Mental Math for RF Engineers: Calculating dB, Power, and Bandwidth on the Fly

RF calculations often span picowatts to watts, so logarithms are unavoidable. A small set of remembered ratios makes most first-pass checks possible without a calculator.

The method below handles three common jobs:

1. Convert dB or dBm to a linear power.
2. Estimate thermal noise from bandwidth.
3. Keep power and voltage ratios straight.

Five dB building blocks

For power, the exact ratio is $10^{dB/10}$. These five approximations are usually enough:

Step	Exact power ratio	Use mentally
$+10\,\mathrm{dB}$	$10.000$	$\times 10$
$+7\,\mathrm{dB}$	$5.012$	$\times 5$
$+3\,\mathrm{dB}$	$1.995$	$\times 2$
$+2\,\mathrm{dB}$	$1.585$	$\times 1.6$
$+1\,\mathrm{dB}$	$1.259$	$\times 1.25$

Take the $10\,\mathrm{dB}$ decades first, then form the remainder from 7, 3, 2, and 1. Negative dB values use the reciprocal: $-3\,\mathrm{dB}$ is approximately $\div 2$.

Worked examples

• $+13\,\mathrm{dBm}$: $+10+3$ gives $10\times2=20\,\mathrm{mW}$.
• $+17\,\mathrm{dBm}$: $+10+7$ gives $10\times5=50\,\mathrm{mW}$; exact is $50.1\,\mathrm{mW}$.
• $+27\,\mathrm{dBm}$: $+10+10+7$ gives $10\times10\times5=500\,\mathrm{mW}$.
• $+43\,\mathrm{dBm}$: $+40+3$ gives $10^4\times2\,\mathrm{mW}=20\,\mathrm{W}$.
• $-6\,\mathrm{dB}$: two $-3\,\mathrm{dB}$ steps give $\div2\div2=\div4$.

These estimates are for checking scale and catching large mistakes. Use exact math for calibration limits and tolerance analysis.

Bandwidth and the 0-30-60-90 ladder

Available thermal noise in a matched bandwidth is

$$N = kTB$$

In dBm, with $B$ in hertz:

$$N_{\mathrm{dBm}} = kT_{\mathrm{dBm/Hz}} + 10\log_{10}(B)$$

The bandwidth term is easy to remember:

Bandwidth	dB-Hz	$kTB$ using $-173.86\,\mathrm{dBm/Hz}$
$1\,\mathrm{Hz}$	$0$	$-173.86\,\mathrm{dBm}$
$1\,\mathrm{kHz}$	$30$	$-143.86\,\mathrm{dBm}$
$1\,\mathrm{MHz}$	$60$	$-113.86\,\mathrm{dBm}$
$1\,\mathrm{GHz}$	$90$	$-83.86\,\mathrm{dBm}$

For a $20\,\mathrm{MHz}$ channel, start at $60\,\mathrm{dB}$-Hz for $1\,\mathrm{MHz}$ and add $13\,\mathrm{dB}$ for the factor of 20:

$$10\log_{10}(20\times10^6) \approx 73\,\mathrm{dB\text{-}Hz}$$

The thermal noise is therefore about $-101\,\mathrm{dBm}$ using the familiar $-174\,\mathrm{dBm/Hz}$ rule.

The constants deserve one clarification. dB Power Converter uses $-173.86\,\mathrm{dBm/Hz}$ for this teaching convention; $-174\,\mathrm{dBm/Hz}$ is the normal mental estimate. Direct evaluation of CODATA $k$ at exactly $290\,\mathrm{K}$ gives approximately $-173.975\,\mathrm{dBm/Hz}$, which IndieRF RF Cascade Analyzer uses. For temperatures other than $290\,\mathrm{K}$, calculate physical $kT$ rather than reusing the room-temperature shortcut.

From total power to power density

Integrated power tells you how much power is in the channel. Power spectral density tells you how that power is distributed across frequency. If the power is uniformly distributed over bandwidth $B$:

$$S_{\mathrm{dBm/Hz}}=P_{\mathrm{dBm}}-10\log_{10}(B_{\mathrm{Hz}})$$

For density per megahertz, either add $60\,\mathrm{dB}$ to the per-hertz result or express the bandwidth directly in megahertz:

$$S_{\mathrm{dBm/MHz}} =S_{\mathrm{dBm/Hz}}+60 =P_{\mathrm{dBm}}-10\log_{10}(B_{\mathrm{MHz}})$$

The subtraction is the reverse of integrating noise over bandwidth. A wider signal has lower average density when total power stays fixed.

