How to Read Transceiver “Lab Test Reports”

I do not start by circling the largest number in a transceiver report. I start by writing one sentence about the operating problem: a quiet VHF receive site, weak signals beside contest stations, a multi-transmitter station, ordinary HF listening, or clean SSB transmission. Then I read only the tests that reproduce the relevant failure mechanism.

Read the Test Recipe Before the Score

A calibrated generator and a 50-ohm load deliberately remove the antenna, propagation and site from the experiment. That is the value of the laboratory: one mechanism can be repeated without waiting for the band to cooperate. The result applies to that controlled reference plane and configuration.

Before comparing two rows—or two laboratories—record:

• test frequency, mode and receiver bandwidth, preferably equivalent noise bandwidth;
• interferer offset or two-tone spacing and whether it lies inside or outside early filtering;
• preamp, attenuator, preselector, roofing filter, AGC, dither and other signal-path settings;
• firmware, supply voltage, temperature and transmitter drive/ALC state;
• the RF reference plane, source/load impedance and fixture or combiner loss;
• the endpoint: a 3 dB noise rise, a 1 dB wanted-signal reduction, an IMD product reaching a reference, or another declared criterion; and
• the dB reference: dBm, dBc, dB below one tone, dB below PEP or dBFS.

If any of those changes, the number may change even though the radio has not. A 2 kHz DR3 result is not a cleaner version of a 20 kHz result; it exercises a different part of the receiver. A preamp-on MDS and a preamp-off blocking result do not describe one simultaneous configuration.

Start with the Receiver’s Own Noise

MDS Is an Endpoint in a Stated Bandwidth

Minimum discernible signal is not a universal “weakest audible station.” In the ARRL laboratory method, MDS is the input level that makes signal power equal receiver-noise power, observed as a 3 dB rise in combined signal-plus-noise, with stated CW filtering and settings. Another report may use a different mode, bandwidth or detection criterion.

At approximately 290 K, a useful first-order relationship is:

Here B_n is equivalent noise bandwidth, not necessarily the filter label on the front panel. That is why an MDS-to-noise-figure conversion becomes unreliable when the actual noise bandwidth or reference conditions are missing.

Noise Figure Describes SNR Degradation

Noise factor is the input SNR divided by the output SNR for defined source conditions; noise figure is that ratio in decibels. It tells us how much noise the receiving chain adds. A calibrated hot/cold or excess-noise source can measure it, but mismatch, source calibration, bandwidth and uncertainty still belong in the result.

A lower receiver NF matters when receiver-added noise is a meaningful part of the system noise. Loss ahead of the receiver, antenna temperature and external noise can change that balance. ITU-R P.372 shows that atmospheric, galactic and man-made noise vary strongly with frequency, place, time and antenna pattern, so “HF is always externally noise-limited” is too broad. Compare the noise delivered by the actual antenna system with the receiver noise in the same bandwidth.

Then Ask Which Strong-Signal Failure Is Being Measured

RMDR, DR3 and blocking are not three names for “dynamic range.” They isolate different ways a strong off-channel signal can hide a wanted one.

Reciprocal Mixing Tests Oscillator and Clock Noise

A strong signal at an offset from the tuned frequency mixes with local-oscillator or sampling-clock phase-noise sidebands and creates noise in the receive passband. An RMDR test raises the blocker until that added noise reaches a declared endpoint; the ARRL method uses a 3 dB noise-floor rise.

Read the offset first. Close-in RMDR is relevant to strong neighbours in a pileup or contest. A wide-offset result may exercise different filtering. Also verify that stimulus phase noise sits below the receiver response being measured; otherwise the generator limits the result.

DR3 Tests Third-Order Products from Two Signals

Two equal off-channel tones at f₁ and f₂ can produce third-order responses at 2f₁−f₂ and 2f₂−f₁. Two-tone third-order dynamic range states how far the input tones can rise above a stated floor or wanted-signal reference before the product reaches the test endpoint.

Tone spacing matters because it determines which filters and stages see the large signals. The receiver’s noise or reciprocal mixing can mask an intermodulation product, making a result noise-limited. Compare DR3 only when spacing, bandwidth, reference level, settings and method match.

Blocking Tests Desensitisation from One Strong Signal

Blocking applies one strong unwanted signal and watches a specified change in the wanted response. The endpoint may be a 1 dB reduction in wanted output, as in the ARRL blocking-gain-compression procedure, or another explicitly defined degradation. Noise rise can become the limiting event before gain compression, so a good report says what actually stopped the test.

