Bob Rothenberg
Associate Editor
This section of the Newsletter provides a simple and effective way to share experience-based knowledge or measurement results with your colleagues in the EMC community. Articles or application notes should be no longer than about 2000 words, and may include up to 6-8 figures (depending on word count). Submit them to this Associate Editor via e-mail, fax or real mail. See addresses and fax number on page 3.
The following articles on absorber design and reverberation chambers should be of interest to many Newsletter readers. Dr. Mayer and his co-authors, Dr. Berthon and Prof. Perini, present a unique approach for creating ferrite absorbers with improved broadband performance characteristics. Kevin Goldsmith explains the nature of Reverberation Chambers and describes their advantages (and disadvantages) compared to anechoic chambers and shielded rooms. In addition, Ken Javor, author of the paper on differential mode conducted emissions in the Winter 1999 issue, responds to comments from Michel Mardiguian in the Summer 1999 issue.
Comments from readers concerning these articles are welcome, either as a letter (or e-mail) to the editor or directly to the authors.
Michel Mardiguian questioned aspects of the subject investigation in the Summer newsletter. My response summarizes his concerns and provides detailed answers.
1) “Javor assumes that...power line CE investigations and modeling were made considering that power return and power earthing reference were a same conductor, such as a one-LISN configuration would represent actual installations (sic). This is not correct...”
CBEMA threshold of interference (TOI) assessment did use a single 5 µH LISN. A limit for conducted emissions for a two wire above ground power bus configuration was analytically derived from the test data garnered. In contrast to the single LISN TOI measurement technique, CBEMA recommended a two LISN conducted emission (CE) measurement technique. CBEMA explicitly addressed the effect on radio TOI due to a second source of CE available when power bus configuration supplies an above ground neutral. The CBEMA report assumed that the conducted voltage at which TOI occurs would decrease 6 dB, due to twice the amount of CE available with this measurement technique. The validity of this last assumption was the entire basis for my investigation. It appears reasonable as long as the second source of rfi is held to be identical to the first. Equation set 1 models the CBEMA assumption.
TOI single LISN (dBµV) ~ Vmeasured Eqn. 1-1
TOI two LISN (dBµV) ~ ½ · Vmeasured Eqn. 1-2
However, the two LISN technique measures two different types of CE, cm and dm. What is measured at each LISN port is a linear combination of these two noise components. Equation set 2 models my assumption.
LISN 1 (dBµV) ~ [Vcm + Vdm] Eqn. 2-1
LISN 2 (dBµV) ~ [Vcm – Vdm] Eqn. 2-2
TOI author’s model (dBµV) ~ [aVcm + bVdm] Eqn. 2-3
With the model of 2-3, CBEMA assumption is only valid if a » b. If this is not the case, as I suspected, then it makes sense to control CE by modes, with the degree of strictness proportional to the ratio of a to b, that is cm to dm susceptibility.
2) “I think that the real danger...is that of mode conversion in the utility wiring...where a purely DM-originated noise creates, after some distance, a CM component...” Mardiguian mentions a bus with an isolated neutral, and provides a figure (reproduced below) to support the contention.
Key issues here are that
There is no mechanism presented here for significantly unbalancing the dm source with respect to ground.
3) “Another aspect is...statistical...it is not sure that radio sets...in the upper percentiles of Vdm TOI (dm mean + 1s) are the same ones which have the highest Vcm TOI (cm + 1s).”
Mardiguian questions whether radio dm and cm TOI is really 20 dB apart, based on statistical concerns. No statistical assumptions need be made, except in projecting the results of the investigation across all radio sets. Figure 9 from the subject article shows actual radio by radio TOI as well as both mean TOI and one standard deviation off the mean. The reader should note that for those radios whose dm TOI is well below the dm mean, it is also the case that the cm TOI is below the present conducted emission limit. Hence these radios are not now absolutely protected by the levied limit. Such will always be the case. A cost-effective limit protects the majority of victims against the majority of culprits. A bullet-proof limit is not cost-effective.
4) “...critical aspect not...addressed by Javor...the conducted limit...also protects receivers from power line radiated rf noise.”
(Mardiguian’s) Figure 2: DM to CM voltage conversion along AC distribution.
Mardiguian apparently did not read the full published report (referenced in the newsletter article) of which the article itself was but a summary. The report provides exhaustive detail on all the controls placed to ensure accurate and valid results. Among these are the use of an alternate power source test to ensure that measured TOIs are due to conducted, not radiated effects. After measuring conducted TOI, each radio was plugged into clean filtered power. Both dm and cm rf noise was injected into a power wire identical in length and routed from the same LISN, along the same serpentine test fixture as the real power cord. Phase and neutral conductors of the radiating power cord were terminated to the green wire through 50 W to ensure realistic current flow. (Note that radiating power wire was in intimate proximity to the victim radio, not three meters away). The radiated TOI often could not be ascertained at levels available using a signal generator output, but in all cases it was well above the conducted TOI.
(Javor’s) Figure 9: TOIs of Figure 8 with CBEMA TOI normalized to 48 dBmV
Mardiguian calculates the electric field resulting from common (actually, antenna) mode noise at the 46 dBµV limit to be 45 dbµV/m on a two wire power bus, and uses the answer to justify the present conducted emission limit. I agree, that to the extent that such wiring still exists, common mode emissions must be controlled so as to also control radiated emissions. However, the subject of my investigation was a relaxation of the differential mode limit. Mardiguian’s analysis is inapplicable to the subject relaxation. Calculation of the field resultant from the relaxed dm limit shows there is no concern about dm-sourced radiation.
Maximum magnetic field from a long wire pair with equal and opposite currents flowing is Hnet = (I/2p)(s/r2), where I is the equal and opposite current, s is wire separation, r is wire distance from field measurement point.
The ratio between Hnet and the antenna mode H is s/r. If power bus wires are separated by as much as one centimeter, and the field point is at 3 m, (per Mardiguian’s analysis) then net magnetic field due to dm current flow is 50 dB down from that induced by antenna mode currents. Therefore a relaxation of the dm limit of 20 dB would still result in radiated emissions 30 dB below those resulting from the presently allowed antenna mode currents. Note that the presence of a green wire allows common mode currents to flow back in close proximity to phase and neutral, hence reducing common mode sourced radiated emissions to the same level as those resulting from dm emissions.
Ken Javor may be reached at 256-650-0646 or ken.javor@emccompliance.com