Conjugate Match Is Real. “Matched Everywhere” Usually Isn’t.

Mark Weaver, K3ZD, with whom I collaborate on articles and YouTube projects, asked what I made of the long-running Maxwell-versus-Straw dispute over conjugate matching, transmission lines and tuners. It is a good question, because both men were defending something worth keeping—and the same phrase was carrying more than one meaning.

Walter Maxwell, W2DU, and Dean Straw, N6BV, are both Silent Keys. Both made substantial contributions to amateur radio. This was not a disagreement between people who knew nothing about the subject, and there is no value in turning it into rival camps. Better formulas, clearer wording and explicit assumptions are more useful than a louder argument.

Read the Exchange as Two Voices

In the annotated exchange Mark asked about, Straw’s article is reproduced in black and Maxwell’s inserted replies appear in red. The black article is dated July 2001; Maxwell’s response is dated January 2005, revised April 2011. Keeping the two voices separate matters: the commentary is not a single jointly agreed explanation.

Straw starts from a lossless-network result in Slater and questions what survives real loss. Maxwell defends the ordinary conjugate-match condition, then argues that retuning restores it throughout a lossy network in the forward direction. Later, their disagreement moves to the source impedance of a power amplifier. Those are distinct questions, not one yes-or-no vote on whether matching exists.

Where Maxwell Is Right: The Port Theorem Is Real

Replace a linear source network, at one chosen port and frequency, by its Thévenin equivalent:

For a load Z_L = R_L + jX_L and a fixed RMS Thévenin voltage V_Th, the average load power is:

That power is maximized when:

The result is exact for the declared linear source model at that port. Kurokawa’s primary power-wave treatment derives the same conjugate condition and relates incident, reflected and exchangeable power for complex port impedances.

Maximum load power is not maximum efficiency. If R_S is a physical dissipative source resistance, the conjugate condition dissipates equal power in R_S and R_L. Different load choices can improve efficiency while delivering less than the maximum available load power.

This is the part of Maxwell’s defence I would not discard. Loss upstream does not abolish the theorem; it becomes part of the equivalent source seen from the chosen load port. Straw’s broader dismissal goes beyond the valid warning about a lossless network’s internal symmetry. A theorem with stated assumptions is useful engineering, not folklore.

A 50 Ω Reference Match Is a Different Statement

For a real positive reference impedance Z₀, the voltage-wave reflection coefficient of a load is:

With Z₀ = 50 Ω, Γ = 0 when Z_L = 50 Ω. That means no reflected travelling wave at that plane in the defined 50 Ω system. It does not independently establish the internal Thévenin impedance of the transmitter, radiation efficiency, or a conjugate relationship at another interface.

Here the terminology can mislead both sides of the exchange. Transforming a load to the transmitter’s required 50 Ω is a definite, useful job. It is not, by itself, a measurement of the transmitter’s internal source impedance.

For complex reference impedances or lossy waveguides, ordinary voltage-wave and power-wave definitions are no longer interchangeable. NIST’s primary general waveguide circuit theory distinguishes travelling waves tied to characteristic impedance from pseudo-waves tied to an arbitrary reference impedance.

Where Straw Is Right: Calculate the Actual Network

Straw’s black-box approach on pages 8–10 of the exchange is practical: start with antenna impedance, transform it through the stated line, then work out the matching network needed at the shack. His TL/TLA/TLW discussion serves that job. I do not need to dismiss conjugate matching to agree that this is a useful way to design and diagnose a station.

A load impedance does not appear unchanged at the input of a finite line. For a lossless line of characteristic impedance Z₀ and electrical length βl:

For a lossy line, β becomes part of the complex propagation constant γ = α + jβ and the tangent form becomes a hyperbolic-tangent transformation:

A tuner can transform this input impedance to 50 Ω at the transmitter while the feedline remains mismatched to the antenna. Standing waves, line loss and elevated voltage/current can remain downstream. The source-facing plane is matched; the distributed system is not thereby “matched everywhere.”

A uniform line is reflection-free throughout when its termination equals the appropriate characteristic impedance and no discontinuity excites another mode. That travelling-wave condition is stronger and more specific than “the transmitter sees 1:1 SWR.”

It is also different from the lossless theorem discussed by Slater: a conjugate relationship between the networks looking in opposite directions can hold at internal cuts even while travelling-wave standing waves exist relative to the line’s characteristic impedance. “Conjugate at a cut” and “flat line” must not silently become synonyms.

Where “Matched Everywhere” Goes Too Far

On page 5, Maxwell argues that retuning the real network restores a conjugate match throughout it in the forward direction. That is the stronger claim I cannot accept as a general rule. Merely cancelling terminal reactance does not guarantee equality of the resistive parts at every internal cut.

At any chosen load port, everything upstream—including loss—can still be reduced to a Thévenin equivalent. Conjugating that equivalent maximizes power delivered to that load. It does not make every internal junction a simultaneous conjugate match.

Consider a 50 Ω source resistance followed by 10 Ω of series loss and then a variable resistive load. Looking upstream from the load, the equivalent resistance is 60 Ω, so maximum load power occurs at 60 Ω. At the internal junction, however, the network looking left is 50 Ω while the network looking right is 70 Ω. Those two impedances are not conjugates.

