Why is conducted noise power measured in dBuV instead of watts?

Conducted-emission limits in CISPR 22, CISPR 32, and FCC Part 15 are written as a voltage developed across the standardized 50 ohm impedance of a Line Impedance Stabilization Network, expressed in decibels relative to one microvolt (dBuV). The receiver or spectrum analyzer reads a voltage, not a power, so the regulatory mask is a voltage mask. The equivalent power into the 50 ohm port is small: 60 dBuV corresponds to 1 mV, which is only 20 nanowatts into 50 ohm, yet that level can still fail Class B limits. Engineers convert with P(dBm) = V(dBuV) minus 107, so the CISPR 22 Class B quasi-peak limit of 56 dBuV in the 0.5 to 5 MHz range equals about minus 51 dBm, and the 60 dBuV limit at 30 MHz equals about minus 47 dBm.

Power Electronics

Conducted Noise Power

Q: How is conducted noise power split into common-mode and differential-mode components?

Differential-mode (DM) noise flows out one power conductor and returns on the other, driven mainly by the ripple current in the input capacitor of a switching converter; it dominates the lower end of the conducted band, roughly 150 kHz to 5 MHz. Common-mode (CM) noise flows in the same direction on both conductors and returns through the safety ground or parasitic capacitance to chassis, driven by the high dV/dt at the switching node coupling through transformer interwinding capacitance and heatsink capacitance; it dominates from about 5 MHz to 30 MHz. A LISN measures the sum on each line, but a CM/DM separator or a current probe with a common-mode choke lets the engineer split them, because each requires a different filter element: an X-capacitor attenuates DM while a CM choke plus Y-capacitors attenuate CM.

Q: What detector and bandwidth does CISPR specify for conducted noise power testing?

For the conducted band of 150 kHz to 30 MHz, CISPR 16-1-1 specifies a measurement receiver with a 9 kHz resolution bandwidth (6 dB) using two weighting detectors: quasi-peak and average. The quasi-peak detector applies defined charge and discharge time constants (1 ms charge, 160 ms discharge in this band) so that infrequent narrow spikes are weighted less than continuous noise, modeling the annoyance to analog receivers. A scan must pass both the quasi-peak limit and the lower average limit. A peak detector is often used for fast pre-scans because peak is always greater than or equal to quasi-peak, so passing peak guarantees passing quasi-peak.

/kuhn-DUK-tid noyz POW-er/

The unwanted radio-frequency energy that a power supply, almost always a switching converter, injects back onto its AC mains or DC supply conductors as a byproduct of fast switching transitions. Measured as a voltage in dBµV across the standardized 50 Ω ports of a line impedance stabilization network over the 150 kHz to 30 MHz band, it separates into common-mode and differential-mode contributions, each suppressed by a different filter element. Quantified this way it is the core metric of a product's conducted emissions, and it must fall below the quasi-peak and average masks of CISPR 22/32 or FCC Part 15 before equipment can ship. It is distinct from radiated emissions, which leave the enclosure as electromagnetic fields rather than traveling out along the power leads.

Category: Power Electronics

Test band: 150 kHz to 30 MHz

Units: dBµV (across 50 Ω)

Where Switching Converters Generate Conducted Noise

Every hard-switched power converter produces sharp voltage and current edges. The drain or collector node of a buck, flyback, or boost stage slews tens to hundreds of volts in a few nanoseconds, and the input loop carries a trapezoidal current waveform with steep di/dt. These edges are rich in harmonics that extend well past 30 MHz, but the conducted-emission regulations focus on the 150 kHz to 30 MHz window because that is where the energy couples efficiently back onto the power leads rather than radiating from the box. The fundamental switching frequency of a modern converter sits between 65 kHz and 500 kHz, so the first several dozen harmonics land squarely inside the regulated band.

