Dielectric Probe
Understanding the Dielectric Probe
Knowing the exact dielectric constant and loss tangent of a material is fundamental to RF design. The impedance of every microstrip line, the resonant frequency of every patch antenna, and the propagation delay of every stripline trace depend directly on the permittivity of the substrate beneath them. The dielectric probe provides the fastest, most convenient way to measure these properties across a wide frequency range without destroying the sample.
The probe itself is a precision-machined coaxial connector (typically 3.5 mm or 7 mm) whose center conductor and outer conductor terminate in a flat, polished plane. When this plane is pressed flush against a material, the fringing electric field lines that normally radiate into open air instead penetrate into the sample. The material's permittivity loads the probe aperture, changing both the magnitude and phase of the reflected signal. A VNA measures S11 at the probe reference plane, and dedicated software (such as Keysight N1500A or SPEAG DAK) applies an electromagnetic model to extract εr′ (real part, dielectric constant) and εr″ (imaginary part, loss) at each frequency.
εr* = εr′ − jεr″
Loss Tangent:
tanδ = εr″ / εr′
Probe Admittance Model:
Y(f, εr*) = G(f, εr*) + jB(f, εr*)
where G captures dielectric loss and B captures the capacitive loading from the material.
Example: Rogers RO4003C substrate measures εr′ = 3.38 and tanδ = 0.0027 at 10 GHz using a 7 mm dielectric probe, matching the datasheet specification of 3.38 ± 0.05.
Measurement Methods Comparison
| Method | Frequency Range | Sample Prep | Best For | Accuracy (εr′) |
|---|---|---|---|---|
| Dielectric Probe | 200 MHz to 50 GHz | Flat surface, no machining | Liquids, semi-solids, soft solids | ±5% |
| Cavity Resonance | Single frequency per mode | Precise sample machining | Low-loss ceramics, PTFE, quartz | ±0.5% |
| Transmission Line (Coaxial) | 100 MHz to 18 GHz | Toroid or cylinder | Absorbers, ferrites, composites | ±2% |
| Free-Space | 5 GHz to 110 GHz | Flat slab, ~30 cm × 30 cm | Large panels, radomes, building materials | ±1% |
| Split-Post Resonator | 1 to 30 GHz (discrete) | Thin flat sheet | PCB laminates, thin films | ±0.3% |
Frequently Asked Questions
How does a dielectric probe actually measure permittivity?
The probe is a short section of coaxial line with a flat, open end. When pressed against a material, fringing fields penetrate into the sample. The material's permittivity changes the boundary conditions at the aperture, altering the reflection coefficient (S11) measured by the VNA. Software uses an electromagnetic model to invert S11 into complex permittivity. Higher permittivity materials produce larger phase shifts, while lossy materials increase the real part of the admittance.
What are the limitations of the dielectric probe method?
The probe requires intimate, bubble-free contact with the sample surface. Any air gap lowers the apparent reading because air has a permittivity of 1.0. The sample must be thick enough that fringing fields do not reach the back surface (typically 5 to 10 mm minimum). The method works best for materials with permittivity above 2 and loss tangent above 0.05. Very low-loss materials like PTFE or quartz are better measured with cavity resonance techniques.
How do you calibrate a dielectric probe?
Standard calibration uses three references: open (air), short (shorting block against the probe face), and a known liquid (typically deionized water at a measured temperature, since water's permittivity is precisely characterized from 0 to 100 degrees C). Some systems add a fourth reference using methanol. The calibration corrects for the probe's own reflection, cable losses, and connector mismatch.