If instead I choose a 40 Ω load, the source sees 10 + 40 = 50 Ω beyond its own resistance. That maximises power leaving the 50 Ω source, but not power reaching the final load. With an illustrative fixed 10 V RMS ideal source, the load receives 0.400 W at 40 Ω and about 0.417 W at 60 Ω. Different reference planes give different optimisation problems, even in this simple resistive circuit.

This example does not refute Maxwell’s port theorem. It refutes the universal extension. Nor does it say that every lossy network must be mismatched at every point: particular networks and terminations can meet particular conditions. The tuner adjustment alone does not prove them.

A practical antenna system adds feedline attenuation, tuner loss, dielectric and conductor loss, and radiation resistance. Each dissipative element changes the available power and the equivalent impedance at the next plane. A low SWR can even improve toward the source because attenuation reduces the returning wave; that is not a performance benefit.

“Tuned” Does Not Specify the Optimization

An adjusted tuner can achieve several different goals:

• present an acceptable impedance to the transmitter;
• minimize reflection at the tuner input;
• maximize power accepted by the downstream network;
• maximize power reaching the antenna feedpoint;
• minimize tuner voltage, current or loss; or
• maximize radiated power after antenna loss.

Those optima need not coincide. A tuner setting that gives 1:1 SWR at its input can have more component loss than another acceptable setting. A feedline length can transform the antenna impedance into a convenient tuner load without reducing line standing waves. A network can maximize antenna-port power without maximizing radiation if the antenna loss or pattern also changes.

Active Sources Add Stability and Large-Signal Constraints

The amplifier part of the exchange deserves the same care. Maxwell correctly distinguishes the tube’s plate region from the external output port, but his page-3 replacement of source resistance with the operating ratio E/I is not a general way to identify a nonlinear transmitter’s Thévenin impedance. Terminal voltage divided by current describes the connected operating load; it does not independently reveal how the source responds to a change in that load.

Straw’s page-17 account instead names the required load line, output power, distortion, harmonic content and safe heat dissipation. That is a useful engineering account of what a power-amplifier matching network must achieve. It does not make linear conjugate matching invalid; it shows why a driven amplifier needs a more specific model.

A passive linear Thévenin source has a fixed small-signal impedance for the operating condition. A driven RF power amplifier is nonlinear and time-varying over its RF cycle. Its preferred load can depend on bias, drive, frequency, waveform and fundamental and harmonic terminations.

For a linear two-port, simultaneous conjugate matching is associated with maximum transducer gain only when the required source and load impedances exist and the network is stable for them. Rollett’s primary stability and power-gain invariant shows why termination choice cannot be separated from stability.

For a large-signal amplifier, load-pull varies the presented impedance and records the resulting output power, power-added efficiency, gain compression, linearity and other metrics. The current Keysight load-pull example explicitly shows that optimum source and load impedances change with the chosen performance target, frequency, bias and harmonic conditions.

Therefore a transmitter’s specified 50 Ω load is an operating requirement at its external reference plane, not permission to infer a fixed 50 Ω internal source resistance. Stability, protection and rated-load limits take precedence over experimental conjugate-loading of an active transmitter.

Exact Matching Is Also a Bandwidth Question

A reactive load can be conjugately matched at one frequency with a lossless network. Achieving a small reflection over a broad band is a different optimization. Passive causal matching networks face integral trade-offs between bandwidth and reflection for a given load.

Fano’s primary MIT report, Theoretical Limitations on the Broadband Matching of Arbitrary Impedances, establishes this boundary. Adding resistive loss can make the displayed match broader, but part of the accepted power then becomes heat. Bandwidth must therefore be reported with insertion loss and delivered power.

The Answer I Would Give Mark

It is a split verdict, not a diplomatic escape. Maxwell is right that conjugate matching is real and useful at a defined port under the appropriate linear-source assumptions. Straw is right that a real lossy station and a driven power amplifier cannot be described completely by carrying an ideal lossless result everywhere.

For the station on my bench, I would use Straw’s network calculation to determine the line input and tuner load, Maxwell’s port theorem where its assumptions apply, and the amplifier’s actual operating requirements at the source. Those are complementary tools once their jobs are stated. They let me choose a workable, lower-loss matching arrangement rather than argue over a slogan.

The useful questions remain short: which reference plane, which source model, and best for what? Maximum load power, minimum reflected power, transmitter comfort, efficiency and linearity are not interchangeable answers. Calibrated complex impedance and a loss budget connect the explanation to the installed station.

That is what I meant by keeping the assumptions attached. A partial truth becomes folklore when the conditions vanish and only “the system is matched” survives. We can respect both men, retain what each contributes and still say exactly where a claim goes too far.

Conjugate matching is not the myth. Promoting a valid port condition into a universal description of the entire station is.

Primary Sources and Scope Anchors

• Kurokawa, “Power Waves and the Scattering Matrix”—complex conjugate matching, exchangeable power and complex-reference power waves.
• Marks and Williams, “A General Waveguide Circuit Theory”—travelling waves, arbitrary reference impedances, lossy lines and network measurement.
• Rollett, “Stability and Power-Gain Invariants of Linear Twoports”—termination-dependent stability and maximum available gain.
• Keysight, Investigating Load Pull and DC Simulations of a FET—large-signal optimum-load trade-offs.
• Fano, Theoretical Limitations on the Broadband Matching of Arbitrary Impedances—passive broadband matching bounds.
• The Saga of Conjugate Match—historical amateur-radio exchange preserved for context.

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