What is the difference between a noise figure analyzer and a spectrum analyzer with noise figure option?

What is the difference between a dedicated noise figure analyzer and a spectrum analyzer with a noise figure measurement option? Both instruments use the Y-factor method with a calibrated noise source to measure noise figure, but they differ in architecture, accuracy, and speed: (1) Dedicated noise figure analyzer: examples: Keysight N8975A (26.5 GHz), legacy Agilent 8970B (1.6 GHz). Architecture: purpose-built radiometer receiver optimized for measuring very small noise power differences. Uses a high-stability internal LO and mixer, optimized for noise measurements. Accuracy: ±0.05-0.15 dB (best-in-class). The receiver is designed specifically for measuring the ratio of hot and cold noise powers with minimal internal noise contribution. Speed: fast (< 1 second per frequency point), suitable for production testing. Second-stage correction: automatic and precise (the analyzer calibration removes its own noise contribution). Frequency range: fixed (typically up to 26.5 GHz). (2) Spectrum analyzer with NF option: examples: Keysight PXA/MXA with N9069A noise figure option, R&S FSW with K30 noise figure option. Architecture: uses the existing spectrum analyzer receiver (wideband IF, ADC, FFT). The NF measurement personality automates the Y-factor procedure: it switches the noise source on/off and calculates Y, NF, and gain. Accuracy: ±0.15-0.40 dB (slightly worse than a dedicated analyzer). The SA receiver is designed for measuring signal power, not optimized for noise power ratios. The DANL and NF of the SA itself limit the measurable DUT NF. Speed: moderate (1-5 seconds per frequency point, depends on averaging). Second-stage correction: software-based, using a calibration sweep without the DUT. Frequency range: same as the SA (up to 50+ GHz for high-end models). (3) When to use each: dedicated NF analyzer: production testing of LNAs (highest accuracy and speed), device characterization where ±0.1 dB NF accuracy is critical, and environments where a dedicated instrument is justified by volume. SA with NF option: R&D labs where the SA is already available (no additional instrument cost, just a software license), occasional NF measurements (not the primary use case), and measurements above 26.5 GHz (where dedicated NF analyzers are not available). (4) PNA-X cold source method: the Keysight PNA-X VNA can measure NF using the cold source method (no noise source needed). It measures S-parameters and noise power in a single sweep. Accuracy: ±0.1-0.2 dB. Advantage: measures S-parameters and NF simultaneously. This is becoming the preferred method in many R&D labs.