Noise Figure for Active HF Receive Antennas: When It Matters

Noise figure describes noise added by a receiving network under stated source and reference conditions. It does not describe the external noise collected by the antenna, the field-to-output transfer of an electrically small sensor or the largest blocker the active stage can tolerate. An active HF antenna must satisfy both ends of the range: enough sensitivity for the quietest credible field and enough linearity for the strongest credible spectrum.

Start with the Reference Plane

At the standard reference temperature T₀ = 290 K, available thermal-noise density at a matched port is approximately −173.98 dBm/Hz. Integrated over equivalent noise bandwidth B:

Those values are not the on-air HF floor. They are the reference noise available from a matched 290 K source. State the reference plane, impedance, equivalent noise bandwidth, detector, averaging and gain state before comparing measurements.

A two-port noise factor is the degradation of signal-to-noise ratio between defined input and output ports. ITU-R SM.1838-1 requires test frequency, receiver state and temperature to be controlled and describes gain, Y-factor and sensitivity methods. A headline such as “0.9 dB NF” is incomplete without frequency range, source impedance or noise parameters, gain state, temperature and uncertainty.

External Noise and Receiver Noise Are Different Terms

ITU-R P.372-17 defines external noise factor f_a from the available noise power of an equivalent loss-free receiving antenna, relative to kT₀B. That antenna is a reference construct, not necessarily the physical port of a short active whip or loop.

For an ideal lossless antenna directly feeding a matched receiver, a useful simplified system expression is:

Lower receiver NF matters little when f_a is very large. But the full P.372 expression also includes antenna-circuit loss, transmission-line loss, their physical temperatures and receiver noise. A lossy, mismatched or very small sensor may deliver much less external noise to the amplifier than the equivalent loss-free antenna would. That moves the receiver contribution back into view.

What ITU-R P.372 Does—and Does Not—Predict

P.372 separates atmospheric noise from lightning, man-made noise and galactic noise. Its outdoor man-made median model uses:

The current recommendation gives different c and d values for city, residential, rural, quiet-rural and galactic curves, plus substantial time and location decile deviations. The man-made curves refer to a short vertical lossless grounded monopole. They are statistical components, not universal antenna-terminal readings.

A lower-band noise estimate must include relevant atmospheric lightning, man-made and galactic components rather than selecting only the largest median curve. Independent noise powers combine in linear or statistical form; P.372 Part 7 supplies the applicable combination method.

The galactic approximation below about 100 MHz also neglects ionospheric shielding. P.372 states that galactic noise is not observed below foF2 and is below the simple curve up to roughly three times foF2. At 14 or 28 MHz the model can be a useful reference in suitable ionospheric conditions, not a guaranteed floor.

A conditional worked example

Using only the P.372 median component curves and the ideal lossless matched system above, the calculated change in total noise when receiver NF improves from 10 dB to 1 dB is:

These results apply to one statistical component, the P.372 reference antenna, a median value, no passive loss, matched ports and no overload. Actual site noise may be higher or lower, and the active sensor may deliver it to the amplifier with a different transfer function.

When a 10 dB System Noise Figure Is Not Enough

With F_a = 18.7 dB, an ideal 10 dB system NF adds about 0.50 dB to the external-only noise. That is a useful design example: receiver-added noise is roughly 9.2 dB below that particular external component.

It is not a universal specification. Ten decibels may be unnecessarily low on a noisy low-band installation, or inadequate when:

• the antenna is inefficient, mismatched or followed by passive loss before the first gain stage;
• the antenna pattern or polarization rejects much of the external-noise field;
• the site and ionosphere are quieter than the selected median component;
• the active device sees a source impedance far from its noise optimum;
• the receiving bandwidth or required SNR changes; or
• a downstream receiver has high NF and the active antenna supplies too little gain.

Specify the allowed system-noise penalty, such as 0.5 dB, and solve from the measured delivered external noise. This makes the trade-off explicit and avoids treating one worked example as a topology-wide rule.

Effective Length and Antenna Factor Connect Field to Output

For a linearly polarized receiving antenna in a locally uniform field, the open-circuit signal voltage is the vector projection of the incident electric field onto the antenna’s complex effective length:

The effective length includes direction, polarization and frequency. The active antenna factor additionally includes the sensor-to-amplifier coupling, active gain, output termination and any cable or bias network inside the calibration boundary. For a small magnetic loop, the corresponding field transfer is based on magnetic flux, loop effective area and the complete coupling network. Either way, calibrated field-to-output transfer is what connects an environmental field to receiver input power.

Antenna factor and effective length must be measured or derived for the installed orientation and declared reference plane. A low amplifier NF cannot compensate for an unknown or lossy field-transfer path, and a large active gain cannot establish sensitivity without output noise and uncertainty.

An Active Whip, Loop and 50 Ω LNA Do Not Share One NF Boundary

Keysight's noise-parameter guidance shows that amplifier noise figure varies with source impedance. A 50 Ω Y-factor result is therefore not enough to characterise an amplifier driven by a highly capacitive whip or inductive loop. Use the device noise parameters, or an input voltage/current-noise model valid over frequency, with the measured complex sensor impedance.

