When Entertainment Outruns Engineering
When Entertainment Outruns Engineering
One lively interview connected polar modulation, switching PA classes, mature components and end-fed return currents. Those are worthwhile questions—but they need four different engineering models.
Why this article exists: I watched Dave G3LRC interview Josef Hoffman, KD7QOW. Dave is an engaging host, and the conversation raises questions worth exploring. The problem begins when several related-sounding subjects are allowed to share one convenient conclusion.
Polar or EER architecture, Class D and Class E switching, component qualification and the current path of an end-fed antenna do not belong in the same drawer. This article follows the interview because that context matters: it keeps the useful curiosity, then tests each technical claim at the boundary where it actually lives. This is about the engineering, not the personalities.
Interview context: the embedded video is G3LRC’s interview with Josef Hoffman, KD7QOW. The article responds directly to its discussion of polar modulation, switching PA classes, component age in commercial radios and end-fed return currents.
Curiosity is welcome. But a transmitter topology cannot be judged from its name, output-power headline or transistor generation, and an antenna return path cannot be settled with a slogan. The complete systems—not the labels—have to be examined.
Evaluation rule: name the signal representation, every power-conversion stage, the measurement plane and the installed RF-current boundary. Then test efficiency, linearity, emissions, device stress and load tolerance under the same declared conditions.
Polar Representation and EER Are Related, Not Identical Labels
A narrowband complex RF signal can be written as:
x(t) = A(t) cos[ωct + φ(t)]
The envelope A(t) and phase φ(t) are two descriptions of the same modulated signal. A polar transmitter processes those quantities in separate paths and recombines them at or around the RF power stage.
In classic Kahn envelope elimination and restoration, a modulated RF input is limited to create a substantially constant-envelope phase path, while an envelope detector recovers the amplitude path. A nonlinear RF power amplifier raises the phase-path power, and high-level amplitude modulation restores the envelope. The architecture can therefore accept an RF input; it is not restricted to a transmitter that generated the modulation internally.
A digitally generated polar transmitter can instead calculate magnitude and phase from baseband I/Q samples before the carrier exists. This can remove analogue envelope detection and share clocks, calibration and control, but it does not remove the two-path problem. EER is a particular envelope-removal/restoration implementation within the broader polar family, not a synonym for every supply-modulated or envelope-tracking PA.
Why an External RF-Input Stage Is Not Automatically Universal
An external EER stage is physically possible. Making one operate transparently behind arbitrary transmitters, modes, drive levels and protection systems is a much stronger requirement. The design still has to control:
- the amplitude and phase transfer functions over the full modulation bandwidth;
- relative delay between the envelope and phase paths;
- envelope-path slew rate, dynamic range and behaviour near envelope minima;
- AM/AM and AM/PM conversion in the switching PA and modulator;
- drive-level variation, ALC interaction and transmitter transients;
- load mismatch, reflected energy and the protection response;
- output filtering and out-of-band emissions; and
- feedback or predistortion at a clearly defined sampling plane.
A timing error does not merely change audio delay. Because the two paths multiply at RF, delay, gain and phase errors generate in-band distortion and spectral regrowth. The acceptable mismatch depends on modulation bandwidth, error-vector or intermodulation target and the applicable emission mask; there is no universal delay number.
Architecture boundary: “external” describes packaging, not independence. A practical external polar/EER stage remains coupled to the exciter’s waveform, bandwidth, control timing and protection behaviour.
Scaling From Integrated QRP to Higher Power
An integrated low-power transmitter can share the digital signal path, reference clock, envelope calibration, PA control, filters and protection inside one known platform. That makes the boundary well defined. It does not make linearity or emissions automatic; they still have to be measured with representative speech, two-tone and other intended waveforms.
Higher RF output increases the absolute value of every loss and magnifies several design constraints:
- the envelope modulator must process greater average and peak current without losing bandwidth;
- switching loss, device output capacitance, package and layout inductance become significant heat and waveform errors;
- drain or collector voltage, current, safe operating area and mismatch energy require wider protection margins;
- the output network must handle current, voltage and harmonic energy while retaining the required bandwidth;
- power combining, current sharing and device matching may become necessary;
- cooling, enclosure, supply, conducted emissions and radiated emissions become system-level constraints; and
- a protection event must coordinate the envelope, phase drive, supply and antenna load quickly enough to avoid damage.
The possible thermal benefit also grows with power: saving 10 percentage points of loss at 500 W output removes much more heat than the same percentage at 5 W. That observation does not create a universal power threshold or business case. Complexity, modulation, duty cycle, production volume, compliance, service objectives and the competing linear-PA design still determine whether polar/EER is worthwhile.
ηsystem = PRF,out / PDC,total
PDC,total = Pphase path + Penvelope path + Pdrivers/control + Pauxiliary
For a modulated transmitter, average the powers over a declared waveform and duty cycle. Drain efficiency of the RF switch alone omits envelope-modulator loss, drivers, control, filters and auxiliaries. Peak efficiency, average efficiency, PEP capability and continuous thermal rating answer different questions.
