Microstrip vs Stripline | Loss, Radiation & Density

Microstrip vs Stripline: The Core Difference

Microstrip is an outer-layer trace sitting on top of a dielectric with a single ground plane beneath it; stripline is an inner-layer trace sandwiched between two ground planes with dielectric on both sides. That structural difference drives every trade-off in loss, radiation, isolation and routing density. If your design needs easy component access and tuning, microstrip is the default. If it needs shielding, low radiation, and tight channel-to-channel isolation, stripline wins.

How Each Topology Works

Microstrip

A microstrip line propagates a quasi-TEM mode: part of the electric field travels through the substrate (Dk typically 3–10 for common RF laminates) and part through the air above the trace (Dk ≈ 1). The effective dielectric constant (Dk_eff) is therefore always lower than the substrate Dk—roughly 2.6–2.8 for a 50 Ω line on RO4350B (Dk 3.48 at 10 GHz). Because the field is partially in air, microstrip radiates, especially at bends, discontinuities, and higher frequencies.

Stripline

A stripline trace is fully enclosed in a homogeneous dielectric between two continuous reference planes. The propagation mode is true TEM (or very close to it), and the effective Dk equals the bulk substrate Dk. The surrounding ground planes act as a Faraday cage, virtually eliminating radiation. The penalty is that all traces are on inner layers, making probing and component mounting impossible on the stripline layer itself.

Head-to-Head Comparison Table

For a broader look at how these topologies sit alongside other options, the RF laminate comparison hub covers material and structure choices side by side.

Loss Breakdown: Where Each Topology Wins

Conductor (Ohmic) Loss

For a given impedance (say 50 Ω) on the same substrate, a microstrip trace is wider than a stripline trace because the effective Dk is lower. A wider trace means lower current density at the conductor edges and therefore lower resistive loss per unit length. On RO4003C at 10 GHz, a 50 Ω microstrip on 10 mil (0.254 mm) substrate is roughly 22 mil (0.56 mm) wide, while a centered stripline between two 10 mil dielectric layers is roughly 10–12 mil (0.25–0.30 mm) wide. The narrower stripline trace carries a measurable conductor-loss penalty, typically 0.01–0.03 dB/cm more at 10 GHz.

Copper roughness amplifies this difference. Standard ED copper (Rz ≈ 6–10 µm) adds more loss on the narrower stripline trace. Switching to VLP or HVLP copper foil (Rz < 2 µm) narrows the gap significantly, which is why low-profile copper is almost mandatory for stripline above 10 GHz.

Dielectric Loss

In microstrip, roughly 50–70 % of the field energy is in the substrate and the rest in air. In stripline, 100 % is in the substrate. Dielectric loss scales with the dissipation factor (Df) and the fraction of field in the lossy medium. On a low-loss laminate like RO4003C (Df 0.0027 at 10 GHz), the dielectric-loss difference between microstrip and stripline is small—on the order of 0.005–0.01 dB/cm. On FR-4 (Df ≈ 0.020), the penalty for stripline becomes substantial and is a key reason FR-4 stripline is rarely used above 2–3 GHz.

Radiation Loss

Microstrip radiates at every discontinuity: bends, T-junctions, open stubs, and even straight runs at millimeter-wave frequencies. Radiation loss rises roughly with frequency squared. At 5 GHz on a well-designed board it is negligible; at 24 GHz it can account for 0.5–1 dB of unexpected loss in a filter or coupler. Stripline, enclosed by ground planes, has essentially zero radiation loss. This is the single biggest reason stripline is preferred for high-isolation RF distribution networks—such as corporate feed networks for phased-array antennas.

Net Verdict on Loss

Below about 6–8 GHz on a low-loss laminate, microstrip usually has lower total insertion loss because its wider trace and partial air propagation outweigh the radiation contribution. Above 10 GHz—and certainly at 24 GHz or 77 GHz—radiation and dispersion losses in microstrip grow fast, and stripline (with VLP copper) can match or beat it in total loss while delivering far better isolation. At millimeter-wave frequencies, grounded coplanar waveguide (GCPW) is also worth evaluating as a surface-layer alternative that offers partial shielding.

