Why does EMC use 20 log for voltage and 10 log for power?

Because the decibel is defined on power, and power goes as the square of voltage into a fixed impedance. A power ratio in decibels is 10 log10(P1/P2). Substituting P = V squared over R, with the same R on both sides, the R cancels and the square comes out of the logarithm as a factor of two, giving 20 log10(V1/V2). So the two formulas are the same formula. The practical consequence worth remembering is that a factor of two in voltage is 6.02 dB while a factor of two in power is 3.01 dB, and both describe the same physical doubling seen through different quantities.

How do dB(uV) and dBm relate?

Only through an impedance, and in EMC that impedance is 50 ohms because it is what an EMC receiver and a LISN present. Working it through, a level in dBm equals the level in dB(uV) minus 106.99, which everyone rounds to 107. So 107 dB(uV) is 0 dBm, and 60 dB(uV) is about -47 dBm. The conversion is meaningless without stating the impedance, which is why field strengths in dB(uV/m) have no dBm equivalent at all: there is no impedance in a field, only in a circuit.

What is antenna factor and why is it added rather than multiplied?

Antenna factor converts what the receiver measures into the field that existed at the antenna. It is the ratio of incident field strength to the voltage the antenna delivers, so in linear terms the field is the voltage times the antenna factor. Because both are expressed logarithmically, that multiplication becomes an addition: field in dB(uV/m) equals measured voltage in dB(uV) plus antenna factor in dB per metre. Cable loss and any preamplifier gain enter the same sum, loss added and gain subtracted. Getting a sign wrong here is one of the few EMC errors that shifts an entire scan by a fixed amount, which is also what makes it recognisable.

How do limits at 3 m and 10 m compare?

In free space the field falls as one over distance, so changing distance by a ratio r changes the level by 20 log10(r). From 10 m to 3 m that is 10.46 dB, usually quoted as 10.5. An FCC Class B limit of 40 dB(uV/m) at 3 m therefore corresponds to about 29.5 dB(uV/m) at 10 m, which is how the FCC and CISPR limit lines are put on the same axis. Treat this as an arithmetic convention for comparing limits, not as a prediction of what a site will measure: a real chamber or open area test site has ground reflection and near-field behaviour that the inverse-distance law does not describe.

Decibels for EMC: dB(uV), dBm and antenna factor

Author Hugues Orgitello Last updated 30 August 2026 ~5 min read

Guide, EMC fundamentals

Every number on an EMC report is a decibel, and almost every avoidable error on one is a decibel handled wrongly. The unit compresses a range no linear scale could show on one graph, turns multiplication into addition, and in exchange demands that you always know what it is a ratio of. This page derives the handful of relations that matter, so that none of them has to be memorised.

Why a logarithmic unit at all

An EMC scan spans an enormous dynamic range. A receiver noise floor and a failing harmonic can be six orders of magnitude apart in voltage, which no linear axis renders usefully. Taking logarithms compresses that into a readable scale.

The second reason matters more in practice. Every element in a measurement chain multiplies: an antenna converts field to voltage by a ratio, a cable attenuates by a ratio, an amplifier gains by a ratio. In logarithms those multiplications become additions, so a measurement chain is a sum rather than a product, and a correction is a term you add.

The definition, and where the 20 comes from

The decibel is defined on power:

Power ratio in dB = 10 log10(P1 / P2)

Voltage enters only through P = V squared over R. With the same impedance on both sides, R cancels and the square leaves the logarithm as a factor of two:

Voltage ratio in dB = 20 log10(V1 / V2)

These are one formula, not two. The consequence to hold on to:

Ratio	In voltage	In power
2x	6.02 dB	3.01 dB
10x	20 dB	10 dB
100x	40 dB	20 dB
1000x	60 dB	30 dB

A doubling is 6 dB when you are looking at volts and 3 dB when you are looking at watts, and both describe the same event.

Absolute units: what the ratio is against

A decibel alone is a ratio and says nothing absolute. EMC fixes a reference and appends it to the unit.

dB(uV) is a voltage referred to one microvolt: 20 log10(V / 1 uV). Conducted emission limits are written in it, because a LISN presents the receiver with a voltage.

dB(uV/m) is a field strength referred to one microvolt per metre: 20 log10(E / 1 uV/m). Radiated emission limits are written in it. So 1 uV/m is 0 dB(uV/m), and 100 uV/m is 40 dB(uV/m).

dBm is a power referred to one milliwatt: 10 log10(P / 1 mW). Radio work uses it; EMC borrows it when discussing transmitters.

dB/m, with no reference in brackets, is the antenna factor below. It is a conversion slope, not a level.

Converting dB(uV) to dBm needs an impedance

This conversion is the one most often done from memory and most often misquoted, so it is worth deriving once.

Take a level in dB(uV) across 50 ohms, the impedance every EMC receiver and LISN presents. Converting to power and then to dBm gives a constant offset:

dBm = dB(uV) + 10 log10( (1e-12) / (50 x 1e-3) ) = dB(uV) - 106.99

Rounded, dBm = dB(uV) - 107. Checks: 107 dB(uV) is 0 dBm; 60 dB(uV) is about -47 dBm.

The offset is not universal. It is 107 because the impedance is 50 ohms, and a different impedance gives a different constant. This is also why dB(uV/m) has no dBm equivalent: a field has no impedance in the circuit sense, so there is no power to convert to. A conversion offered without an impedance is not a conversion.

