RF spectrum analyzers

01

Set resolution bandwidth and selectivity

Agilent's Spectrum Analysis Basics (Application Note 150) describes resolution filters by their 3 dB bandwidth and bandwidth selectivity, the ratio of the 60 dB to the 3 dB bandwidth, which sets how well unequal signals close together can be resolved. Its analog four-pole synchronously tuned filters have selectivity of about 12.7:1; digital resolution filters reach about 4:1, so they resolve more closely spaced signals; in its ESA-E example the narrowest bandwidths, 1 Hz to 300 Hz, are realised digitally. [1]

The note warns that phase noise from the analyzer's local oscillators, specified in dBc normalised to a 1 Hz bandwidth, can hide a small signal next to a large one even when the resolution bandwidth is narrow enough, and that each tenfold reduction in resolution bandwidth lowers the displayed phase noise level by 10 dB. [1]

A current bench analyzer shows these terms on a datasheet front page: SIGLENT lists its SSA3000X Plus series from 9 kHz to 1.5, 2.1, 3.2 or 7.5 GHz by model, with a 1 Hz minimum resolution bandwidth, phase noise below −98 dBc/Hz at 10 kHz offset, displayed average noise level of −165 dBm/Hz (typical), amplitude accuracy better than 0.7 dB, a standard preamplifier and tracking generator, and options for RF power measurement, vector signal modulation analysis, reflection measurement and EMI pre-compliance testing. [3]

Give the supplier the frequency range, signal levels, the closest spacing and level difference between signals to be resolved, required noise floor or sensitivity, measurement types (phase noise, harmonics, channel power, EMI pre-scan), sweep-time needs, bench or handheld use and interfaces.

02

Balance noise floor against sweep time

AN 150 shows that sweep time for analog filters follows ST = k × Span / RBW², with k of 2 to 3 for near-Gaussian filters, so each step in a 1-3-10 resolution sequence changes sweep time by a factor of about 10; analyzers couple sweep time to span and RBW automatically and flag uncalibrated settings. The displayed noise level changes as 10 log(BW2/BW1), falling 10 dB for a tenfold narrower RBW, while the video filter smooths noise without changing the average noise level. [1]

The note derives noise figure from the displayed noise: the thermal (kTB) noise at room temperature in a 50 Ω termination is −174 dBm in 1 Hz, so a reading of −110 dBm in a 10 kHz RBW gives NF = −110 − 10 log(10,000) + 174 = 24 dB. [1]

Editorial calculation: applying AN 150's kTB relation to SIGLENT's typical −165 dBm/Hz gives a noise figure of about −165 + 174 = 9 dB, and the same level read at a 10 kHz RBW would be about −165 + 10 log(10,000) = −125 dBm. SIGLENT's page does not state the preamplifier and attenuator settings behind the typical figure, so ask for the DANL specification with and without preamplifier. [1] [3]

Editorial calculation: with k = 2.5, a 100 MHz span at 10 kHz RBW sweeps in about 2.5 × 10⁸ ÷ 10⁸ = 2.5 s; narrowing to 1 kHz RBW lowers the noise floor 10 dB but stretches the analog sweep to about 250 s. [1]

A USB real-time analyzer's datasheet shows the same terms split by band. Tektronix lists its RSA306 at 9 kHz to 6.2 GHz with a measurement range of +20 to −160 dBm, displayed average noise of −160 dBm/Hz (−163 typical) from 5 MHz to 1 GHz rising to −150 dBm/Hz (−153 typical) from 4 to 6.2 GHz, phase noise on a 1 GHz carrier of −84 dBc/Hz at 10 kHz offset (−87 typical), warranted amplitude accuracy of ±2.0 dB from 9 kHz to 3 GHz at 18–28 °C (±1.25 dB typical), 100% probability of intercept for signals of 100 µs in spans up to 40 MHz, −10 to +55 °C operation, MIL-PRF-28800 Class 2 environmental ratings and a weight of 0.59 kg. [4] Editorial calculation: by the kTB relation above, −163 dBm/Hz typical corresponds to a noise figure of about 11 dB, 2 dB worse than the 9 dB worked out for the −165 dBm/Hz bench analyzer; its warranted −84 dBc/Hz at 10 kHz compares with < −98 dBc/Hz on that bench model, so the portable unit trades noise and phase-noise margin for size, real-time capture and ruggedness. [4] [3] [1]

03

Compare what suppliers offer

The suppliers linked to this page list spectrum analyzers; their cards link to the pages reviewed. SUIN, for example, lists spectrum analyzers alongside signal generators and power analyzers. [2]

Ask any supplier for frequency range, RBW range and filter type, displayed average noise level with and without preamplifier, phase noise at stated offsets, amplitude accuracy, maximum input level, sweep or FFT speed, tracking generator and measurement options, and calibration.

04

Trace the analyzer blocks

This is a functional reference built from the sources cited; it is not one maker's specification.

05

Fill the specification fields with values for your signals

A spectrum analyzer quotation needs values for your signals. The table explains each field, gives published values where a reviewed source states them, and says what to obtain from the supplier.

06

Send the signal and measurement data together

Paste it into your enquiry below or into an email to a maker, then replace every bracket.

Signals: [frequency range; levels]
Resolution: [closest spacing; level difference]
Sensitivity: [noise floor needed]
Measurements: [phase noise; harmonics; channel power; EMI]
Speed: [sweep time limit]
Use: [bench or handheld; interfaces; calibration]

See manufacturer records Send an enquiry through CNFX