RF and microwave (to 75 GHz) calibration
Up to 75 GHz the quantities are power, attenuation, reflection, frequency and noise. Most of the uncertainty comes from connectors and mismatch, not from the instrument. This page covers coaxial and waveguide measurement from 10 MHz to 75 GHz.
Instruments and typical accuracy
| Instrument | Typical range | Typical accuracy | Where it is used |
|---|---|---|---|
| Thermistor and thermocouple power sensors | 10 MHz to 50 GHz, about -35 to +20 dBm | 1 to 3 % (about 0.05 to 0.13 dB) | Reference power measurement |
| Diode power sensors | 10 MHz to 75 GHz, -70 to +20 dBm | 2 to 5 % plus mismatch | General power measurement and low levels |
| Microcalorimeter (primary) | Coaxial and waveguide | Better than 0.5 % | National institutes: sets the efficiency of sensors |
| Vector network analyser (VNA) | 10 MHz to 70 GHz and more | Return loss and S-parameters, depends on calibration | Reflection, insertion loss, phase |
| Step attenuator and standard attenuator | DC to 50 GHz and waveguide | 0.01 to 0.05 dB per 10 dB | Attenuation and linearity references |
| Frequency counter and timebase | Up to 40 to 70 GHz direct | Timebase accuracy (10⁻⁹ to 10⁻¹²) | Frequency of signal generators |
| Signal generator and spectrum analyser | 9 kHz to 75 GHz | Level ±0.5 to ±2 dB, frequency 10⁻⁷ to 10⁻⁹ | Production and general test |
| Noise figure meter and noise source | 10 MHz to 50 GHz and above | ±0.1 to ±0.3 dB | Receiver and low-noise amplifier testing |
Everything above is a typical figure. At microwave the connector and adapters add as much uncertainty as the instrument. The accuracy figures are typical ranges to help you compare instruments. Always use the specification of the instrument in front of you.
Waveguide bands up to 75 GHz
| Waveguide | Broad wall a | Standard band (GHz) | TE10 cutoff (GHz) | Band letter (approx.) |
|---|---|---|---|---|
| WR-284 | 72.136 mm | 2.6 to 3.95 | 2.08 | S |
| WR-187 | 47.549 mm | 3.95 to 5.85 | 3.15 | C |
| WR-137 | 34.849 mm | 5.85 to 8.2 | 4.30 | C |
| WR-112 | 28.499 mm | 7.05 to 10 | 5.26 | X |
| WR-90 | 22.86 mm | 8.2 to 12.4 | 6.56 | X |
| WR-62 | 15.799 mm | 12.4 to 18 | 9.49 | Ku |
| WR-42 | 10.668 mm | 18 to 26.5 | 14.05 | K |
| WR-28 | 7.112 mm | 26.5 to 40 | 21.08 | Ka |
| WR-22 | 5.69 mm | 33 to 50 | 26.34 | Q |
| WR-19 | 4.775 mm | 40 to 60 | 31.39 | U |
| WR-15 | 3.759 mm | 50 to 75 | 39.88 | V |
Letter designations vary between standards (IEEE, EIA, NATO), so quote the WR number and the frequency range. Cutoff is c / 2a.
Coaxial connector families
| Connector | Highest single-mode frequency | Mates with |
|---|---|---|
| Type N | 18 GHz | Rugged, general lab and field |
| 7 mm (APC-7) | 18 GHz | Sexless, reference-grade metrology |
| 3.5 mm | 26.5 GHz | SMA compatible |
| 2.92 mm (K) | 40 GHz | 3.5 mm and SMA compatible |
| 2.4 mm | 50 GHz | Precision, mates with 1.85 mm (limited to 50 GHz) |
| 1.85 mm (V) | 67 GHz | Mates with 2.4 mm |
| 1.0 mm | 110 GHz | Smallest common coax |
Mating connectors of different families is only safe where the datasheet says so, and the lower-frequency one sets the limit.
Reference standards and traceability
- Power. A calorimeter or microcalorimeter compares RF heating with DC heating, so power traces back to DC voltage, current and resistance. The calibration factor of each sensor is given against the reference at a set of frequencies.
- S-parameters. The VNA is calibrated with a precision calibration kit (short, open, load, thru, or TRL lines in waveguide) whose definitions trace back to dimensional measurements.
- Attenuation. Reference attenuators are calibrated with a dual-channel IF substitution or a calibrated receiver, with traceability to ratio standards.
- Frequency. Compare the instrument with a rubidium, cesium or GPS-disciplined reference through the external timebase input.
- Documents. IEEE 287 for precision coaxial connectors, ISO/IEC 17025 for the lab, and national guides for the uncertainty of RF measurements.
How it is calibrated
Typical procedure
- Warm up the instruments and the sensor for the time in the manual, and let the room be stable to ±1 °C.
- Inspect and clean connectors with lint-free swabs and isopropyl alcohol, and check them with a gauge.
- Tighten connectors with a torque wrench: typically 0.9 N·m for 3.5, 2.92, 2.4 and 1.85 mm connectors, and less for SMA.
- Calibrate the VNA (SOLT, TRL or LRM) at the plane of the device with the calibration kit coefficients loaded.
- Zero and calibrate the power meter at the 0 dBm, 50 MHz reference, then measure the source at each frequency.
- Measure source and sensor match, calculate the mismatch uncertainty, and add the calibration factor uncertainty.
- Check with a verification device (a known attenuator or mismatch standard) before and after.
Where it goes wrong
- Dirty or damaged connectors. A bad connection can change return loss by 10 dB and cannot be calibrated out.
- Wrong torque. Over-tightening damages the pin, under-tightening changes the contact.
- Cable movement. Phase and loss change when flexible cables are moved after calibration.
- Mismatch ignored. At 1.3 VSWR on both ports the power error is about ±0.15 dB.
- Wrong adapters. A cheap adapter adds loss and reflection that were not in the calibration.
What goes in the uncertainty budget
- Power sensor calibration factor and its uncertainty.
- Mismatch between source and sensor (see the calculator).
- Connector repeatability and adapter effects.
- Linearity, noise and drift of the power meter.
- Cable flex, temperature and frequency accuracy.
Combine the contributions as in the uncertainty budget calculator, then decide pass or fail with a guard band as explained on the calculators page.
Calculators
Everything runs in your browser. Nothing you type is sent anywhere.