Radiation and Isolation

Radiation from microstrip is not just a loss mechanism—it is a coupling mechanism. Two parallel microstrip traces separated by three line-widths may still show −20 dB of near-end crosstalk at 10 GHz. The same spacing in stripline typically yields −35 dB or better, because the ground planes confine the fields. For multi-channel receivers, radar IF distribution, and any design where channel-to-channel isolation matters (better than −30 dB), stripline is the safer choice.

Where microstrip must be used for component access, engineers mitigate radiation with via fences along both sides of the trace, ground coplanar gaps, and absorber-loaded shields. These techniques help but add fabrication cost and board area—which partly offsets the simplicity advantage of microstrip.

Routing Density and Stackup Impact

Microstrip traces live on outer layers. A four-layer board has at most two microstrip routing layers (top and bottom), each referencing its adjacent ground plane. Stripline traces live on inner layers, so a six-layer board can have two or more stripline routing layers plus the surface layers for components. This makes stripline the natural fit for dense RF routing in boards with four or more layers.

The trade-off is lamination complexity. Each stripline layer requires precise dielectric-thickness control on both sides, which tightens prepreg selection and press-cycle parameters. In a hybrid Rogers/FR-4 stackup, the stripline core is typically the expensive low-loss laminate while outer prepreg layers may be standard FR-4—saving cost while keeping the critical RF path on the right material.

Copper Weight, Etch Compensation and Trace Geometry

Both microstrip and stripline impedance depend critically on the finished trace width, which in turn depends on copper weight and the etch compensation applied during fabrication. A 1 oz (35 µm) copper layer etches differently from a ½ oz (17.5 µm) layer: the thicker copper undercuts more, requiring a wider artwork trace to hit the same finished width. Getting this wrong shifts impedance by several ohms—enough to push return loss from −20 dB to −12 dB on a 50 Ω line.

Use the calculator below to see how copper weight affects etch compensation for your target trace width. Enter your copper weight in ounces and your desired finished trace width; the tool returns the artwork compensation you should apply.

For example, a 50 Ω microstrip on 6.6 mil RO4350B with 1 oz copper needs a finished width near 14.4 mil. The etch compensation for 1 oz copper is typically 0.5–1.0 mil per side, so the artwork trace should be drawn at roughly 15.4–16.4 mil. For a stripline on the same material at ½ oz copper, the compensation drops to about 0.3–0.5 mil per side—but the target width is also narrower (perhaps 8 mil finished), making the percentage error larger. Always confirm these numbers with your fabricator’s specific etch process and tolerance data.

Dispersion: Why It Matters at mmWave

Microstrip’s quasi-TEM mode becomes increasingly dispersive above roughly 20 GHz: the effective Dk rises with frequency because more field concentrates in the substrate. This means a wideband pulse or modulated signal sees different phase velocities across its bandwidth, distorting the waveform. Stripline’s true TEM mode has negligible dispersion up to the point where higher-order waveguide modes can propagate between the ground planes—typically well above 40 GHz for ground-plane spacings under 10 mil.

For wideband radar front-ends—particularly 77 GHz automotive radar—this dispersion difference matters. Most 77 GHz designs use GCPW or microstrip on very thin substrates (3–4 mil) to manage dispersion, but stripline feed networks remain common for the lower-frequency IF and LO distribution on the same board.

When to Choose Microstrip

• Component interface: Any trace that connects directly to a surface-mount component (amplifier, mixer, filter, antenna feed) must be on an outer layer, making microstrip the only option at that interface.
• Tuning and prototyping: Microstrip traces can be trimmed, probed, or modified with copper tape during development. Stripline cannot.
• Lower frequencies (< 6 GHz): Radiation is manageable, conductor loss is lower, and fabrication is simpler.
• Cost-sensitive designs: Two-layer boards with microstrip avoid the lamination steps stripline requires.
• Antenna feed transitions: Patch and slot antennas on the outer layer naturally connect to microstrip feeds.

