A low noise figure (NF) is valuable only when amplifier noise is a meaningful part of the system noise. Across much of HF, atmospheric, man-made, and galactic noise already dominate. Once the front end is quiet enough, further NF improvement produces little usable SNR—while overload, instability, and common-mode pickup can still ruin reception.
The design target: make the amplifier quiet enough for the antenna and band, then maximize strong-signal headroom, stability, and system-level common-mode rejection. The best device is the one that meets all four requirements at the same time.
Start with the Antenna’s Source Impedance
NF is always measured for stated conditions, including source impedance and frequency. That matters because an active antenna’s sensing element is rarely a simple 50 Ω source:
- E-field probes are commonly high-impedance, capacitive sources. Input capacitance, leakage, bias current, and current-noise density can matter as much as voltage-noise density. Protection components must also be included because their capacitance and leakage load the probe.
- Small H-field loops present resistance plus inductive reactance. Depending on whether the circuit senses voltage or current, amplifier voltage noise, current noise, transformer ratio, loop loss, and input termination all affect sensitivity.
- A 50 or 75 Ω output does not imply a 50 or 75 Ω sensor input. The amplifier at the antenna head performs that conversion, so its input must be evaluated against the actual loop or probe impedance across the full operating band.
“H-field” and “E-field” describe an antenna’s dominant coupling mechanism, especially in the near field. For a distant plane wave, the electric and magnetic fields are linked. A shielded loop can reduce local capacitive E-field pickup, but it does not receive a separate kind of radio wave.
Practical consequence: compare total input-referred noise in the finished circuit—not one voltage-noise or NF number copied from a datasheet.
When Does Lower NF Still Improve SNR?
ITU-R P.372 documents the external radio-noise sources seen by receiving systems and how they vary with frequency, location, season, and time. In many HF installations, this external noise is well above the amplifier’s added noise. The exceptions include very quiet sites, the upper end of HF and low VHF, lossy or very small sensors, and front ends that load the antenna incorrectly.
A useful design quantity is the external-noise margin at a common reference plane. If antenna-delivered noise is M dB above the electronics’ added noise, the approximate SNR penalty from the electronics is:
SNR penalty = 10 log10(1 + 10−M/10)
With a 6 dB margin, the penalty is about 1.0 dB. With a 10 dB margin, it is about 0.4 dB. Beyond that point, improving NF delivers rapidly diminishing returns.
This comparison is valid only when both noise contributions are referred to the same impedance, bandwidth, gain state, and point in the signal chain.
Match the Symptom to the Limitation
| What you observe | Likely mechanism | What to examine |
|---|---|---|
| Band noise barely rises above the electronics’ own noise | Insufficient sensitivity, excessive loss, or a poor noise match | Source impedance, voltage/current noise, gain, input loading, and loss ahead of the first stage |
| Phantom signals, broadband “hash,” or broadcast splatter | Compression, cross-modulation, or intermodulation | Input P1dB, IIP2/IIP3, blocker levels, gain distribution, and filtering |
| Noise changes when the coax is touched or rerouted | Feedline common-mode pickup; the coax is participating as an antenna | Balance, system CMRR, isolation, choking, bonding, enclosure, and cable routing |
| Spurs or peaks move when the cable, load, or supply changes | Oscillation or marginal stability | Stability margin, decoupling, feedback, source/load impedance, shielding, and PCB layout |
MMIC or Wideband Op-Amp? Neither Wins by Category
“MMIC versus op-amp” is not a complete design decision. A monolithic microwave integrated circuit (MMIC) gain block and a wideband op-amp can both be low-noise and highly linear; the surrounding topology, bias, impedance environment, gain, filtering, and layout determine the finished front end.
| Design factor | RF gain block or MMIC | Wideband op-amp or discrete front end |
|---|---|---|
| Characterization | Often specified with S-parameters, NF, P1dB, and IP3 in a defined RF impedance environment—commonly 50 Ω, with some 75 Ω CATV devices. | Often specified with voltage/current noise, slew rate, distortion, and stability conditions; RF behavior must be evaluated in the intended closed-loop circuit. |
| Sensor interface | Convenient after a transformer or matching network, but a direct connection can load a high-impedance probe or miss the optimum noise match. | Flexible for high-impedance buffering, transimpedance, active termination, and differential sensing. |
| Strong-signal performance | Many RF gain blocks offer excellent, clearly specified headroom for their cost and current. | Can also be excellent, but depends strongly on closed-loop gain, supply voltage, output loading, common-mode range, and feedback topology. |
| Stability | Predictable when the specified source/load, biasing, and reference layout are respected; reactive antennas and long cables can still cause trouble. | Noise gain, feedback parasitics, capacitive sources/loads, and layout require careful control; use the manufacturer’s stability guidance. |
| Best use | A strong candidate when a defined RF interface, repeatable gain, and characterized linearity are priorities. | A strong candidate when the sensor requires high input impedance, transimpedance conversion, custom gain, or a fully differential interface. |
Practical takeaway: choose the architecture that produces the best measured system result with the real antenna, blockers, supply, enclosure, and feedline—not the part family with the most attractive single datasheet number.