Example: $+17\,\mathrm{dBm}$ over $20\,\mathrm{MHz}$

The previous sections give $+17\,\mathrm{dBm}\approx50\,\mathrm{mW}$ and $20\,\mathrm{MHz}\approx73.01\,\mathrm{dB}$-Hz:

$$S\approx17-73.01=-56.01\,\mathrm{dBm/Hz}$$ $$S\approx-56.01+60=+3.99\,\mathrm{dBm/MHz}$$

That is about $2.5\,\mathrm{mW}$ in each megahertz. Across 20 MHz, those twenty slices sum to approximately $50\,\mathrm{mW}$.

More useful reference points

Total power	Bandwidth	Average density per Hz	Average density per MHz
$0\,\mathrm{dBm}$	$20\,\mathrm{MHz}$	$-73.01\,\mathrm{dBm/Hz}$	$-13.01\,\mathrm{dBm/MHz}$
$+17\,\mathrm{dBm}$	$20\,\mathrm{MHz}$	$-56.01\,\mathrm{dBm/Hz}$	$+3.99\,\mathrm{dBm/MHz}$
$+27\,\mathrm{dBm}$	$100\,\mathrm{MHz}$	$-53\,\mathrm{dBm/Hz}$	$+7\,\mathrm{dBm/MHz}$
$+30\,\mathrm{dBm}$	$100\,\mathrm{MHz}$	$-50\,\mathrm{dBm/Hz}$	$+10\,\mathrm{dBm/MHz}$

At $1\,\mathrm{MHz}$, total dBm and dBm/MHz have the same numerical value. At $100\,\mathrm{MHz}$, the average dBm/MHz value is $20\,\mathrm{dB}$ below total power.

The thermal-noise example works in reverse. A density of $-173.86\,\mathrm{dBm/Hz}$ integrated across $20\,\mathrm{MHz}$ becomes approximately $-100.85\,\mathrm{dBm}$. Expressed per megahertz, the same density is $-113.86\,\mathrm{dBm/MHz}$.

These conversions describe average density. They are appropriate for flat noise or as a first-order estimate for a spread waveform. Do not divide a CW tone by an arbitrary bandwidth, and do not use average density as a substitute for peak spectral density when a shaped modulation or regulatory mask matters.

10log versus 20log

Power ratios use $10\log_{10}$:

$$\mathrm{dB}_{P}=10\log_{10}\left(\frac{P_2}{P_1}\right)$$

Voltage ratios use $20\log_{10}$ when the impedances are equal:

$$\mathrm{dB}_{V}=20\log_{10}\left(\frac{V_2}{V_1}\right)$$

The distinction follows from $P=V_{\mathrm{rms}}^2/R$. At a matched $50\,\Omega$ reference plane:

$$V_{\mathrm{rms}}=\sqrt{50P},\qquad V_{\mathrm{pp}}=2\sqrt{2}\,V_{\mathrm{rms}}$$

Therefore $0\,\mathrm{dBm}=1\,\mathrm{mW}=0.224\,\mathrm{V_{rms}}=0.632\,\mathrm{V_{pp}}$ into $50\,\Omega$.

The two rules engineers most often mix up are:

• Twice the power is $+3\,\mathrm{dB}$.
• Twice the voltage is $+6\,\mathrm{dB}$ and four times the power.

The $20\log$ relation only applies directly when the two voltage measurements use the same impedance. If the impedances differ, convert each voltage to power first.

Bench reference

dBm	Approximate power	Useful decomposition
$-30$	$1\,\mu\mathrm{W}$	three $-10\,\mathrm{dB}$ decades
$-20$	$10\,\mu\mathrm{W}$	two $-10\,\mathrm{dB}$ decades
$-10$	$100\,\mu\mathrm{W}$	one $-10\,\mathrm{dB}$ decade
$0$	$1\,\mathrm{mW}$	reference
$+10$	$10\,\mathrm{mW}$	$\times10$
$+17$	$50\,\mathrm{mW}$	$\times10\times5$
$+20$	$100\,\mathrm{mW}$	$\times100$
$+27$	$500\,\mathrm{mW}$	$\times100\times5$
$+30$	$1\,\mathrm{W}$	$\times1000$
$+43$	$20\,\mathrm{W}$	$\times10^4\times2$

Where mental math stops

These shortcuts are useful for checking a measurement or reviewing a budget. They do not replace a path calculation when gain, noise figure, filtering, compression, and routing all interact. That is where a deterministic cascade model earns its keep.