All insights
RF & Microwave3 min read

Spectrum analyzer calibration: beyond the frequency reference

Checking the 10 MHz reference is only the start. Amplitude accuracy, frequency response, noise floor and phase noise all determine whether your analyzer tells the truth.

This quarter we commissioned a dedicated RF and microwave bench, extending our capability to 26.5 GHz. The instrument we expect to see most on it is the spectrum analyzer, used by telecom installers, broadcasters, EMC engineers and R&D teams across the region.

Spectrum analyzers are complex, and a meaningful calibration covers much more than the frequency reference. Here is what a full performance verification involves.

Frequency reference accuracy

The internal reference oscillator, usually 10 MHz, sets the frequency accuracy of every measurement. We compare it against our GNSS-disciplined standard. For analyzers with an oven-controlled oscillator (OCXO), ageing can be specified at around 1 × 10⁻⁷ per year or better; a standard temperature-compensated oscillator may be an order of magnitude looser.

Absolute amplitude accuracy

We apply a signal of precisely known power, measured with a calibrated power sensor, and compare the analyzer’s displayed level. This is typically done at a reference frequency, such as 50 MHz, and is the anchor for all other amplitude checks.

Frequency response (flatness)

Displayed amplitude should be independent of frequency. We sweep a leveled source across the analyzer’s range, using a power splitter and power sensor to know the actual level at the input, and measure deviation. Microwave analyzers often specify flatness in bands, with tighter limits at lower frequencies.

Reference level and input attenuator

Changing the reference level or the input attenuator should not change the measured amplitude of a signal. We step through attenuator settings with a known signal and record the switching error. An attenuator with a worn or contaminated contact can produce errors of a dB or more.

Resolution bandwidth

We check the accuracy of each resolution bandwidth (RBW) filter’s 3 dB width and the switching uncertainty when changing between filters. RBW errors directly affect noise and channel-power measurements.

Displayed average noise level (DANL)

With the input terminated, we measure the analyzer’s noise floor, normalised to 1 Hz bandwidth. DANL determines the smallest signal the analyzer can see. A degraded noise floor often points to a damaged front-end mixer, frequently caused by excessive input power.

Residual and spurious responses

Analyzers generate small internal signals, called residuals, and can create false responses from strong input signals. We check that these are below their specified limits, because a spur mistaken for a real signal can send an interference investigation in the wrong direction.

Phase noise

For analyzers used on oscillators, radar or digital modulation, the local oscillator’s phase noise matters. We measure it at standard offsets, commonly 10 kHz and 100 kHz, using a low-phase-noise source.

Protecting your analyzer

The most common cause of failure we see is a damaged input mixer. Always check the maximum safe input level printed beside the RF input, use an external attenuator when measuring transmitters, and remember that DC on the input can be just as damaging as RF power.

Start a calibration

Your next calibration is due sooner than you think.

Send us your instrument list and we will come back within one working day with a fixed quote, turnaround date and collection slot.