Transmission Lines

Contactless SIW

Q: How do via spacing and diameter prevent leakage in a contactless SIW?

The periodic via fence behaves as a leaky parallel-plate boundary, so the design rules keep the gaps between vias electrically small. The common criteria are via pitch p less than 2 times the via diameter d (which is the same as a diameter-to-pitch ratio d/p above 0.5), and pitch p less than 0.1 times the wavelength in the dielectric. Holding these limits keeps leakage and radiation suppression within acceptable bounds, typically better than 20 dB. For a 28 GHz design on Rogers RO3003 (Dk 3.0), the dielectric wavelength is roughly 6.2 mm, so a pitch near 0.5 mm with 0.3 mm vias satisfies both rules. Tightening the pitch reduces leakage but raises drill count and fabrication cost.

Q: What is the difference between contactless SIW and gap waveguide?

Both confine waves without metal-to-metal contact, but the mechanism differs. Classic SIW uses two solid metal planes joined by via fences inside a single dielectric substrate. A contactless or gap-waveguide implementation replaces one of those contacting interfaces with a periodic bed of pins or mushroom-shaped EBG unit cells separated from the opposing plate by an air gap smaller than a quarter wavelength. Inside the stopband of that pin structure, no parallel-plate mode can propagate across the gap, so the signal stays confined to the ridge or groove even though the two halves never touch. This removes the assembly screws, solder, and contact-resistance variability that limit ordinary split-block waveguide above 60 GHz.

Q: What insertion loss can a contactless SIW achieve at 60 GHz?

On a low-loss laminate such as RO3003 or a fused-silica carrier, ridge-type contactless SIW and gap-waveguide lines demonstrate roughly 0.05 to 0.15 dB per centimeter at 60 GHz, comparable to hollow metal waveguide and several times better than open microstrip on the same substrate. Because there is no soldered or screwed seam, the loss does not degrade with thermal cycling, and the absence of a contacting joint removes passive intermodulation sources. The main contributors that remain are dielectric loss tangent, copper surface roughness, and radiation through the via or pin gaps, which is why pitch control and a high d/p ratio matter.

/kon-takt-luhs ess-eye-double-yoo/

A class of substrate integrated waveguide in which the solid conducting sidewalls are replaced by periodic via fences, beds of pins, or mushroom-shaped EBG unit cells that confine the guided wave through an electromagnetic bandgap rather than continuous galvanic metal contact. Within the stopband of that periodic structure, no parallel-plate mode can leak across the boundary, so the two metal layers, or the two halves of a gap waveguide block, never need to physically touch. Eliminating the screwed or soldered seam removes contact-resistance loss, passive intermodulation, and assembly tolerance sensitivity, which is why contactless SIW has become a preferred packaging approach above 60 GHz where conventional split-block joints become unreliable. Measured ridge-type lines on RO3003 reach 0.05 to 0.15 dB/cm at 60 GHz, rivaling hollow waveguide.

Category: Transmission Lines

Typical Band: 30 to 110 GHz

Loss (60 GHz): 0.05 to 0.15 dB/cm

How Periodic Boundaries Replace Solid Walls

Conventional substrate integrated waveguide sandwiches a dielectric slab between two solid copper planes and stitches the sidewalls with a dense row of plated through-vias. That arrangement works beautifully on a single board, but it fails the moment a designer must join two separate pieces of metal: a screwed or soldered seam in the high-current region of a parallel-plate boundary introduces unpredictable contact resistance, leakage slots, and passive intermodulation. Contactless SIW sidesteps the problem by making the boundary itself reflect through an electromagnetic bandgap instead of through conduction. A periodic array of vias, pins, or mushroom patches presents a high surface impedance over a designed frequency band, so any wave that tries to escape across the gap sees an open circuit and is reflected back into the guiding region.