Class D and Class E Are Different Switching Networks
Both classes seek low simultaneous voltage and current in the active devices. They shape that condition differently and impose different network and stress requirements.
| Property | Class D RF PA | Class E RF PA |
|---|---|---|
| Typical switching structure | Push-pull, half-bridge or full-bridge voltage- or current-mode switching | Often a single switch with shunt capacitance and a tuned load network |
| Waveform principle | Devices commutate between states; the output network selects the wanted RF component | The load network shapes switch voltage so turn-on occurs at low voltage and, in the nominal case, low voltage slope |
| Critical non-idealities | Dead time, overlap, switching charge, output capacitance, transformer or combiner imbalance and layout inductance | Load and duty-cycle sensitivity, shunt capacitance, device output capacitance, network Q, peak switch voltage and component loss |
| Load network role | Filters switching harmonics and transforms the load; topology depends on voltage- or current-mode implementation | Defines the switch waveform as well as filtering and load transformation |
| What the class label does not prove | Efficiency, linearity, bandwidth, harmonic compliance, mismatch survival, device temperature or suitability for envelope restoration | |
The classic Class E conditions are conditional on its chosen load, frequency, duty cycle and component values. Device capacitance can be part of the intended shunt capacitance, but its nonlinear voltage dependence and layout parasitics still matter. Class D likewise depends on commutation, dead time and the realised output network. Hybrid classes such as Class DE further show why “switching PA” is only the start of a description.
Component Qualification Is an Application Record
Publication date is not a qualification metric. A mature component can be the strongest choice when its behaviour, variation, availability, production process and field performance are understood. A newer component can deliver a real advantage, but only after the surrounding design and qualification evidence are updated.
A defensible selection record covers:
- voltage, current, power, junction temperature, transient and safe-operating-area margins;
- large-signal RF behaviour, nonlinear capacitance, switching charge, gain, stability and ruggedness;
- package, PCB layout, cooling, matching network and manufacturing tolerances;
- qualification and reliability data for the relevant process, package and application class;
- process-change notification, errata, second-source strategy and lifecycle risk;
- firmware, calibration, protection and production-test dependencies; and
- compliance, repair stock, documentation and service impact.
Semiconductor quality systems formalize qualification, change notification and reliability evidence because a part change can affect form, fit, function, quality or reliability. In a transmitter it can also force new PCB, thermal, matching, filtering, firmware, protection, calibration, EMC and production validation. Retaining a qualified component or adopting a new one can both be rational outcomes.
The Installed End-Fed Antenna Still Needs a Complete Current Path
A matching transformer changes the impedance presented at its ports; it does not remove current continuity. At an installed end-fed antenna, current can close through a deliberate counterpoise, the exterior of the feedline shield, a mast or other conductors, and distributed displacement current to the surroundings. Several paths can exist simultaneously.
Inside an ideal coaxial transmission-line mode, centre-conductor current is paired with opposite current on the shield’s inner surface. Exterior-shield current is a separate common-mode path. If it is significant, the feedline can affect radiation pattern, receive-noise pickup, feedpoint impedance and RF in the operating area.
A common-mode choke inserts frequency-dependent complex impedance into that exterior circuit. Its result depends on the existing common-mode impedance, choke placement, cable electrical length, transformer parasitics and all alternate paths. Moving the choke can increase or decrease current at a particular point and can move voltage or heating elsewhere; it does not erase the need for a complete installed path.
Measure safely: begin at low power with the transmitter isolated whenever wiring or instruments are changed. A calibrated RF current probe around the complete coax can map the uncancelled exterior current at several positions. Stay within instrument, choke, connector and exposure limits, and never disconnect required protective earth to alter RF behaviour.
Evidence Checklist for the Complete System
| Boundary | Minimum evidence |
|---|---|
| Signal architecture | Block diagram, signal planes, sample rates or bandwidths, path-delay calibration and recombination point |
| RF and envelope power stages | Efficiency at the same waveform, output and supply planes; device waveforms; temperature; linearity; emissions |
| Scaling | Peak and average current, voltage, thermal, mismatch, filter, protection and duty-cycle results at the stated output power |
| Components | Current data, qualification basis, derating, change control, availability, production variation and service plan |
| Antenna installation | Feedpoint impedance, exterior cable-current map, choke impedance/temperature, cable routing, exposure and protective-earthing checks |
No single wattmeter, SWR trace, efficiency peak or component date can validate all five boundaries. A credible result states what was measured, at which plane, with which waveform and uncertainty, and which parts of the installed system remain conditional.
Engineering References
- Kahn: Single-Sideband Transmission by Envelope Elimination and Restoration
- Raab et al.: Power Amplifiers and Transmitters for RF and Microwave
- Sokal and Sokal: Class E Switching Power Amplifiers
- Pedro, García and Cabral: Nonlinear Distortion Analysis of Polar Transmitters
- Documented Integrated QRP Polar-Modulation Implementation
- Manufacturer Application Report: GaN Driver and Class D Envelope Amplifier
- Manufacturer Qualification and Reliability Process
- Manufacturer Change Management and Product-Change Notification
- ITU-T K.37: Common-Mode Control, Earthing, Bonding and Cabling
Mini-FAQ
- Must a polar or EER power stage be integrated inside the exciter? No. Classic EER can accept a modulated RF input, but an external stage must still extract, align and recombine envelope and phase while controlling distortion, emissions and protection interactions.
- Does a successful QRP polar transmitter scale directly to higher power? No. The architecture can scale, but envelope current, switching loss, parasitics, voltage stress, filtering, cooling, protection and mismatch energy must be qualified at the new power and waveform.
- Are Class D and Class E interchangeable names? No. Both are switching approaches, but Class D commonly uses bridge or push-pull commutation, while Class E uses a tuned network to shape the switch voltage at turn-on.
- Does a mature transistor make a transmitter technically inferior? No. Part age is not a performance metric. Selection depends on electrical and thermal margins, RF behaviour, qualification, availability, change control, production variation and service requirements.
- Does an end-fed matching transformer eliminate the return path? No. It transforms impedance. Current can still use a deliberate counterpoise, feedline-shield exterior, nearby conductors and distributed displacement-current paths.
- Does adding a common-mode choke guarantee zero feedline current? No. A choke adds finite complex impedance to the installed common-mode circuit. Its effect depends on placement, cable length, alternate paths, frequency, voltage, loss and temperature.