Active MMIC or Passive Transformer Phasing for RX Arrays?

The old low-band receive arrays built with transformers, delay lines and passive combiners were not primitive mistakes. They demonstrated how much can be done with geometry and carefully controlled phase. Modern active networks add useful design freedom, but they also put semiconductor noise, overload, bias, stability and calibration into every channel. The sensible comparison is therefore not old versus new. It is measured passive network versus measured active network inside the same installed array.

The Array Adds Complex Channel Signals

For an N-element receive array, the combiner output at frequency f can be represented as:

Vn is the signal induced in element n. Hn is the complete channel transfer from that element to the combining reference plane: matching network, transformer or amplifier, filter, cable, switch and connectors included. wn is the intended complex weight. It contains both amplitude and phase.

The array pattern comes from the complete sum. Replacing a transformer with an MMIC cannot repair a wrong element spacing, an uncontrolled cable-current path or an incorrect phase convention. Conversely, a well-designed active stage can isolate variable element impedances and make calibrated weights easier to apply. Both statements can be true.

Small Channel Errors Fill Deep Nulls

A cancellation null is especially unforgiving. For two normalized channels, let one have fractional amplitude error ε and phase error δ radians from perfect opposition. The residual relative to one channel is exactly:

For small errors this is approximately √(ε2 + δ2). That is why a small connector change, termination error, temperature drift or element mismatch can be visible in a deep null even when it barely changes ordinary receive level.

Quoting a component's phase tolerance is not enough. The claim needs the complete channel, frequency range, element impedances, unused-port terminations, operating temperature and measurement uncertainty. An installed null then adds mutual coupling, ground, nearby structures and all unintended receiving conductors.

What Passive Transformer Networks Do Well

A passive transformer, transmission-line transformer, hybrid or combiner needs no bias supply and adds no semiconductor compression or active-device intermodulation. With the correct core, winding, transmission-line impedance and terminations, it can provide useful transformation, splitting, combining or phase relationships across a substantial band.

That does not make a wire-wound network ideal. At the low-frequency edge, magnetising inductance and core loss matter. At the high-frequency edge, leakage inductance, interwinding capacitance, conductor loss and layout become important. Applied signal and DC current can also move a magnetic component toward a different operating region. Mini-Circuits' transformer measurement note correctly treats bandwidth and insertion loss as properties of the exact transformer under declared source and load conditions, not of the word transformer.

A passive divider also has an unavoidable division relationship. Excess insertion loss is the additional real loss and mismatch beyond that ideal division. Keep those quantities separate. Channel-to-channel amplitude imbalance can damage the pattern even when total loss is acceptable.

What an Active MMIC Channel Can Add

An MMIC is a manufacturing form, not one circuit function. Depending on the exact device and surrounding network, an active channel may provide low-noise gain, impedance buffering, reverse isolation, variable attenuation, switching or phase control. Buffering can reduce interaction between a variable element impedance and the combiner. Gain placed before a later lossy network can reduce that later network's contribution to receiver noise.

Those advantages are conditional. Every active channel has a specified frequency range, noise figure, gain flatness, input and output match, reverse isolation, compression point, intercept behavior, bias range and stability boundary. A low noise figure does not imply high blocker tolerance. A large output third-order intercept does not by itself state input headroom, and an intercept is an extrapolated small-signal metric rather than a safe operating point.

Active gain can make matters worse when strong broadcast, local-transmitter or lightning-induced signals drive a device, filter, switch or following ADC into compression. Intermodulation products may then enter the wanted band. Preselection, attenuation, limiting and sequencing can be more valuable than another decibel of gain. Their placement must be included in the channel transfer and overload test.

Noise Figure Must Be Read Against External Noise

Loss before the first low-noise gain stage normally degrades the receiver noise factor. Early gain can make later losses less significant through the cascade described by Friis' relation. That is the valid reason to consider a feedpoint amplifier or active channel—not a promise that every active array hears better.

On much of HF, atmospheric, galactic and man-made noise arriving through the antenna can exceed receiver noise. ITU-R P.372-17 supplies statistical radio-noise information and explicitly distinguishes noise received through the antenna/feed system from noise coupled through other cables, inadequate screening or poor balance. Where external noise dominates by a healthy margin, reducing preamplifier noise figure further may not improve received SNR. On a quiet band, with a small inefficient element or after substantial passive loss, receiver noise can again matter.