Antenna factor, and the sign that catches people

A receiver measures volts. A limit is written in volts per metre. The bridge is the antenna factor, the ratio of the incident field to the voltage the antenna delivers, published per frequency on the antenna's calibration certificate in dB/m.

Because everything is logarithmic, the chain is a sum:

E [dB(uV/m)] = V [dB(uV)] + AF [dB/m] + cable loss [dB] - preamplifier gain [dB]

Losses add because they were divisions; gains subtract because they were multiplications. Worked example: a receiver reading 32 dB(uV), an antenna factor of 13.2 dB/m, 2.1 dB of cable loss and a 20 dB preamplifier gives 32 + 13.2 + 2.1 - 20 = 27.3 dB(uV/m).

A sign error here shifts an entire scan by a constant. That is what makes it both dangerous and, once suspected, easy to spot: a curve the wrong shape is a real problem, a curve the right shape sitting at a suspiciously uniform offset is usually arithmetic.

Distance scaling between limit sets

In the far field the field falls as one over distance, so moving from distance r1 to r2 changes the level by 20 log10(r1 / r2).

The case that comes up constantly is 10 m against 3 m, because CISPR and the FCC do not always use the same reference distance: 20 log10(10 / 3) = 10.46 dB, usually quoted as 10.5. An FCC Class B limit of 40 dB(uV/m) at 3 m therefore corresponds to about 29.5 dB(uV/m) at 10 m, which is how the two limit sets are put on one axis in CE versus FCC on EMC.

Treat this as a convention for comparing limits, not a prediction of measurement. A real site has ground reflections, and below a few hundred megahertz a 3 m separation is not comfortably far field for a large product, so measured levels do not track the inverse-distance law exactly. The standards permit the scaling for limit comparison; they do not promise the site will agree.

Margin, in the same currency

Since 20 log10(2) is 6.02 dB, a 6 dB margin means emitting half the permitted field strength. That is the usual internal target, and it is not conservatism for its own sake. It absorbs measurement uncertainty, which CISPR 16-4-2 puts as high as 5.2 dB for radiated measurements below 1 GHz, plus unit-to-unit spread and ageing.

A result exactly on the limit has not passed with zero margin. It has produced a value whose uncertainty band lies half above the line.

Key takeaways

20 log for voltage, 10 log for power, because they are the same definition seen through P = V squared over R.
6 dB is a factor of two in voltage; 3 dB is a factor of two in power.
dBm = dB(uV) - 107 in 50 ohms, and the constant is meaningless without the impedance.
dB(uV/m) does not convert to dBm. A field has no circuit impedance.
E = V + AF + cable loss - preamp gain, all in decibels. A sign error offsets the whole scan.
20 log10(10/3) = 10.46 dB puts 3 m and 10 m limits on one axis, for comparison rather than prediction.

Frequently asked questions

Why does EMC use 20 log for voltage and 10 log for power?: Because the decibel is defined on power, and power goes as the square of voltage into a fixed impedance. A power ratio in decibels is 10 log10(P1/P2). Substituting P = V squared over R, with the same R on both sides, the R cancels and the square comes out of the logarithm as a factor of two, giving 20 log10(V1/V2). So the two formulas are the same formula. The practical consequence worth remembering is that a factor of two in voltage is 6.02 dB while a factor of two in power is 3.01 dB, and both describe the same physical doubling seen through different quantities.
How do dB(uV) and dBm relate?: Only through an impedance, and in EMC that impedance is 50 ohms because it is what an EMC receiver and a LISN present. Working it through, a level in dBm equals the level in dB(uV) minus 106.99, which everyone rounds to 107. So 107 dB(uV) is 0 dBm, and 60 dB(uV) is about -47 dBm. The conversion is meaningless without stating the impedance, which is why field strengths in dB(uV/m) have no dBm equivalent at all: there is no impedance in a field, only in a circuit.
What is antenna factor and why is it added rather than multiplied?: Antenna factor converts what the receiver measures into the field that existed at the antenna. It is the ratio of incident field strength to the voltage the antenna delivers, so in linear terms the field is the voltage times the antenna factor. Because both are expressed logarithmically, that multiplication becomes an addition: field in dB(uV/m) equals measured voltage in dB(uV) plus antenna factor in dB per metre. Cable loss and any preamplifier gain enter the same sum, loss added and gain subtracted. Getting a sign wrong here is one of the few EMC errors that shifts an entire scan by a fixed amount, which is also what makes it recognisable.
How do limits at 3 m and 10 m compare?: In free space the field falls as one over distance, so changing distance by a ratio r changes the level by 20 log10(r). From 10 m to 3 m that is 10.46 dB, usually quoted as 10.5. An FCC Class B limit of 40 dB(uV/m) at 3 m therefore corresponds to about 29.5 dB(uV/m) at 10 m, which is how the FCC and CISPR limit lines are put on the same axis. Treat this as an arithmetic convention for comparing limits, not as a prediction of what a site will measure: a real chamber or open area test site has ground reflection and near-field behaviour that the inverse-distance law does not describe.
What does a 6 dB margin actually mean?: A factor of two in voltage, since 20 log10(2) is 6.02 dB. A design sitting 6 dB under a limit is emitting half the permitted field strength. That is the usual target because it has to absorb measurement uncertainty, which CISPR 16-4-2 puts as high as 5.2 dB for radiated measurements below 1 GHz, plus unit-to-unit spread and ageing. A product measured exactly at the limit has not passed with no margin, it has produced a result whose uncertainty band sits half above the line.