I've also tried to orient a directional receive antenna at 90º to the strongest ambient, which tends to reduce the pickup. The only other solution is to try shielding the room you're working in with aluminum foil taped to the walls. I've also seen shielded tents that are used for tempo- rary shield rooms. Just a hint: Generally speaking, domi- nant emissions below 200–300 MHz are likely due to common-mode cur- rents on cables; above that frequency, emissions are more likely due to slot, seam (differential-mode emissions), or sometimes cable emissions. In this spe- cific case of 30 to 60 MHz harmonics, you could probably do most of your troubleshooting with a current probe, which tends to not be as sensitive to far- off ambient signals. Current probes Question: I need to make a current probe, but one thing that isn't clear to Test & Measurement World | JULY/AUGUST 2012 | –25–

Articles in this issue

Links on this page

Archives of this issue

view archives of EETimes - Test & Measurement World, July/August 2012