Measure the system with the antenna connected and replaced by a suitable termination at a declared plane. Confirm that sky or environmental noise rises above the receiver floor without consuming the headroom needed for real blockers. Noise figure, antenna factor, gain and overload belong in one budget.

Isolation, Balance and CMRR Are Different Claims

Port-to-port isolation tells us how much a signal applied at one declared port appears at another under specified terminations. Amplitude and phase balance compare intended channel transfers. Common-mode rejection ratio compares differential and common-mode responses of a declared differential circuit. None of those numbers, alone, proves low current on a coax exterior.

An active buffer may improve reverse isolation between channels. A Wilkinson or transformer network may provide useful passive isolation when all ports see their required impedances. Either can perform poorly when a real antenna presents a changing complex load. Isolation resistors and unused ports must be terminated exactly as the topology requires.

Feedline-exterior current is an installed current-path problem. Coax, DC power, control wiring, Ethernet, protective earth, mast and enclosure can all become unintended receiving conductors. A 1:1 transformer may break direct-current continuity, but capacitance can still bridge the RF path; a separate choke may impede a measured exterior-current mode, but neither device should be called a guaranteed CMRR cure.

Fixed Phase, Delay and Electronic Control Solve Different Problems

An ideal delay τ produces phase -2πfτ. A nominal fixed-phase network produces a chosen angle only over the band where its measured response supports that description. An electronic phase shifter adds adjustability; it does not automatically become true time delay or remove beam squint.

Coaxial delay lines are frequency-dependent in phase by design. Their delay, loss and temperature coefficient still need characterization. Passive hybrids can provide accurate relative phase over a specified band. Active all-pass, vector-modulator or switched networks can provide recalibration and steering, but their phase, amplitude and group delay vary with frequency and control state.

Decide whether the array requires one fixed pattern, switched directions, continuous null steering or simultaneous digitized channels. Then choose hardware whose measured transfer supports that job. Do not demand continuously adjustable electronics from a fixed broadside array, and do not demand a fixed passive box to perform adaptive cancellation.

Temperature Stability Comes from Measurement

Ferrite permeability, cable delay, semiconductor gain and phase, PCB dielectric properties, bias regulators and terminations all vary with temperature. A low-drift substrate can help an active design, just as stable cable and controlled magnetic material can help a passive one. Neither construction style justifies a blanket phase-drift number.

Characterize each complete channel over the required temperature and supply range. If the system recalibrates, state how often, at which reference plane and with what reference signal. Calibration can remove repeatable complex error inside its fitted conditions; it cannot remove an antenna pattern that changed with wet soil, ice, a moved cable or a new nearby conductor.

Compare the Two Architectures at the Same Planes

A Measurement Plan That Can Decide

• Declare the array objective. Name the bands, wanted headings, rejected directions, polarization, bandwidth, SNR target and blocker environment.
• Fix reference planes and terminations. Show where the element, network, cable, receiver and calibration begin and end.
• Measure every complex channel. Record gain or insertion loss, phase, group delay, input/output match, isolation and state-to-state repeatability.
• Test noise and headroom. Measure noise figure or noise temperature where it matters, plus compression, two-tone intermodulation and recovery with the actual preselection.
• Sweep temperature and bias. Include cable, transformer, active stage, regulator and enclosure in the same test.
• Map unintended currents. Check coax exteriors, power and control leads, mast and bonds on every operating band.
• Verify the installed pattern. Measure azimuth and the relevant elevation region, or use a controlled wanted-signal/noise A/B/A test with the baseline restored.
• Keep uncertainty visible. A null below the measurement floor or a tiny SNR difference is not a quantified advantage.

Joeri's Bottom Line

Active phasing is powerful when I need buffering, gain, controllable weights or repeatable electronic calibration. Passive phasing remains powerful when a fixed network already meets the loss, balance, bandwidth and isolation requirement. I will not call transformers obsolete, and I will not call an MMIC modern evidence.

The winner is the architecture that keeps the wanted signal-to-noise ratio, survives the real blocker environment and produces the intended installed pattern across the required band. Measure both candidates at the same planes. Then the array—not the fashion—gets the final word.

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