In a direct-sampling receiver, ADC clipping or another architecture-specific overload can set the limit. That does not make the test invalid; it means a classical extrapolated IP3 model may no longer describe the first failure. Measured blocking, reciprocal mixing, intermodulation and overload behaviour are more useful than arguing from an architecture label.

Use IP3 as a Model, Not a Promise

Third-order intercept is an extrapolated crossing point derived from assumed slopes in a nonlinear model. It is useful when the measured region follows those assumptions. It is not normally an input level at which the receiver can actually operate, and it does not replace the underlying tone levels, products, bandwidth and spacing.

For radios with switched gain, preselection, AGC transitions or ADC limits, one IP3 number can conceal several regimes. Keep the measured DR3 and overload results beside the derived intercept.

Read the Transmitter Half with Equal Care

Receiver tables tell you what the operator can tolerate. Transmitter plots tell you what everyone else may have to tolerate. Read them at the stated band, power, supply voltage, modulation, drive, ALC state and duty cycle.

Two-Tone IMD Needs Its Reference

An equal two-tone SSB test exposes odd-order intermodulation products. The audio tones, microphone gain, compression, RF power and ALC behaviour all influence the result. So does the reference convention: each equal tone is 6 dB below the two-tone peak-envelope-power reference, which means a product quoted “below either tone” differs by 6 dB from the same product quoted dB below PEP.

Compare like with like. Look at third-, fifth- and higher-order products and the shape of the shoulders, not only the best single component. Confirm that the coupler, attenuator and analyser were sufficiently linear and that resolution bandwidth and detector settings are declared.

Occupied Bandwidth Is Not the Whole Emissions Test

ITU terminology separates several ideas:

• Occupied bandwidth contains a specified fraction of total mean emission power; the standard 99% case leaves 0.5% below and 0.5% above its limits.
• Necessary bandwidth is the bandwidth required to convey the information at the intended quality under specified conditions; it is not simply read from one spectrum trace.
• Out-of-band emissions immediately outside the necessary bandwidth arise from modulation.
• Spurious emissions include harmonics, parasitic products, intermodulation and conversion products in the spurious domain.

A radio can have an acceptable 99% occupied bandwidth and still show objectionable close-in shoulders or a discrete harmonic. A useful report therefore shows spectrum at more than one span and states measurement bandwidth, detector, averaging, reference power, fundamental rejection, load and uncertainty.

Composite or Sideband Noise Deserves Its Own Plot

Transmitted noise at offsets from the carrier can affect stations that are not inside the nominal audio passband. Read offset, measurement bandwidth and dBc/Hz or integrated reference carefully. Do not merge this result with two-tone IMD: one describes noise-like skirts, the other deterministic nonlinear products under the chosen stimulus.

Translate the Bench Result into Your Station

The laboratory did not forget your antenna; it removed it on purpose. Put the station back into the analysis after the radio mechanisms are understood.

• Quiet receiving system: MDS, NF and loss before the receiver may matter strongly.
• Crowded band or nearby transmitter: close-in RMDR, narrow-spaced DR3, blocking and front-end filtering deserve priority.
• Multi-radio site: also examine antenna isolation, band-pass filtering, transmitter noise and timing; a single receiver number is not a site plan.
• Typical antenna-fed HF station: compare receiver noise with delivered external noise before paying for a small sensitivity difference.
• Responsible transmitting: inspect IMD, noise shoulders, occupied bandwidth, out-of-band behaviour and spurious emissions at realistic power and drive.

Feedline loss, preamplifiers, filters, antenna pattern, common-mode paths, local noise, signal statistics and AGC settings can all change what reaches the radio. None of that invalidates the bench test. It defines whether the tested limit is likely to become your limit.

A Practical Reading Method

For every candidate radio, make a four-column note: metric, conditions, failure mechanism, my operating case. Then work through the report:

• reject comparisons that lack compatible settings, spacing, bandwidth or endpoints;
• mark whether each result is measured directly, calculated or extrapolated;
• identify what limited the measurement—radio, generator, fixture, analyser noise or overload;
• look across bands and gain settings instead of selecting the best row;
• read transmitter plots as seriously as receiver tables; and
• prefer meaningful margin over the station’s expected signals and noise, not a league-table rank.

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