Where the watts really go

Every watt needs a name tag: location, waveform, and averaging time. With those three details, the number on the radio becomes useful engineering information instead of shack folklore.

01 / STATION POWER

First, ask the watts where they work

The big knob is persuasive, but it does not tell the whole story. A 1,000-watt transmitter is ten times the RF output power of a 100-watt transmitter when both numbers describe the same quantity at the same reference plane. It does not follow that the signal sounds ten times louder, that the antenna radiates all of it, or that every direction receives ten times the field strength. This guide follows the energy and separates signal peaks from sustained operation.

A quick tour of the shack

Sections 2–4 define power and the feed-system budget. Sections 5–7 compare operating modes, including conventional AM. Sections 8–10 cover electrical supply, component stress, and RF exposure. Sections 11–12 turn the calculations into practical station worksheets and decisions. References follow at the end.

Meet the five power levels

*At a fixed far-field point, with unchanged antenna, propagation, and impedance. Power density scales with power; electric-field amplitude scales with its square root. These are calculated ratios, not guaranteed reception improvements.

Scope: general amateur-radio engineering, with U.S. regulatory examples clearly identified. The example powers are comparison values, not permission to use those powers on every band or licence. [1–3]

02 / STATION POWER

Watts have peaks, pauses, and a schedule

PEP: a peak with an average hiding inside

The name sounds like a contradiction, but the two averages happen on different clocks. PEP is the RF-cycle average at the crest of the modulation envelope. It is not the sub-cycle maximum of instantaneous voltage multiplied by current. The U.S. amateur definition appears in 47 CFR 97.3(b)(9). A clean, unmodulated carrier has equal PEP and RF-cycle average power; speech and other changing envelopes do not maintain their crest level continuously. [1, 8]

Pmean = (1 / T) × integral of pRF(t) over T

For operating comparisons, define a waveform factor F as mean RF power during a stated transmit interval divided by PEP. Define a separate transmit fraction D as time transmitting divided by the complete observation interval. If output level and pattern are stable, the interval average is PEP × F × D. State exactly which pauses F already includes so that D does not subtract them a second time.

Three clocks walk into a radio shack

Two tones are not a chatty operator

For two equal, clean tones in a linear SSB system, mean RF power is one-half the PEP: a 500 W PEP two-tone test averages 250 W. That follows from adding the two sinusoidal voltage envelopes, then averaging their squared sum. Real speech has no fixed universal peak-to-average ratio. Compression, clipping, microphone gain, and the speaker change it. A two-tone test is useful for linearity; it is not a model of every voice transmission.

Your meter needs the right job description

A wattmeter is helpful company, provided you ask it a question it can answer. An averaging wattmeter may read much less than the PEP of speech. A peak-reading instrument needs adequate envelope response and the correct operating mode. A slow display can miss leading-edge overshoot. Record detector type, calibration date, frequency range, load, measurement uncertainty, and whether the value is forward, reflected, or net power. [8, 9]

03 / STATION POWER

Follow the watts (and the warm bits)

Think of the station as an energy ledger: the watts need to balance, even when the story gets complicated. Use a calibrated antenna-port plane. For real reference impedance, incident and reflected traveling-wave powers yield net accepted power by subtraction. Keep the same waveform quantity throughout the chain. The following first-order equations assume a linear, stable system at one frequency. [4, 5]

Paccepted = Pincident − Preflected = Pincident × (1 − |Γ|²)

Pradiated = ηradiation × Paccepted

EIRP(direction) = Paccepted × G(direction)

Gain G includes radiation efficiency but excludes port mismatch. Realized gain includes port mismatch, so the alternative is EIRP = Pincident × Grealized. Multiplying accepted power by realized gain counts mismatch twice. Gain and realized gain also require a declared reference impedance and direction. [4, 5]

A tidy example before things get interesting

Assume measured one-way dissipative loss of 1 dB, a matched antenna port, radiation efficiency 80%, and directional gain 3 dBi referenced to accepted power. Gain already includes that 80% efficiency. Thus EIRP is accepted power × 1.995; it is not radiated power × 1.995.