When to Choose Stripline

• High isolation (> 30 dB): Multi-channel receivers, duplexers, and corporate feed networks where crosstalk must be minimized.
• EMI-sensitive environments: Stripline does not radiate and does not pick up external interference, critical in dense mixed-signal boards.
• Wideband and mmWave distribution: Lower dispersion and no radiation loss make stripline ideal for LO and IF routing in radar and 5G systems.
• Dense routing: When a four-layer or six-layer stackup is already justified, stripline layers add routing channels without extra board area.
• Broadside-coupled structures: Stripline couplers and filters benefit from the symmetric field distribution and tighter coupling control.

Design Checklist: Microstrip vs Stripline

1. Identify every RF net and classify it: needs surface access → microstrip; needs isolation or is internal distribution → stripline candidate.
2. Set the stackup: choose dielectric thickness and Dk to hit 50 Ω (or your target Z₀) on both microstrip and stripline layers. Verify with a 2D field solver, not just closed-form equations.
3. Select copper foil roughness: standard ED is acceptable for microstrip below 6 GHz; VLP or HVLP is strongly recommended for any stripline above 10 GHz.
4. Apply etch compensation per copper weight (use the calculator above) and confirm with the fab shop.
5. Add via stitching along microstrip traces above 10 GHz; add ground vias around stripline transitions to suppress parallel-plate modes.
6. Simulate transitions: every microstrip-to-stripline via transition adds 0.1–0.3 dB and a return-loss bump. Use back-drilled or controlled-depth vias and anti-pads sized for impedance continuity. Tools like HFSS, CST, or ADS can model these transitions accurately.
7. Validate impedance on the finished board with TDR or VNA; IPC-2141A provides guidance on controlled-impedance test coupons for both topologies.

Surface Finish Considerations

Both topologies are affected by surface finish on the conductor, but microstrip is more sensitive because the current flows on the exposed top surface of the trace. ENIG adds a thin nickel barrier layer (typically 3–6 µm) that is ferromagnetic and lossy at microwave frequencies, contributing 0.02–0.05 dB/cm of extra loss above 10 GHz. Immersion silver or bare copper with OSP avoids this penalty. For stripline, the inner-layer copper is not plated with ENIG—it contacts only the laminate—so the finish choice matters less. Details on finish-related loss are covered in the ENIG vs immersion silver comparison.

Frequently Asked Questions

Is stripline always lower loss than microstrip?

No. Below about 6–8 GHz on a low-Df laminate, microstrip often has lower total insertion loss because its wider trace reduces conductor loss and part of the field propagates in lossless air. Stripline’s advantage appears at higher frequencies where microstrip radiation loss and dispersion become significant.

Can I mix microstrip and stripline on the same board?

Yes, and most complex RF boards do. Surface-mount components connect through microstrip, then signals transition to stripline through vias for internal routing. Each via transition must be carefully impedance-matched; expect 0.1–0.3 dB of added loss per transition if well designed.

What is the maximum frequency for each topology?

Microstrip is used routinely to 77 GHz (automotive radar) on thin substrates, though GCPW is often preferred above 30 GHz. Stripline is practical to 40–60 GHz with tight ground-plane spacing; above that, higher-order modes and fabrication tolerances become limiting. Substrate integrated waveguide (SIW) is an alternative for the highest frequencies.

Does stripline require more layers?

Stripline needs a minimum of three conductor layers (two ground planes plus the signal layer), so the simplest stripline board is a four-layer stackup. Microstrip works on a two-layer board. The added layers increase cost, but also provide more routing and better power-plane decoupling.

How do I control impedance on stripline?

Impedance depends on trace width, trace thickness, and the dielectric thickness to each ground plane. Asymmetric stripline (signal layer not centered) is common in practice and requires a field solver that accounts for both dielectric gaps independently. Request impedance-controlled fabrication per IPC-2141A and specify test-coupon requirements on your drawing.