Use Linearity Specifications Correctly
- Input P1dB indicates where gain has compressed by 1 dB. It is a real large-signal operating point.
- IIP3 is an extrapolated third-order intercept used to predict two-tone IMD in the small-signal region. It is not a damage rating or an allowable input level.
- IIP2 matters when second-order products from strong signals can fall in-band. Good differential balance can improve it substantially.
- Compare input-referred figures at the same frequency, gain, supply, impedance, and test method. OIP3 and output P1dB can look larger simply because gain has been added.
Filtering also has to be placed deliberately. A filter at the receiver protects the receiver, but it cannot undo overload that has already occurred in the active antenna head. When local MW, FM, or transmit signals are strong enough to stress the head amplifier, rejection must occur before—or within—that first active stage.
Why Differential and Push-Pull Stages Help
For a balanced loop or probe, a well-executed differential front end can offer important advantages:
- Even-order suppression: ideal symmetry cancels even-order distortion, improving second-order performance.
- Common-mode rejection: interference coupled equally into both input paths is rejected—within the circuit’s frequency-dependent CMRR.
- Differential signal swing: some architectures can deliver more differential output swing for a given supply.
- Repeatable phase and amplitude: useful in arrays when both channels are tightly matched and calibrated.
These benefits are not automatic. Device mismatch, unequal protection capacitance, transformer imbalance, asymmetric PCB geometry, and uneven antenna coupling all convert common-mode energy into differential error. A differential stage also does not guarantee better IIP3; third-order performance still depends on device linearity, bias, degeneration or feedback, and signal swing.
CMRR Is a System Property
Amplifier CMRR is only one link in the chain. The complete receive system includes the sensing element, input protection, transformer or matching network, PCB, enclosure, power feed, coax, common-mode choke, station bonding, and cable routing. Noise coupled onto the feedline after the balanced input cannot be removed by the input stage’s datasheet CMRR.
Ask for CMRR or balance versus frequency. A single low-frequency figure says little about performance across HF or VHF.
Selection Checklist
- Define the sensor impedance versus frequency. Include the antenna element, transformer, protection network, and PCB parasitics.
- Set an external-noise margin. Verify that the finished front end is quiet enough on the quietest intended band and site.
- Build a blocker profile. Include MW and shortwave broadcast, FM, nearby amateur transmitters, and multi-transmitter operation.
- Compare the right linearity numbers. Use IIP2/IIP3 and input P1dB under comparable gain, supply, frequency, and impedance conditions.
- Plan filtering and gain distribution. Protect the earliest stage that can overload; extra gain is useful only when it preserves system headroom.
- Test stability at the corners. Vary the antenna, cable length, load, supply, temperature, and enclosure state.
- Measure system-level common-mode behavior. Preserve symmetry through protection, layout, cabling, and grounding.
- Engineer protection separately from linearity. ESD/surge survival, nearby-TX survival, and clean reception during TX are different requirements.
Bottom Line
E-field probes: prioritize input impedance, input capacitance, current noise, leakage, linearity, and protection. Their high-impedance sensor interface often favors FET-input or other purpose-built buffering, but the complete noise calculation decides.
H-field loops: prioritize the loop-to-amplifier noise match, symmetry, input-referred linearity, and magnetic-field overload tolerance. A small or lossy loop can still make amplifier noise important.
For both: once the required external-noise margin is achieved, spend the remaining design budget on headroom, filtering, stability, common-mode control, and protection.
Mini-FAQ
Why isn’t noise figure always the main specification?
Because the amplifier’s noise adds to noise already delivered by the antenna. When external noise is much larger, reducing amplifier noise changes total SNR only slightly. NF matters again when that margin becomes small.
Do MMICs always beat wideband op-amps?
No. MMICs are often convenient and well characterized in standard RF environments; wideband op-amps and discrete stages can be better for high-impedance, transimpedance, or differential sensor interfaces. Compare the finished circuits under the same conditions.
What matters most for an H-field loop?
The loop’s impedance and loss, the amplifier’s voltage/current noise, input-referred linearity, electrical symmetry, common-mode control, and stability. There is no universal NF target for every loop.
Why use a push-pull or differential stage?
Good balance can suppress even-order distortion and reject common-mode interference. It may also increase differential swing, but it does not automatically improve third-order linearity.
How close can an active RX antenna be to a transmitter?
There is no universal distance. Coupling depends on frequency, antenna type, current and voltage maxima, orientation, feedline routing, common-mode current, gain, and duty cycle. Use RF.Guru’s model-specific proximity guidance, start tests at low power, and use greater spacing, filtering, or a PTT-controlled disconnect/shorting relay where required. “Survives” does not mean “remains linear while transmitting.”
Technical References
- ITU-R P.372-17: Radio noise
- Analog Devices: Op-amp voltage noise, current noise, and source impedance
- Analog Devices: Differential interfaces in RF designs
- Mini-Circuits: P1dB and IP3 in receiver front ends
Interested in more technical content? Subscribe to RF.Guru updates for deep-dive RF articles and lab notes, or contact RF.Guru with questions and measurements.