What if the antenna sends some back?

At 3:1 SWR, |Γ| = (3 − 1)/(3 + 1) = 1/2. At the antenna port, 25% of the incident wave power is reflected on that encounter, and 75% is accepted. If a directional measurement there shows 500 W incident and 125 W reflected, accepted power is 375 W. With 80% radiation efficiency, radiation is 300 W. These are port measurements; they do not specify transmitter output without a line-and-source model.

A reflected watt has changed direction; that alone does not tell you its final destination. Reflected power is not automatically dissipated or permanently missing. Re-reflections, tuner behavior, transmitter foldback, and loss on repeated line traversals determine the complete result. Do not use matched cable attenuation as if it fully described a high-SWR line, or apply a second mismatch penalty to a net-transfer measurement that already includes it. [6]

04 / STATION POWER

The cable would like a word about heat

Decibels are wonderfully consistent. Unfortunately, so are their losses. For a matched transmission path with loss L in decibels, the remaining fraction is 10^(−L/10). The table is calculated directly from that equation. It applies to the stated path loss, not to an unknown cable merely because its catalogue lists a similar number.

Values are watts remaining at the output plane. With matched terminations, the difference becomes heat in the path.

One decibel, an increasingly warm bill

The percentage stays polite; the heat bill does not. At 100 W, 1 dB dissipates 20.6 W. At 1,000 W, it dissipates 205.7 W. The fractional penalty is unchanged, but the thermal burden is ten times larger. Heat may be distributed among cable, connectors, filters, switches, and a matching network. A single poor connection can concentrate part of the loss into a small hot spot.

The tuner is an impedance translator

A happy SWR meter at the desk does not guarantee a carefree cable outdoors. A shack tuner can present a good load to the amplifier while the outdoor line still operates at high SWR. A tuner at the antenna may reduce mismatch on the main cable, but its own losses, insulation, enclosure, and duty rating still matter. Compare net power before and after each stage under the intended load; record the antenna-side impedance and frequency. [6]

A picture from the shack

Some watts take the scenic route into heat

The cable is not keeping a secret stash of watts. In a matched feed system, dissipative loss turns part of the transmitted RF power into heat along the path.

At 1 dB matched loss, 100 W becomes 79.4 W at the output plane, while 1,000 W becomes 794.3 W. The same percentage loss produces ten times the heat at ten times the power.

Conceptual illustration: teal parcels represent energy continuing toward the antenna; orange wisps represent dissipation. They are not a map of radiation leaking from the cable.

05 / STATION POWER

Same radio, very different workout

The radio has several ways to spend its working day. These normalized traces show envelope power conceptually. They are not RF voltage oscillograms, measured duty factors, or a promise about the output of a particular radio.

A digital signal through “USB” is still a digital workload

The display says USB; the heatsink would like to know what you are actually transmitting. The radio may display USB while a sound-card tone produces sustained RF. The front-panel mode name does not impose a voice duty factor. ALC action, audio clipping, or amplifier compression can broaden emissions. Set drive and audio levels according to the actual radio and amplifier instructions, and verify a clean signal rather than treating a specific ALC indication as universal advice.

No fixed power percentage is correct for every digital mode and transmitter. Some equipment supports full rated continuous output; other equipment specifies reduced output or time limits. Follow the continuous-duty rating, cooling conditions, SWR constraints, and manufacturer instructions.

A picture from the shack

The heatsink reads the whole transmission

Speech takes breaths. Morse has gaps. A sustained carrier keeps the RF stage working throughout its active interval. The same peak rating can therefore produce very different average thermal demands.

Compare waveform factor and transmit fraction separately. A sustained burst can have a waveform factor near 1 even when pauses between bursts lower its longer-term average.

