Measurements and Limits of Conducted EMI: The Conducted Emission Limits

What are we designing our filter to really achieve? In Table 1, we have provided the CISPR 22 test limits and the FCC limits. We have plotted out the CISPR limits in Fgr. 6. Note that in the table, dBµV has also been expressed in terms of mV, in deference to engineers who may have trouble thinking "logarithmically." By definition, a decibel is dB = 20 x log (voltage ratio). So "dBµV" implies a logarithmic comparison to a reference voltage of 1 µV. Note that for power, a decibel is 10 x log (ratio).

Note: The FCC does not state average limits, only quasi-peak. However, the FCC does accept certification to CISPR 22. But in that case, "mix-and-match" of the standards is not allowed - it's either one or the other.

Note: The published CISPR 22 Class B limits in the range 150 kHz to 500 kHz are themselves actually 'rounded-up' versions of the 'easier to remember' forms provided above. That is why the "exact" equations above look surprisingly stranger than the "almost exact" ones. In fact, what has been done in CISPR 22 is a line was drawn with a slope of -20 dB/dec, passing through the 1 MHz point (extrapolated, see Fgr. 6) -- at 40dBµV for the average limits and 50 dBµV for the quasi-peak limits. Thereafter, the calculated y-coordinates at 500 kHz (point of truncation of this portion) and at 150 kHz were rounded off to the nearest decibel.

Example: What are the CISPR Class B conducted emission limits at a frequency of 300 kHz? A frequency of 300 kHz is 0.3 MHz. Therefore, the average limit is:

19.07 × log (0.3) + 40.28 = 50.25 dBµV

The quasi-peak limit (which is defined to be 10 dB higher in this region) is automatically 60.25 dBµV.

Looking at the FCC and CISPR 22 quasi-peak limits in Table 1, we can justifiably ask - do the numbers imply that the FCC standards are actually more stringent than those of CISPR 22 (above 450 kHz)? Not really! The first difference is that FCC measurements are done at the lower U.S. line voltages, whereas the CISPR measurements are done at roughly twice that. So we may be in the position of comparing apples to oranges. Further, though FCC has no defined average detection limits, the language allows for a relaxation of the limits (by 13 dB) if the quasi-peak reading exceeds the average by more than 6 dB.

Therefore, practically speaking, equipment compliant to CISPR will always be found compliant to FCC limits.

Fgr. 7: Average, Quasi-peak and Peak Readings of a Pulsed Wave --- Peak Quasi-Peak Average PULSED INTEREFERENCE Internal Attack and Release times.