Field Guide · hardware

Also known as: noise source, noise diode, ENR source

A noise source is a calibrated broadband noise generator used to measure the noise figure of receivers, amplifiers, and other RF components.1 Instead of a clean carrier like a signal generator, it emits flat, noise-like power across a wide band, characterized by its excess noise ratio (ENR) — the amount by which its “on” noise exceeds the reference thermal noise of a resistor at room temperature.

Noise source OFF (cold) / ON (hot), ENR Device under test Receiver / power meter Y = P(hot) / P(cold)
Noise-figure measurement by the Y-factor method: switch the source between its calibrated hot and cold states and read the ratio of output powers; ENR and Y together give the device's noise figure.

How it works

At the heart of a noise source is a reverse-biased avalanche (Zener) noise diode that, when energized (typically by a 28 V supply), produces broadband shot noise well above the thermal floor. Switch the bias off and the source presents a matched resistor at ambient temperature — the “cold” state at roughly 290 K. The two states are the “hot” and “cold” references the measurement needs. The source’s ENR, supplied as a calibration table versus frequency, states in dB how far its hot-state noise temperature exceeds the 290 K reference.

The Y-factor method

Noise figure is most commonly measured by the Y-factor technique:

  • Connect the noise source to the device under test and its output to a receiver, noise- figure meter, or spectrum analyzer.
  • Record the output power with the source on (hot) and off (cold).
  • The ratio Y = P_hot / P_cold combines with the calibrated ENR to solve for the device’s added noise. A low-noise device barely changes Y as the input toggles; a noisy one swamps the difference — which is exactly what noise figure quantifies.

A two-stage “measure the analyzer alone, then the analyzer plus DUT” calibration removes the meter’s own noise contribution, so the result reflects the device under test rather than the instrument.

In practice

  • ENR grade matters. Low-ENR sources (~5–6 dB) suit sensitive, low-noise-figure devices; high-ENR sources (~15 dB) suit lossy or high-NF paths.
  • Match and cabling. Poor impedance match between states adds error; keep connections short and well-characterized.
  • Frequency limits. ENR is calibrated per frequency; use the table, not a single number, across a wide sweep.

Relevance to SDR

Noise figure sets a receiver’s ultimate sensitivity, and a noise source is the standard way to measure it — for instance to confirm that adding a low-noise amplifier actually improves the system NF rather than the attenuation of a long feedline dominating it. For the SDR scanner enthusiast this is bench-lab territory: quantifying an LNA or front-end so weak trunking signals clear the noise floor. GopherTrunk is a receive-only decoder that neither generates nor measures noise figure; a noise source is a specialized RF-lab aid used upstream to characterize the hardware that feeds the SDR, not part of the decode chain.

Sources

  1. Noise generator — Wikipedia, on calibrated noise sources, ENR, and their use in noise-figure measurement. ↩

See also