Conceptual illustration: the three envelope strips suggest speech, keyed Morse, and a sustained transmission. Shapes are illustrative, without measured scales or prescribed duty factors.

06 / STATION POWER

Average power: the longer story

This is where the peaks settle down and the averages get a turn. The following are calculated comparisons at equal PEP, not measured values for a specific operator. F applies over the selected transmit interval; D is set to 1 here. The 20%, 40%, and 50% choices illustrate common planning conventions, and 100% is the sustained-carrier case. Actual measurements override convenient assumptions. [6, 7]

All numeric entries are mean RF watts during the stated transmit interval. Common associations are unprocessed conversational speech near the 20% planning example, conversational Morse near 40%, heavily processed speech near 50%, and steady FM/FSK near 100%. They are not guaranteed properties or manufacturer derating instructions.

Now let the operator take a turn

For a 600 W PEP SSB example, choose F = 0.20 and D = 0.50. Mean transmitter output over the complete interval is 60 W. With 1 dB matched loss, mean antenna-port power is 47.7 W. A 600 W steady digital burst has F approximately 1 during its active portion; at D = 0.50, the corresponding antenna-port average is 238.3 W. The same PEP produces five times the average in this example.

Count each pause once. It has earned its rest.

Pauses deserve credit, but they do not get to claim it twice. A hypothetical schedule with 12 seconds of full-envelope transmission in every 30 seconds has D = 0.40. If your measurement already averages over all 30 seconds, do not multiply by 0.40 again. Real software timing, aborted transmissions, retries, operator behavior, and sequence selection change the schedule. Use observed active RF time instead of assuming that every nominal transmit slot contains a carrier for its entire duration.

The heatsink and the exposure clock have different jobs

A fan and heatsink must withstand each sustained transmit burst even if the longer operating average is modest. Ferrite and small matching components can warm quickly. Exposure evaluation uses the legally applicable interval and actual access conditions; a casual average over an entire evening does not establish compliance during a concentrated run of transmissions. [2, 6]

07 / STATION POWER

AM brings a carrier—and some fine print

AM travels with a carrier, so its power arithmetic comes with an extra suitcase. This section describes conventional double-sideband AM with a full carrier and sinusoidal amplitude modulation. Let Pc be unmodulated carrier power and m the modulation index, from 0 to 1. The envelope voltage is proportional to 1 + m cos(ωt). Squaring it and averaging gives the relationships below. These are ideal waveform derivations, not ratings for any particular transmitter.

PEP = Pc × (1 + m)²

Pmean = Pc × (1 + m² / 2)

At 100% sinusoidal modulation, m = 1: PEP = 4Pc and mean RF power = 1.5Pc. The carrier remains present through speech pauses. Ordinary speech is not a perpetual 100%-modulated sine wave, so its long-term mean must be measured or derived from the actual waveform. Peak behavior also changes with asymmetrical processing. [8, 10]

If the comparison rating means maximum PEP

If the quoted number instead means carrier watts

Thus “1,000 watts AM” could mean a 250 W carrier with 1,000 W PEP, or a 1,000 W carrier needing 4,000 W PEP under the ideal test condition. That distinction affects amplifier headroom, component insulation, and permitted operation.

Getting AM ready without unwelcome surprises

Do not assume an SSB-rated amplifier may run one quarter of its PEP indefinitely as an AM carrier. Carrier dissipation, modulation topology, power-supply capability, cooling, and manufacturer restrictions govern. Overmodulation and amplifier clipping produce distortion; evaluate the envelope and occupied spectrum under the intended audio conditions.

08 / STATION POWER

The watts your power supply knows about

The signal gets the applause. The power supply pays for the performance. RF output is only part of the electrical demand. For a simplified amplifier stage with efficiency η, DC input is RF output / η, and stage heat is DC input − RF output. These calculations exclude receive electronics, drivers, fans, control circuits, and supply losses unless those are explicitly included in the efficiency measurement.

