RF Engineer Interview Questions and Strong Answers

What is the difference between return loss and VSWR?

Both express the same mismatch: return loss is the ratio of incident to reflected power in dB, where HIGHER return loss means less reflection and a better match — 10 dB return loss is about 10% power reflected (VSWR ≈ 1.92), 20 dB is about 1% (VSWR ≈ 1.22). VSWR states it as a standing-wave ratio of at least 1, where lower is better. The convention trap interviewers listen for: "high return loss" is good, "high VSWR" is bad — a candidate who says both "point to mismatch" has the sign of one of them backwards.

How do you sanity-check a link budget?

Received power = transmit power + antenna gains − path loss − implementation losses; the margin is received power minus receiver sensitivity. Sanity anchors: free-space path loss is about 80 dB at 100 m and 2.4 GHz and grows 6 dB per doubling of distance or frequency; sensitivity comes from −174 dBm/Hz + 10·log10(bandwidth) + noise figure + required SNR. Treat a 40 dB margin as a red flag to audit the assumptions — short, high-EIRP, narrowband links can legitimately carry huge margins, but more often a number that large hides an error like free-space path loss applied indoors. Real deployments add fade and implementation margin on top.

How would you diagnose a bad antenna result?

Verify the measurement before the antenna: calibration current, reference plane at the antenna feed, cable and fixture loss accounted for. Then check the antenna environment — ground plane size, enclosure detuning, feed routing — because on compact embedded antennas, a small ground plane or nearby plastic often shifts resonance more than the design errors you are hunting for. Only then compare measured return loss and pattern against the simulated or expected electrical behavior, and reconcile each discrepancy to a physical cause rather than re-tuning blindly.

Why can good S11 still result in poor range?

Good S11 only means the port is reasonably matched. It does not guarantee good radiation efficiency, pattern quality, polarization alignment, output power, receiver sensitivity, or link-budget margin.

How do you debug poor receiver sensitivity?

Validate calibration first, then check cable loss, insertion loss before the LNA, LNA gain, noise figure, filter loss, mismatch, desense, interference, and the measurement reference plane.

How does the Friis formula guide front-end design?

The cascaded noise figure is dominated by the first stage, with later stages divided down by the gain ahead of them. That tells you to put a low-noise, reasonably high-gain LNA first and to minimize any lossy components (cables, switches, filters) before it, because every dB of loss there is roughly a dB of noise figure.

What does the Smith chart help you reason about?

The Smith chart maps complex impedance and reflection coefficient onto one plane, so you can see mismatch, design matching networks, and reason about how series or shunt L and C move the impedance toward the center. It makes the relationship between impedance, VSWR, and return loss visual instead of purely algebraic.

Why is desensitization a system-level problem, not just an RF one?

Desense happens when an aggressor signal, such as a nearby transmitter, digital clock harmonic, or DC-DC converter, raises the noise floor in the victim receiver band. Diagnosing it requires looking beyond the RF path into board layout, shielding, grounding, and timing, so it is a cross-domain debugging exercise.

How does phase noise limit a receiver?

Two main ways. Reciprocal mixing: the local oscillator’s phase-noise skirt mixes a strong adjacent blocker onto the wanted channel, raising the in-channel noise floor — so when reciprocal mixing dominates, close-in phase noise (for example at 10 kHz to 1 MHz offsets, depending on channel spacing) sets blocking performance, though front-end compression, filtering, and ADC headroom can also be the limit. Constellation rotation: integrated phase noise contributes a floor to error vector magnitude — roughly the RMS phase error in radians for small angles — capping the highest usable modulation order even with perfect SNR. That is why phase-noise requirements scale with both the blocking spec and the modulation order, and why a synthesizer choice can silently decide receiver performance.

What is the difference between sensitivity and selectivity?

Sensitivity is the weakest signal the receiver can demodulate with no interference present: −174 dBm/Hz + 10·log10(bandwidth) + noise figure + required SNR. Selectivity is how well it demodulates a wanted signal while rejecting strong signals on adjacent or nearby channels — set mainly by filtering, linearity, and phase noise; noise figure still matters for the wanted signal’s post-filter SINR, but it is not what rejects the blocker. A receiver can have excellent sensitivity and still fail in the field because selectivity, not sensitivity, is what crowded spectrum actually tests.

What does EVM measure, and what degrades it?

Error vector magnitude (see /topics/evm-error-vector-magnitude) is the RMS magnitude of the error vector between received and ideal constellation points, normalized to the reference constellation and reported in percent or dB — a single figure of merit that rolls up every impairment in the chain. Typical contributors split by signature: phase noise smears points tangentially, compression pulls outer points inward, IQ imbalance produces an image and asymmetric spread, and additive noise spreads points uniformly. Reading the constellation shape before reaching for a spectrum analyzer is the fast way to localize which impairment dominates.