
What Is an RF Transmission Line on a PCB?
An RF transmission line is a conductor geometry on a printed circuit board that carries a signal at a controlled characteristic impedance, typically 50 Ω, while confining electromagnetic fields between the trace and one or more reference planes. The four structures used most often—microstrip, stripline, grounded coplanar waveguide (GCPW), and substrate-integrated waveguide (SIW)—each trade off loss, bandwidth, isolation, and manufacturability differently. Choosing the right one for a given frequency band, board stackup, and production volume is one of the earliest and most consequential decisions in any RF layout.
The Four Main RF Transmission Line Types

Microstrip
A microstrip is the simplest RF transmission line: a trace on the top or bottom layer, separated from a ground plane by one dielectric layer. Fields travel partly in the substrate and partly in air, so the effective dielectric constant is lower than the bulk Dk of the laminate. On RO4350B (Dk 3.48 at 10 GHz), a 50 Ω microstrip is roughly 18.6 mil (0.47 mm) wide over a 10 mil (0.25 mm) dielectric—dimensions you can verify with a 50 Ω trace width reference. Microstrip is easy to probe, easy to route surface-mount components onto, and the lowest cost to fabricate. Its main weaknesses are higher radiation loss above about 10 GHz and limited isolation from adjacent traces because the fields are partially unshielded.
Stripline
Stripline buries the signal trace between two ground planes, fully enclosing the fields in a homogeneous dielectric. This eliminates radiation loss and provides excellent isolation, making it the default choice for multilayer RF boards carrying sensitive receive paths or tightly spaced channels. The penalty is that the trace is inaccessible for probing, vias are needed to reach surface components, and the narrower trace width (roughly 30 % narrower than microstrip for the same impedance and Dk) demands tighter etch tolerances. Stripline also supports a true TEM mode, so effective Dk calculation is straightforward—it equals the bulk Dk of the fill material.
Grounded Coplanar Waveguide (GCPW)
GCPW adds coplanar ground pads on either side of the signal trace, stitched to the ground plane below with closely spaced vias. The result is a quasi-TEM line whose impedance is set by both the trace width and the gap to the coplanar ground. This gives the designer an extra degree of freedom: you can hit 50 Ω with a narrower trace than microstrip, which is valuable on dense mmWave layouts. GCPW also offers better isolation than open microstrip and a natural launch geometry for coaxial connectors. The via stitching pitch must stay below λ/20 at the highest frequency of interest to suppress parallel-plate modes; at 28 GHz on RO3003 (Dk 3.00), that means vias every 0.54 mm or closer. Proper RF via design is therefore critical in any GCPW layout.
Substrate-Integrated Waveguide (SIW)
SIW mimics a rectangular waveguide inside a PCB by using two rows of plated-through vias as the side walls, with the top and bottom copper acting as the broad walls. It supports a TE₁₀-like mode and is inherently high-pass: the cutoff frequency depends on the effective width between the via rows. On a 10 mil RO3003 substrate, a typical SIW for 77 GHz automotive radar has an effective width near 1.6 mm and a cutoff around 55 GHz. SIW offers the lowest loss per unit length of any planar line above 30 GHz and excellent isolation, but it is the widest structure on the board and requires transitions (typically tapered microstrip-to-SIW) at every port.
Comparison Table: RF Transmission Line Selection
| Parameter | Microstrip | Stripline | GCPW | SIW |
|---|---|---|---|---|
| Dominant mode | Quasi-TEM | TEM | Quasi-TEM | TE₁₀-like |
| Practical upper frequency | ~15 GHz (on FR-4); ~40 GHz (on low-loss laminates) | ~40 GHz | ~110 GHz | ~110 GHz+ |
| Dielectric loss at 10 GHz (typical) | 0.02–0.08 dB/cm | 0.02–0.06 dB/cm | 0.02–0.07 dB/cm | 0.01–0.03 dB/cm |
| Radiation loss | Moderate; rises with f² | Negligible | Low | Negligible |
| Isolation from neighbors | Low | High | Medium-high | Very high |
| Component access | Direct | Via required | Direct | Transition required |
| Relative board area | Narrow | Narrowest | Medium (gap + ground pads) | Wide |
| Fabrication complexity | Low | Medium | Medium-high (via stitching) | High (via fence tolerances) |
How to Choose: Frequency, Loss, and Layout Constraints
Selection starts with three questions: what is the operating frequency, how much insertion loss can the link budget tolerate, and how dense is the layout?
- Below 6 GHz — Microstrip on standard low-loss laminates (Dk 3.0–3.7, Df ≤ 0.004) handles nearly every application. Stripline is used only when isolation between channels is paramount, such as in multichannel receivers.
- 6–30 GHz — GCPW becomes attractive because it suppresses radiation and eases connector launches. Stripline remains the go-to for buried feed networks in mmWave PCB designs that use antenna arrays on outer layers.
- 30–77 GHz and above — SIW offers the lowest loss for longer runs (filters, feed networks). GCPW is still practical for short interconnects between ICs and transitions. Microstrip is limited to very short stubs and matching elements. At these frequencies, copper roughness effects add measurably to conductor loss regardless of line type, so smooth-foil laminates (e.g., Rz ≤ 1 µm rolled copper on low-profile treatments) are essential.
Loss Budget: Where the dBs Go
Total loss in any RF transmission line has three components: conductor loss (ohmic, proportional to √f), dielectric loss (proportional to f × Df), and radiation loss (proportional to f² for open structures). Below 6 GHz, conductor loss usually dominates. Above 30 GHz, dielectric and radiation losses catch up or overtake it—which is exactly why SIW, which virtually eliminates radiation, wins at millimeter-wave frequencies.