The 13.8 V column is an arithmetic comparison, not a suggested supply architecture. Many higher-power amplifiers use other internal voltages or AC supplies. At 1,000 W output and 50% stage efficiency, the stage requires 2,000 W DC and rejects 1,000 W as heat while transmitting.

Efficiency does not wear the same hat all day

Efficiency is a moving target, not a permanent personality trait. A linear amplifier operated below maximum output may have different efficiency from its rated full-output figure. Bias current, drive, frequency, load, and envelope shape matter. Estimate average electrical input from an appropriate measurement or a waveform-aware model; do not simply apply one full-power efficiency number to all speech conditions.

Supply cables can have expensive appetites

Voltage drop = current × loop resistance; cable heat = current² × loop resistance

A little resistance can become a surprisingly enthusiastic heater. For an illustrative 0.020-ohm supply-cable loop, 20 A causes a 0.40 V drop and 8 W heating. At 80 A, that becomes 1.60 V and 128 W. The loop includes both outgoing and return conductors and contacts. Cable ampacity, protection, connector ratings, and permissible voltage drop are separate constraints; size the system to the actual equipment instructions.

A kilowatt visits the wall socket

If the RF stage requires 2,000 W DC and the supply is 90% efficient, AC input is about 2,222 W before additional station loads. Supply loss is about 222 W. This is a declared hypothetical example; use measured AC input or manufacturer data for branch-circuit planning, including power factor where applicable.

A picture from the shack

The power supply buys the whole meal

Only part of the electrical input becomes RF output. The rest of the amplifier-stage energy budget becomes heat, which is why the fan gets a supporting role in the story.

In the guide’s 50%-efficient stage example, 1,000 W RF output requires 2,000 W DC input and produces 1,000 W of stage heat while transmitting. Drivers, fans, and supply losses are additional unless explicitly included.

Conceptual illustration: the energy parcels and orange heat wisps show an accounting split, not a measured efficiency or a wiring arrangement.

09 / STATION POWER

Small print, big volts

The watts are only part of the component shopping list. For a sinusoidal traveling wave delivering P watts into a matched 50-ohm load: Vrms = √(50P), Irms = √(P/50), and Vpeak = √2 × Vrms. If P is PEP, these describe the RF sinusoid at the envelope crest. They do not describe every point inside a mismatched tuner.

Standing waves have their favorite hot spots

At a given incident traveling-wave amplitude, voltage maximum is multiplied by 1 + |Γ| and current maximum by the same factor, at different positions on an ideal lossless line. At 3:1 SWR, |Γ| = 0.5, so each maximum is 1.5 times its incident-wave amplitude. If instead net delivered power is held fixed, the comparison changes: maxima rise by √SWR relative to a matched line carrying that net power. State which power is held fixed.

Inside the tuner, things can get lively

Inside a tuner, a calm front panel can hide a very busy electrical neighborhood. Coils and capacitors may carry circulating current or voltage far above the simple matched-port values. High transformation ratios and network Q can create serious stress even when the input SWR is excellent. The current and voltage limits, duty cycle, frequency, and load range of a tuner matter as much as a large wattage printed on its enclosure.

Read the rating before it becomes a story

A large wattage printed on a box is the start of a conversation, not the end. For coax, connectors, switches, filters, baluns, chokes, and dummy loads, check continuous versus PEP ratings, maximum voltage, frequency, ambient temperature, ventilation, mismatch, and operating duration. A 1,000 W PEP voice rating does not establish 1,000 W continuous data capability. A dummy load can have a short-time energy limit as well as a steady thermal rating.

A choke suppresses unwanted common-mode current; it is not a universal cure for a missing antenna return path. Its resistive impedance and actual common-mode current determine heating. Treat feedline routing, antenna balance, and connected structures as part of the station rather than assuming that a good differential SWR guarantees a quiet or cool feed system.

