An RF measurement without an uncertainty is an opinion. An opinion with two decimal places is no better than an opinion with one. The number that matters is how much you can defend.
When an RTSA reports a signal at -47.3 dBm, what does that number actually mean?
Absolute amplitude accuracy is how close that -47.3 dBm reading is to the true power at the antenna connector, measured against a calibrated reference. Typical RTSAs achieve absolute accuracy of plus or minus 1 to 2 dB across their full frequency range, narrower (plus or minus 0.5 dB) within a defined "calibrated range" near room temperature.
Relative amplitude accuracy is how close the difference between two readings is to the true difference. If you measure two tones and report a 6.0 dB difference, the relative accuracy says how close that 6.0 dB matches reality. Relative accuracy is often much better than absolute, often plus or minus 0.1 to 0.3 dB, because most amplitude errors are common to both readings and cancel.
Knowing which kind of accuracy applies to your measurement matters. Compliance work demands absolute accuracy because the regulator cares about absolute emission levels. Comparative work only needs relative accuracy.
Half a dozen contributors stack up to produce the published accuracy spec: reference power source uncertainty (calibrator tolerance, 0.1 to 0.5 dB), frequency response variation (gain through the analyzer's RF chain varies with frequency), temperature drift (front-end components shift gain as they warm or cool), attenuator step accuracy (when AGC switches in or out attenuators), detector linearity (FFT magnitude detection chain), and window function correction (ENBW and equivalent power normalization).
A practical SPECTRAN V6 PLUS deployment combines factory calibration with optional field calibration against a user-supplied reference. The factory calibration covers the full frequency range; the field calibration tightens accuracy at a specific frequency or temperature to within 0.1 dB.
When the RTSA tunes to 3.5000 GHz, how close is the actual measured center to that number?
Frequency accuracy is set almost entirely by the reference oscillator. A free-running crystal oscillator drifts on the order of 10 to 100 parts per million (ppm) over temperature and time. At 3.5 GHz, 10 ppm is 35 kHz of error. A temperature-compensated crystal (TCXO) tightens this to 1 to 5 ppm. An oven-controlled crystal (OCXO) reaches 0.01 to 0.1 ppm, well under 1 kHz at 3.5 GHz.
For demanding work, an external 10 MHz reference is locked to the analyzer. A rubidium standard gives 0.001 ppm (a few hertz at 3.5 GHz). A GPS-disciplined oscillator (GPSDO) gives similar accuracy, with the long-term stability of GPS satellites.
The SPECTRAN V6 PLUS accepts a 10 MHz external reference input, so users requiring high-precision frequency measurements can connect a rubidium standard or GPSDO. With a GPSDO locked, frequency accuracy improves from the internal TCXO's few ppm to fractional ppb (parts per billion). The same 10 MHz reference output can be daisy-chained to other equipment in the test bench, ensuring everything sees the same time base. For lab calibration, the reference oscillator is the foundation. Every other measurement traces back to it.
A real oscillator does not produce a single perfect tone. Its output spectrum has a Gaussian-shaped phase noise skirt around the nominal frequency. The skirt level is published as dBc/Hz at specific offset frequencies from the carrier. Why does phase noise matter? Three concrete consequences.
Reciprocal mixing. Phase noise on the analyzer's local oscillator mixes with strong signals in the band, producing artifacts that look like real signals. A strong tone with the LO's phase noise smeared 100 Hz away appears as a smeared, lower-amplitude version of itself near the original. This limits the dynamic range for finding weak signals close to strong ones.
EVM degradation. In modulation analysis (Chapter 6), LO phase noise rotates the constellation by a small random amount per symbol. The constellation clusters smear in arcs around the origin, and EVM degrades. For 5G NR 256-QAM at 28 GHz, the LO phase noise budget is around -110 dBc/Hz at 1 MHz offset.
Spectrum analyzer dynamic range. Trying to measure a weak signal 100 kHz away from a strong carrier? The carrier's phase noise spreads to the weak signal's frequency. The weak signal's apparent amplitude is lower-bounded by the carrier's phase noise level at that offset. Cleaner LO equals better dynamic range close to carriers.