CISPR 16-1-6 covers calibrated antenna factors and uncertainty for relevant monopole and loop measurements from 9 kHz upward; installation and active electronics require their additional conditions. IEEE 145-2025 supplies the antenna terminology. A gain number cannot replace a field-to-output transfer function.

Gain Only Helps Until Following Noise Is Subordinate

For matched cascaded stages, the Friis noise-factor relation is:

Enough first-stage gain makes the receiver's added noise small. Extra gain after that does not improve the signal-to-noise ratio already present at the antenna output; it raises wanted signal, external noise and blockers together. It can drive the active-antenna output stage, receiver mixer or ADC closer to compression or clipping.

Choose gain from a budget containing minimum wanted field, delivered external noise, cable loss, receiver NF, strongest in-band and out-of-band fields, ADC full scale and required margin. A switchable gain or attenuation state can outperform one fixed “maximum sensitivity” setting across changing bands.

Sensitivity Is a Defined Quality Threshold

The familiar matched-input estimate:

describes a receiver driven from a stated matched source. It is not an on-air active-antenna sensitivity by itself. ITU-R SM.1840 emphasizes that receiver sensitivity depends on test frequency, modulation, IF bandwidth, quality criterion such as SINAD and temperature. Keysight likewise notes that signal-generator sensitivity is degraded by real background noise when an antenna is connected.

For an active antenna, publish the minimum incident E or H field that yields the stated output SNR or decoder criterion, with bandwidth, modulation, frequency, orientation, termination, gain state and uncertainty. Also publish output noise spectral density so users can determine whether their receiver remains the limiting stage.

Dynamic Range and Overload Need More Than One Number

A wideband HF active antenna may simultaneously receive medium-wave broadcasters, shortwave transmitters, local amateur signals and impulsive interference. The amplifier can create false signals or a raised floor long before the wanted channel itself is strong.

ITU-R SM.1837-1 defines a reproducible IP3 procedure for 9 kHz–30 MHz monitoring receivers. ITU-R SM.575-3 shows why strong-signal protection also depends on antenna gain, cable loss, bandwidth, interfering fields, sensitivity and external noise. IP3 is a theoretical extrapolation, not a maximum input rating, and P1dB does not predict every multitone spur.

Spurious-free dynamic range derived from IP3 also depends on measurement bandwidth and the two-tone model. A narrower FFT bin lowers displayed noise but does not increase the amplifier's blocker headroom. State the bandwidth and reference plane whenever quoting dynamic range.

Filtering and Common Mode Are Part of the Active Antenna

A high-pass, notch or band-pass filter ahead of the first nonlinear stage can improve usable dynamic range when a known broadcast service dominates. Its insertion loss and physical temperature also change system noise. Place and select filtering from a measured blocker survey rather than applying one filter to every site.

Common-mode current on coax, power or control wiring can bypass the intended small sensor. It may raise received noise, change antenna factor and deliver strong fields directly to active electronics. Measure current and common-mode-to-output transfer. A quiet-looking trace after adding loss or reducing gain is not proof of improved input SNR.

A Complete Active-Antenna Test Programme

1. Declare boundaries. Identify the sensing element, active unit, bias tee, feedline, receiver and every calibration plane.
2. Measure complex source and port impedances. Include the sensor, enclosure, mast or shield, operating bias and final geometry.
3. Calibrate field transfer. Measure E- or H-field-to-output response, pattern, polarization, orientation, termination and uncertainty.
4. Measure noise correctly. Use a valid Y-factor, cold-source or source-substitution method where applicable; otherwise derive equivalent input voltage/current noise with the real source impedance.
5. Measure gain and output noise. Verify that following cable and receiver noise are subordinate without exhausting output or ADC headroom.
6. Map external noise on site. Follow ITU-R SM.1753 principles: calibrated antenna, receiver, bandwidth, detector, time statistics and uncertainty.
7. Test two-tone linearity. Sweep tone pairs across in-band and out-of-band cases, levels, spacings and gain states; record actual products before compression.
8. Test compression, blocking and composite spectra. Include realistic broadcasters, nearby transmitters, impulses and supply conditions.
9. Test common mode. Inject or measure current on coax, power and control leads; verify balance and choke performance over frequency.
10. Repeat over environment. Temperature, supply voltage, termination and component tolerances can shift noise, gain and headroom.

Bottom Line

On many lower-HF installations, delivered external noise is so far above receiver-added noise that another decibel of 50 Ω amplifier NF will not improve on-air SNR. On a quiet site, with a small or lossy sensor, passive loss ahead of gain, unfavorable source impedance or inadequate cascade gain, NF can matter materially.

P.372 component curves are statistical reference-antenna models. Relevant atmospheric, man-made and galactic terms must be combined correctly, and a short active sensor does not necessarily present or deliver a 50 Ω lossless source. Use calibrated antenna factor or effective length to determine what field and noise the sensor actually delivers.

Specify the complete field-to-output system. Keep receiver-added noise below the quietest delivered external noise by a chosen margin, use only enough gain to overcome following noise, and preserve headroom with measured IP3, compression, blocking, filtering and common-mode performance. That tells an operator what the antenna can hear—and what strong signals will make it invent.

Questions, antenna-factor records or height trials to share? Contact RF.Guru.