A picture from the shack

A calm meter can hide a busy tuner

A suitable match at the tuner input helps the transmitter see its intended load. It does not establish low standing waves beyond a shack tuner or low voltage and current inside the matching network.

Component limits depend on the actual load, frequency, transformation, network Q, and duty cycle. Check the applicable continuous-power and peak ratings; a reassuring input SWR cannot replace those checks.

Conceptual cutaway: the enlarged coil, capacitor, glow, and arrows illustrate internal electrical stress. They are not temperature readings, field plots, or a circuit design.

10 / STATION POWER

Where the signal goes—and who is nearby

An antenna can aim the performance; it cannot conjure extra energy out of thin air. EIRP expresses directional radiation relative to an isotropic source. It is not additional physical energy created by antenna gain. With the same total radiation, concentrating the pattern raises intensity in some directions and reduces it in others. ERP uses a half-wave dipole reference: EIRP is approximately 1.64 × ERP, or 2.15 dB higher for the same physical radiation. [1, 4]

A useful equation—with a defined job

S = EIRPaverage / (4πr²); r = √[EIRPaverage / (4πSlimit)]

Use watts, metres, and watts per square metre. This free-space far-field estimate excludes ground reflection, near-field structure, coupling to buildings, and multiple sources. It is an educational screening model, not a site-specific compliance distance. At HF, accessible positions may lie in a near field where separate electric and magnetic field evaluation is needed. [2, 6]

Calculated illustration: 28 MHz, sustained carrier

Assume 1 dB matched feed loss, antenna gain 6 dBi in the direction of the person, continuous transmission, and a free-space far field. At 28 MHz, the U.S. general-population MPE plane-wave-equivalent power-density limit is 180/f² = 180/28² = 0.2296 mW/cm² = 2.2959 W/m², with f in MHz. [2]

These radii omit reflections and do not establish compliance. At 28 MHz, wavelength is about 10.7 m, and these distances may lie outside the valid far-field region of the actual antenna. In a near field, evaluate electric and magnetic fields separately against the applicable limits: 824/28 = 29.43 V/m and 2.19/28 = 0.0782 A/m for U.S. general-population exposure. Antenna dimensions and geometry matter; use a suitable validated method for the installation. [2]

Frequency and people both get a vote

U.S. exposure limits depend on frequency and exposure category. The rule provides averaging periods up to six minutes for occupational/controlled exposure and up to thirty minutes for general population/uncontrolled exposure. Controlled exposure requires awareness and the ability to control exposure; ownership of the station alone is insufficient. Do not apply these U.S. examples as rules for another country. [2]

11 / STATION POWER

Give your station a useful paper trail

The best station notes save your future self from having to play detective. A meaningful evaluation combines measurements with the actual operating plan. Assess accessible positions, every operating band, antenna direction, and the operating condition producing the highest relevant exposure. A low transmitter setting alone does not establish an exemption or compliance. Current U.S. amateur guidance describes station evaluation obligations following the updated exposure rules. [3, 7]

A sensible walk around the station

1. Begin with the equipment manual and an appropriately rated dummy load. Verify output with a meter suitable for the waveform, then check a sustained condition only within the stated equipment duty limit.

2. Establish the measurement planes. Measure forward and reflected power at the same plane; distinguish net transfer from matched attenuation. Estimate loss under the intended frequency and load rather than borrowing a generic cable figure.

3. Check thermal behavior through the intended transmission sequence. Look for output drift, protection action, and heating at connectors and matching components. Investigate anomalies before escalating output.

4. Evaluate the actual antenna installation with a method appropriate to its field region. Where modeling assumptions are poor, obtain suitable RF field measurements or competent technical assessment. Apply the applicable rules and record assumptions.

5. Revisit the record after changes in power, band, antenna, feedline routing, operating mode, access, or transmission schedule.

Averaging is not a cure for contact hazards

RF contact burns, high voltage, and electrical installation hazards require their own controls. Keep people away from energized antenna conductors and matching components. A favourable time-averaged field estimate does not make touching a transmitting antenna acceptable.