The most widely deployed form is the gap-waveguide family, where a bed of nails (a regular lattice of metal pins) faces a smooth metal lid separated by an air gap smaller than a quarter wavelength. Inside the pin stopband the parallel-plate mode cannot propagate laterally, so energy is trapped along a ridge (ridge gap waveguide) or in a groove (groove gap waveguide). The result is a low-loss channel that tolerates a non-contacting, even slightly misaligned, lid. This mechanical freedom is decisive at E-band and W-band, where a 10 micrometer burr on a flange face can cost a full dB.

Design proceeds in two stages. First the unit cell, the via or pin geometry, is dispersion-engineered so its stopband brackets the operating band with margin. Then the guiding feature is dimensioned for the target impedance and cutoff, exactly as in an ordinary rectangular waveguide. Full-wave eigenmode simulation of the periodic cell is essential because the achievable bandgap, typically 40 to 60 percent fractional, sets the usable bandwidth of the whole structure.

Leakage and Bandgap Design Rules

Via-fence leakage criteria (classic SIW wall):
p < 2d (equivalently d/p > 0.5) and p < 0.1 × λ_d

Wavelength in the substrate dielectric:
λ_d = c / (f × √ε_r)

Pin-bed stopband (gap waveguide, air gap, approximate):
f_low ≈ c / (4h) to f_high set by pin period a < λ₀/2

Where p = via pitch, d = via diameter, λ_d = wavelength in the dielectric, λ₀ = free-space wavelength, ε_r = substrate permittivity, h = pin height (in air), a = pin lattice period, c = speed of light. Example: 28 GHz on RO3003 (ε_r ≈ 3.0) gives λ_d ≈ 6.2 mm, so p ≈ 0.5 mm with d ≈ 0.3 mm satisfies both rules.

Contactless SIW Versus Conventional Guiding Structures

Structure	Wall mechanism	Loss at 60 GHz	Needs metal contact?	Best application
Contactless SIW / gap waveguide	EBG pin bed or via fence	0.05 to 0.15 dB/cm	No	mmWave packaging, split-block-free modules
Conventional SIW	Plated via fence, single board	0.1 to 0.3 dB/cm	No (monolithic)	Planar filters, antenna feeds
Hollow rectangular waveguide	Solid metal walls	0.02 to 0.1 dB/cm	Yes (flange joints)	High-power, lowest-loss links
Microstrip	Open, ground plane only	0.5 to 2 dB/cm	No	Low-cost integrated routing
Grounded coplanar waveguide	Side grounds + vias	0.3 to 1 dB/cm	No	Probe-friendly IC interconnect

Common Questions

Frequently Asked Questions

How do via spacing and diameter prevent leakage in a contactless SIW?

The via fence must look electrically continuous, so the standard rules keep gaps small: via pitch p < 2d (the same as a diameter-to-pitch ratio d/p > 0.5) and p < 0.1λ_d, which holds radiation suppression better than 20 dB. At 28 GHz on RO3003 (ε_r ≈ 3.0), the dielectric wavelength λ_d is about 6.2 mm, so a 0.5 mm pitch with 0.3 mm vias works. Tighter pitch lowers leakage but increases drill count and cost.

What is the difference between contactless SIW and gap waveguide?

Both confine waves without metal-to-metal contact. Classic SIW joins two solid planes with via fences in one substrate; a gap-waveguide implementation replaces a contacting interface with a periodic bed of pins or mushroom EBG cells separated by an air gap under λ/4. Inside the pin stopband no parallel-plate mode crosses the gap, so the two halves never touch, removing the screws, solder, and contact-resistance variability that limit split-block waveguide above 60 GHz.

What insertion loss can a contactless SIW achieve at 60 GHz?

On low-loss laminates like RO3003, ridge-type contactless SIW and gap-waveguide lines measure roughly 0.05 to 0.15 dB/cm at 60 GHz, near hollow waveguide and far better than open microstrip. With no soldered or screwed seam, loss is stable over thermal cycling and PIM sources are removed. Remaining contributors are dielectric loss tangent, copper surface roughness, and gap radiation, which is why pitch control and a high d/p ratio matter.

Related Terms

← Contactless Rotary Container →