For a quick sanity check, Rogers RO4350B datasheets list Df = 0.0037 at 10 GHz. On a 50 Ω microstrip with 0.5 oz copper (17 µm), expect roughly 0.04 dB/cm of total loss at 10 GHz. At 28 GHz on the same material, that climbs to about 0.10–0.12 dB/cm. If your trace runs 5 cm, that is 0.5–0.6 dB—significant in a low-noise receive chain where every tenth of a dB raises the noise figure.
Transitions Between Line Types

Real boards rarely use a single RF transmission line type everywhere. A common architecture routes GCPW from the SMA launch to a filter, transitions to stripline through a buried via, then exits back to microstrip at the IC pad. Every transition introduces a small impedance discontinuity and a potential mismatch. Keep transitions short, match impedance through the via antipad geometry, and validate each one with a 3-D EM simulation. The basics of S-parameters are the language you need for quantifying those transition losses: S₂₁ shows insertion loss, S₁₁ shows return loss, and both should be checked across the full operating band.
Microstrip-to-SIW Taper
The standard microstrip-to-SIW transition is a linearly tapered microstrip section whose width gradually increases to match the SIW aperture. A well-optimized taper on RO3003 at 77 GHz achieves better than −15 dB return loss over a 5 GHz bandwidth. The taper length is typically λg/4 to λg/2 at the center frequency. Because this transition is a distributed structure, its performance is sensitive to etch tolerance and dielectric thickness variation—one more reason to specify tight fabrication tolerances for mmWave boards.
Cost and Production Considerations
Line-type choice affects PCB cost in ways that are not always obvious. Microstrip on a two-layer board is the cheapest option. Adding stripline requires at least four layers and a second lamination cycle. GCPW adds drilling density for via stitching, which increases drill time and tool wear. SIW demands even more vias—often hundreds per centimeter of waveguide—and tighter positional tolerances (±25 µm or better) to keep the cutoff frequency predictable.
For prototypes, the per-board cost difference between microstrip and SIW can be 3–5× due to setup charges and small-panel utilization. In volume production, that gap shrinks because drill time amortizes across thousands of panels. Use the tool below to explore how line-type complexity, layer count, and volume interact with overall PCB cost:
As an example, a four-layer GCPW board on RO4350B with 0.2 mm via pitch might show a 40 % cost premium over a two-layer microstrip board at prototype quantities of 10 pieces. At 1 000 pieces, that premium drops to roughly 15 %, because the dominant cost driver shifts from setup and drilling to raw laminate area. If your design can tolerate microstrip loss and isolation at the target frequency, staying with a simpler stackup saves real money—especially in early prototyping where multiple board spins are likely.
Design Rules of Thumb
- Start with microstrip unless loss or isolation force you to a more complex line. It is the easiest to fabricate, probe, and rework.
- Use GCPW for connector launches and mmWave ICs. The coplanar ground pads provide a natural ground-signal-ground interface that matches both coaxial connectors and wafer-level chip-scale packages.
- Reserve stripline for buried routing where you need shielding—e.g., LO distribution in a multi-channel receiver, or feed networks under an antenna array.
- Deploy SIW for low-loss, high-isolation paths above 30 GHz—filters, couplers, and long feed runs where distributed loss matters most.
- Budget 0.3–0.5 dB per transition between line types in your initial link budget, then tighten that number with EM simulation. Refer to IEEE publications on microwave transitions for peer-reviewed transition designs you can benchmark against.
- Validate with S-parameter measurements. Fabricate a set of TRL (thru-reflect-line) calibration standards on the same panel as your circuit to de-embed fixture effects.
Material Selection Interplay
The RF transmission line type and the laminate choice are coupled decisions. A microstrip on FR-4 (Dk ≈ 4.3, Df ≈ 0.020 at 1 GHz) is acceptable for Wi-Fi front-end modules at 2.4 GHz but unusable at 24 GHz. GCPW on RO4350B handles 5G FR2 bands (24.25–29.5 GHz) well. SIW at 77 GHz almost always calls for RO3003 or a comparable ultra-low-loss PTFE laminate (Df ≤ 0.001). Isola Astra MT77 is another option at mmWave, offering Df of 0.0017 at 10 GHz with a thermoset resin system that processes more like conventional FR-4 than PTFE.
Frequently Asked Questions
Which RF transmission line has the lowest loss?
SIW has the lowest loss per unit length above roughly 30 GHz because it eliminates radiation loss and confines fields entirely within the substrate. Below 10 GHz, stripline and microstrip on low-Df laminates are comparable, and the simpler geometry is usually the better choice.
Can I mix transmission line types on one board?
Yes, and most mmWave boards do. The key is to design each transition carefully—match impedances through the via or taper, simulate the transition in 3-D EM, and include TRL calibration structures so you can measure transition loss independently.
Is GCPW always better than microstrip?
Not always. GCPW requires via stitching that consumes board area and adds drilling cost. Below 6 GHz, microstrip with adequate ground-plane clearance provides comparable performance at lower complexity. GCPW’s advantages become decisive above 10 GHz and for connector-launch geometries.
What via pitch do I need for SIW?
The via-to-via pitch (center to center) should be ≤ 2× the via diameter to prevent leakage, and the pitch should be ≤ λg/5 at the highest operating frequency. At 77 GHz on RO3003, that typically means 0.4–0.5 mm pitch with 0.2 mm drills.
How do I simulate an RF transmission line transition?
Use a 3-D electromagnetic solver (HFSS, CST, or openEMS) to model the physical geometry including vias, antipads, and tapers. Extract S-parameters across the full band and import the resulting Touchstone file into your circuit simulator to verify the link budget end to end.
Ready for a quotation?
Send your requirements for a material, stackup and manufacturing review. Price and schedule are confirmed for your project.
