What errors does calibration correct?

VNA calibration corrects six systematic error terms per port (12-term for full 2-port): Directivity error (e_df): finite isolation between the reference and test channels in the directional coupler. Without correction, a device with 40 dB return loss appears to have only 30 dB if the coupler directivity is 30 dB. Source match (e_sf): impedance mismatch at the VNA test port. Creates multiple reflections with the DUT that distort measurements. Load match (e_lf): imperfect termination at port 2, causing reflections back through the DUT. Reflection tracking (e_rf): frequency-dependent gain and phase variations in the reflection measurement path. Transmission tracking (e_tf): frequency-dependent gain and phase variations in the transmission measurement path. Isolation (e_xf): crosstalk leakage between ports. After calibration, these errors are mathematically removed using the measured cal standard data, improving measurement accuracy from approximately plus or minus 1 dB to plus or minus 0.01 dB.

What is the difference between SOLT and TRL?

SOLT (Short-Open-Load-Thru) uses four precisely characterized standards. The Open has a known fringing capacitance (specified as polynomial coefficients in the cal kit definition). The Short has a known offset inductance. The Load is a precision 50-ohm termination (VSWR less than 1.01). The Thru is a zero-length connection or known-length adapter. SOLT requires accurate knowledge of each standard's electrical behavior, which becomes difficult above 50 GHz where manufacturing tolerances are significant. TRL (Thru-Reflect-Line) uses two transmission standards (a thru connection and a line of known impedance but unknown length) and one reflection standard (an unknown short or open, only needs to be highly reflective, actual value unimportant). TRL is self-calibrating: it derives the error model from the measurements themselves, requiring only the line's characteristic impedance and approximate length. TRL is more accurate than SOLT above 20 GHz and is the standard method for on-wafer measurements and waveguide calibration.

How often should calibration be performed?

Calibration validity depends on environmental stability and measurement accuracy requirements. General lab: recalibrate every 4 to 8 hours, or when temperature changes by more than 2 degrees Celsius. The VNA's internal temperature compensation helps, but the cables and adapters are the main drift sources. Production test: recalibrate at the start of each shift. Many production VNAs support ECal (electronic calibration) modules that automate the process to under 30 seconds. High-accuracy metrology: recalibrate before each measurement session and verify with a known check standard. After any cable change: always recalibrate. The cables are part of the calibrated measurement system. Changing a cable invalidates the calibration. After connector maintenance: if a connector is cleaned, re-gauged, or replaced, recalibrate. Signs of bad calibration: poor port-to-port isolation, ripple in thru measurement, non-physical results (gain in passive devices).

Measurement

Calibration Kit

/kal-ih-bray-shun kit/ — Cal Kit

Precision impedance standards for VNA error correction. SOLT: Open (C_fringe polynomial), Short (L_offset), Load (50Ω, VSWR<1.01), Thru (zero-length). TRL: self-calibrating, better >20 GHz. Corrects 12 error terms: directivity, source/load match, tracking. Accuracy: ±0.01 dB IL, ±0.1° phase. Connector-specific. Recalibrate: every 4-8 hrs or after cable change.

SOLT: 4 standards

TRL: Self-cal

Accuracy: ±0.01 dB

Understanding Calibration Kits

Without calibration, a VNA's raw measurements are corrupted by systematic errors from the instrument's internal couplers, cables, adapters, and connectors. A precision cal kit provides known impedance standards that allow the VNA to mathematically model and remove these errors, transforming a mediocre measurement into one accurate to hundredths of a dB.

The cal kit is the single most important investment for any RF measurement lab. A $50,000 VNA with a worn-out cal kit produces worse data than a $5,000 VNA with a pristine cal kit. Treat cal standards with extreme care: clean connectors before every use, store in protective cases, never drop, and verify annually against traceable standards.

Error Model

12-term error model:
S_11m = e_df + (S_11a×e_rf) /
(1 − e_sf×S_11a)
Where: e_df=directivity, e_rf=tracking
e_sf=source match

Cal standard models:
Open: Γ = −1 × e^−j2πfτ × 1/(1+j2πfRC)
Short: Γ = +1 × e^−j2πfτ × (1+j2πfL/Z₀)
Load: Γ = 0 (ideal), <0.01 (real)

Uncal vs cal accuracy:
Uncal: ±1-3 dB IL, ±5-15° phase
After SOLT: ±0.01 dB, ±0.1°

Calibration Method Comparison

Method	Standards	Accuracy	Freq Range	Application
SOLT	O, S, L, T	Good	DC-50 GHz	General lab
TRL	T, R, L	Best	1-110 GHz	On-wafer, WG
ECal	Electronic	Good	DC-67 GHz	Production
SOLR	O, S, L, R	Good	DC-50 GHz	Unknown thru
Response (S/O)	S or O	Poor	Any	Quick check

Common Questions

Frequently Asked Questions

Error terms?

6 per port (12 total): directivity (coupler leakage), source match (port impedance), load match (far port), reflection tracking (gain/phase vs f), transmission tracking, isolation (crosstalk). After cal: ±1 dB → ±0.01 dB accuracy improvement. Mathematical removal using measured standards.

SOLT vs TRL?

SOLT: 4 known standards with polynomial models. Good to 50 GHz. Needs accurate standard characterization. TRL: self-calibrating from 2 TL sections + unknown reflect. Only needs line Z₀. Best above 20 GHz. Standard for on-wafer/waveguide. TRL = gold standard for accuracy.

How often?

Lab: every 4-8 hours or 2°C temp change. Production: every shift (ECal: 30 seconds). After ANY cable change. After connector cleaning. Signs of bad cal: ripple in thru, gain in passive devices, poor isolation. Verify with check standard. Annual traceable verification recommended.

Related Terms

← Calibration Cal Plane →

RF Measurement

Request a Quote

Need cal kits, VNA accessories, or measurement services? Contact our team.

Get in Touch