Dynamic range is the difference between the strongest and weakest signals that an analyzer can simultaneously distinguish. It has three sub-specifications.
Noise floor. The bottom of dynamic range. Below this, signals are buried in thermal and quantization noise.
For a 1 dB system NF and 100 kHz RBW, the noise floor is about -123 dBm. Drop the RBW to 1 kHz and the noise floor drops to -143 dBm. Long FFTs (small RBW) give better sensitivity, at the cost of slower measurements.
Spurious-free dynamic range (SFDR). The difference between the strongest signal and the strongest spur generated by the analyzer itself. Premium RTSAs achieve 80 to 100 dB SFDR. SFDR is what matters when you have one strong signal and you are looking for weak signals nearby.
Third-order intercept point (IIP3). A measure of resistance to intermodulation distortion when two strong signals are present. IIP3 is the (extrapolated) input power at which third-order intermod products would equal the input signals in amplitude:
A typical RTSA front end has IIP3 of +10 to +20 dBm. A 6 dB increase in input causes an 18 dB increase in IM3 products, halving the dynamic range.
Every RF measurement has a number, an uncertainty, and a confidence interval. A measurement of "-47.3 dBm" means nothing without "+/-0.5 dB at 95 percent confidence."
For an absolute power measurement on a SPECTRAN V6 PLUS at 3 GHz:
| Source | Value | Distribution | Standard uncertainty |
|---|---|---|---|
| Reference calibrator | ±0.2 dB | Normal (k=2) | 0.10 dB |
| Frequency response | ±0.5 dB | Rectangular | 0.29 dB |
| Temperature drift | ±0.3 dB | Rectangular | 0.17 dB |
| Cable / connector | ±0.1 dB | Rectangular | 0.06 dB |
| Window correction | ±0.05 dB | Rectangular | 0.03 dB |
Combined standard uncertainty (sum in quadrature):
Expanded uncertainty (k=2, 95 percent confidence):
So the measurement is reported as -47.3 dBm +/- 0.72 dB at 95 percent confidence. This is the level of rigor that compliance bodies, defense laboratories, and serious R&D groups demand.
A SPECTRAN V6 PLUS arrives from the factory with a multipoint calibration covering the full frequency range and supported reference standards. Periodic recalibration (annually for general work, more often for compliance use) maintains accuracy.
Self-calibration. The analyzer can run self-tests on power-up: internal noise diodes verify the noise floor, internal frequency references verify the LO chain, and digital correction tables remap any drift.
User calibration. The user can connect a known reference signal and tune the analyzer's correction tables for their specific use case. RTSA Suite PRO supports this through a guided calibration wizard.
Factory recalibration. For traceable accuracy, periodic factory recalibration through Aaronia or an authorized lab provides full multipoint calibration with certificate. The certificate documents the accuracy at every calibrated point and references the calibration standards used (often NIST-traceable in the US, PTB-traceable in Germany).
RTSA Suite PRO ships with image rejection calibration, frequency response calibration, and amplitude calibration wizards. The tools accept user-supplied references and apply corrections that persist across measurement sessions. For users who need traceable accuracy, the wizards produce calibration reports that document the procedure and the resulting uncertainties. Factory calibration sets the baseline, user calibration tightens the spec for specific deployments, and software automates both processes so engineers can focus on the measurement, not the metrology.
The Chapter 9 questions are now an interactive quiz. Pick an answer for each, get instant scoring, and see why each answer is right. Your progress is saved on this device.
Take the interactive quiz →Chapter 10 covers RTSA form factors, packaging, and environmental engineering. Benchtop, rack-mount, OEM-embedded, portable, and ruggedized RTSAs each fit different deployment patterns. We close with a snapshot of the SPECTRAN V6 PLUS family at a glance: ECO 100XA, PLUS 250XA, PLUS 500XA, and PLUS 2000XA, with their target applications and pricing tiers.