12 / STATION POWER

Before you turn the big knob

100 watts: establish a measured baseline

Verify power delivery, feed loss, and antenna performance before assuming the rated number reaches the antenna. A 3 dB path improvement doubles delivered power at every output level and may also improve receive sensitivity when feed loss is a meaningful part of the receive-noise budget.

200 watts: a 3.01 dB step

The output doubles relative to 100 W, while the matched-port voltage and current each rise by √2. Check supply demand, tuner ratings, and permitted operation rather than assuming all 100 W accessories have adequate margin.

500 watts: a 6.99 dB step

Five times the output of the baseline makes existing fractional loss five times the absolute heat. Amplifier drive, linearity, cooling, and the feed path need attention. Sustained digital operation may impose a substantially larger average load than conversational speech at the same PEP.

600 watts: a 7.78 dB step

Moving from 500 to 600 W adds only 0.79 dB while raising RF output and approximately proportional loss heating by 20% under unchanged conditions. Compare that modest signal increment with available thermal and electrical margin.

1,000 watts: a 10 dB step

The ideal received power is ten times the 100 W baseline for an unchanged link, while field amplitude is 3.16 times higher. Moving from 600 to 1,000 W adds 2.22 dB. Check every intervening component, continuous-duty capability, exposure evaluation, and applicable operating limit before using the additional output.

The ionosphere has not read your power budget

Propagation has a talent for making tidy predictions look optimistic. For an unchanged linear path, received power scales with transmitted power. Actual HF reception varies with propagation, polarization, fading, local noise, and interference. S-meter calibration also varies, so do not promise a fixed number of S-units from a wattage change. Antenna pattern or noise improvements can matter more than a modest power increase.

The note to leave beside the big knob

State output quantity and plane; state the mode and time interval; account for feed and antenna loss; preserve peak headroom and thermal margin; and evaluate the real installation. Those statements make each of the five power levels comparable and repeatable.

13 / STATION POWER

References and calculation notes

Sources checked October 22, 2026. Numbered references are linked below. Engineering examples, tables, and charts are original calculations from the stated assumptions; they are not measurements or certifications of a station. Cartoon illustrations are conceptual and carry no calibrated field or circuit information.

1. eCFR — 47 CFR 97.3: PEP, EIRP, and ERP definitions.

2. eCFR — 47 CFR 1.1310: exposure limits, categories, and averaging.

3. eCFR — 47 CFR 97.313: transmitter power standards and exceptions.

4. IIT Madras — Wireless System Design: antenna gain, efficiency, realized gain.

5. Ansys HFSS — Peak Realized Gain: incident-power reference.

6. ARRL — RF Exposure Calculator Instructions: input conventions and mode factors.

7. ARRL — RF Exposure: updated U.S. station evaluation guidance.

8. Rohde & Schwarz — Application Note 7TS02: crest factor and envelope power, especially pages 3–5.

9. NTIA ITS — Gain characterization of the RF measurement path: reference planes and loss.

10. Rohde & Schwarz — AM modulation information: carrier headroom and 6 dB peak relationship.

11. ARRL — RF Exposure Calculator: practical screening tool and limitations.

12. RF.Guru — Follow-up reading: 100-watt station power-budget guide.

Rounding and conventions

Tables use unrounded intermediate calculations and display rounded results. Gain in dBi is converted using 10^(gain/10); power ratios use 10 log10; field-amplitude ratios use 20 log10. EIRP always retains the waveform convention of its input power. Matched-line examples intentionally separate dissipative attenuation from antenna-port mismatch.

Image production note

Image production: original conceptual cartoon sheet generated with the built-in image tool. Art direction: six text-free magazine-style scenes showing a station, coax loss, speech, Morse keying, burst data, and an elevated antenna; navy, teal, and orange on ivory. Quantitative charts were plotted directly from the stated models.