Two coupling mechanisms dominate. Differential-mode (DM) noise is the ripple voltage the input capacitor cannot fully bypass; it appears between the two power conductors and scales with loop inductance and the magnitude of the switched current. Common-mode (CM) noise is driven by the high dV/dt at the switching node coupling through parasitic capacitance, primarily the transformer interwinding capacitance and the heatsink-to-chassis capacitance, into the safety ground. DM tends to dominate from 150 kHz to roughly 5 MHz and CM from 5 MHz to 30 MHz, although the crossover shifts with layout. Because the two modes need opposite filter topologies, separating them is the first diagnostic step in any failed scan.

The measurement is deliberately repeatable. A LISN inserts a known 50 Ω / 50 µH impedance between the utility and the equipment under test, blocks ambient mains noise, and routes the converter noise to a CISPR-compliant receiver. The receiver applies quasi-peak and average detectors with a 9 kHz resolution bandwidth, producing the two traces that are compared against the Class A (industrial) or the stricter Class B (residential) limit lines.

Governing Relationships

Voltage-to-power conversion (50 Ω reference):
P(dBm) = V(dBµV) − 107 → 60 dBµV = 1 mV ≈ −47 dBm (20 nW)

Total vs. CM and DM (per conductor):
V_line = V_CM + V_DM, V_neutral = V_CM − V_DM

CM filter corner (CM choke L_CM with two Y-caps C_Y):
f_c ≈ 1 / (2π × √(L_CM × 2C_Y))

Where V is in dBµV at the LISN 50 Ω port. Example: an X-cap forms the DM pole; an L_CM = 4.7 mH choke with C_Y = 2.2 nF gives f_c ≈ 35 kHz, providing >40 dB CM attenuation at 1 MHz.

CISPR and FCC Conducted-Emission Limits

Frequency	CISPR 22 Class B QP	CISPR 22 Class B Avg	CISPR 22 Class A QP	FCC Part 15 Class B QP
150 kHz	66 dBµV	56 dBµV	79 dBµV	66 dBµV
500 kHz	56 dBµV	46 dBµV	73 dBµV	56 dBµV
1.5 MHz	56 dBµV	46 dBµV	73 dBµV	56 dBµV
5 MHz	56 dBµV	46 dBµV	73 dBµV	56 dBµV
30 MHz	60 dBµV	50 dBµV	73 dBµV	60 dBµV

Common Questions

Frequently Asked Questions

How is conducted noise power split into common-mode and differential-mode components?

Differential-mode noise flows out one power conductor and returns on the other, driven mainly by input-capacitor ripple current; it dominates roughly 150 kHz to 5 MHz. Common-mode noise flows the same direction on both conductors and returns through safety ground via transformer interwinding and heatsink capacitance, dominating about 5 MHz to 30 MHz. A LISN reads the sum per line, but a CM/DM separator splits them, because an X-capacitor attenuates DM while a CM choke plus Y-capacitors attenuate CM.

Why is conducted noise power measured in dBµV instead of watts?

CISPR 22/32 and FCC Part 15 conducted limits are written as a voltage developed across the LISN's standardized 50 Ω impedance, in decibels relative to one microvolt. The receiver reads a voltage, so the mask is a voltage mask. The equivalent power is tiny: 60 dBµV is 1 mV, only about 20 nW into 50 Ω, yet still enough to fail Class B. Convert with P(dBm) = V(dBµV) − 107, so the 56 dBµV Class B quasi-peak limit (0.5 to 5 MHz) equals roughly −51 dBm, while the 60 dBµV limit at 30 MHz equals about −47 dBm.

What detector and bandwidth does CISPR specify for conducted noise testing?

For 150 kHz to 30 MHz, CISPR 16-1-1 specifies a 9 kHz resolution bandwidth (6 dB) receiver with two weighting detectors: quasi-peak and average. The quasi-peak detector uses defined charge and discharge time constants (1 ms charge, 160 ms discharge in this band) so infrequent spikes weigh less than continuous noise. A scan must pass both the quasi-peak and the lower average limit. A peak detector is used for fast pre-scans because peak is always ≥ quasi-peak.

Related Terms

← Conducted Emissions Conducted Susceptibility →