RF PCB Design Resources

50 Ohm Trace Width | Width Table for Every Laminate

This guide provides ready-to-use 50 ohm trace width values for common RF laminates, explains the variables that shift the number, and gives you a checklist to lock it down before fabrication.

Gold-plated microstrip trace on Rogers high-frequency laminate

What Determines 50 Ohm Trace Width?

A 50 ohm trace width is not a single number—it depends on the dielectric constant (Dk), substrate height (h), copper thickness (t), and whether you use microstrip or stripline topology. On standard 1.0 mm (39.4 mil) FR-4 with Dk ≈ 4.3, a 50 Ω outer-layer microstrip is roughly 1.85 mm (73 mil) wide. Switch to Rogers RO4350B (Dk 3.48) at the same height and the width jumps to about 2.2 mm (87 mil). These differences matter because even a 5 % width error can push return loss above −20 dB at a few GHz and wreck a filter or LNA match.

The four governing variables interact as follows:

  • Dk (dielectric constant): Higher Dk concentrates the field, requiring a narrower trace for 50 Ω.
  • Substrate height (h): Thicker dielectric needs a wider trace; thinner needs narrower.
  • Copper thickness (t): Thicker copper slightly lowers impedance for the same width, because the effective cross-section grows.
  • Topology: Stripline, buried between two ground planes, needs a narrower trace than microstrip for the same impedance because more field is captured.

Understanding how effective Dk differs from the datasheet bulk value is essential—microstrip fields fringe through air above the trace, lowering the effective Dk and widening the required trace compared to a naive calculation using bulk Dk alone.

50 Ohm Microstrip Width Table

Cross-section of multilayer PCB showing buried stripline trace

The table below gives approximate 50 Ω microstrip trace widths for popular laminates at common substrate heights. All values assume 1 oz (35 µm) copper, outer-layer microstrip with solder-mask removed over the signal trace, and the Dk values published at 10 GHz by each vendor.

Laminate Dk (10 GHz) h = 0.254 mm (10 mil) h = 0.508 mm (20 mil) h = 1.016 mm (40 mil)
FR-4 (generic) 4.3 0.46 mm (18 mil) 0.94 mm (37 mil) 1.90 mm (75 mil)
Isola I-Tera MT40 3.45 0.54 mm (21 mil) 1.10 mm (43 mil) 2.22 mm (87 mil)
Rogers RO4350B 3.48 0.53 mm (21 mil) 1.08 mm (43 mil) 2.20 mm (87 mil)
Rogers RO4003C 3.55 0.52 mm (20 mil) 1.06 mm (42 mil) 2.14 mm (84 mil)
Rogers RO3003 3.00 0.62 mm (24 mil) 1.26 mm (50 mil) 2.56 mm (101 mil)
Taconic TLY-5 2.20 0.78 mm (31 mil) 1.58 mm (62 mil) 3.20 mm (126 mil)

These are starting-point values derived from closed-form Hammerstad–Jensen equations. For production, always verify with a 2D field solver (the one built into your EDA tool or a standalone solver like Altium’s integrated impedance calculator or ANSYS 2D Extractor) and then confirm against the fab house’s impedance coupon data.

Why the Numbers Shift in Practice

Fabrication tolerances move the real impedance away from the calculated target. Typical etching tolerance is ±0.5 mil (±12.7 µm) on standard processes and ±0.3 mil on fine-line processes. On a 10 mil wide trace at 10 mil height on RO4350B, a +0.5 mil etch error drops impedance by roughly 1.5 Ω. Dielectric thickness tolerance (often ±10 % on thin cores) has an even larger effect. The practical rule: if your 50 Ω line is narrower than about 6 mil, you must specify tighter fab tolerances or accept wider impedance bands.

Surface roughness also matters above 10 GHz. Rougher copper increases effective conductor loss and slightly shifts the apparent Dk, which changes the width needed for 50 Ω. Our guide on mmWave copper roughness effects covers the Huray and Hammerstad models in detail.

50 Ohm Stripline Width

For inner layers, stripline is the standard topology. A symmetric stripline (signal centered between two ground planes) on RO4350B with total dielectric height b = 0.508 mm (20 mil) needs roughly 0.23 mm (9 mil) trace width for 50 Ω—about half the microstrip width on the same material at the same total height, because the trace is fully immersed in dielectric.

Asymmetric stripline (signal not centered) is common in multilayer stackups. Moving the trace closer to one ground plane lowers the impedance; compensate by narrowing the trace. Every stackup change demands a fresh solver run.

Trace Width at mmWave Frequencies

At 24 GHz and above, trace widths shrink because designers choose thinner substrates to suppress surface waves and higher-order modes. A 77 GHz automotive radar front-end on RO3003 at 5 mil height uses a 50 Ω microstrip only about 0.31 mm (12 mil) wide. At these dimensions, the mmWave PCB design rules for etch tolerance, via placement, and ground-plane continuity become critical.

Via transitions near 50 Ω lines also need attention—an uncompensated through-via stub can destroy return loss above 20 GHz. Backdrilling or using blind microvias is standard practice, as covered in our RF via design guide.

Length Matching on 50 Ohm Pairs

Once you have the correct 50 Ω trace width, differential pairs and clock-data buses often require length matching. Serpentine (trombone) routing adds electrical length to the shorter trace, but each bend introduces a small impedance bump. Keep the serpentine amplitude below 3× the trace width and the gap between bends at least 3× the line spacing to minimize mismatch.

Use the calculator below to find the extra physical length or serpentine parameters needed for a given delay target. Enter your trace velocity factor (typically 0.55–0.60 for microstrip on FR-4, 0.50–0.54 on RO4350B) and the required time-delay difference.

Loading Length Matching and Serpentine Delay…

For example, on RO4350B microstrip (velocity factor ≈ 0.54), matching a 10 ps skew requires adding about 1.6 mm of serpentine. If the serpentine amplitude is 0.5 mm and trace width is 0.27 mm, that works out to roughly three meander periods—easy to fit in most layouts without violating the 3× spacing rule.

The Formulas Behind the Table

mmWave radar PCB with controlled-impedance traces and via stitching

The most widely used closed-form model is the Hammerstad–Jensen equation set, documented in the original IEEE Transactions on Microwave Theory paper. For microstrip, the characteristic impedance Z₀ is:

Z₀ = (60 / √ε_eff) × ln(8h/W_eff + W_eff/4h) for W/h ≤ 1
Z₀ = (120π / √ε_eff) / [W_eff/h + 1.393 + 0.667 × ln(W_eff/h + 1.444)] for W/h ≥ 1

Where ε_eff is the effective dielectric constant and W_eff is the effective width adjusted for copper thickness. These equations are accurate to about ±1 % for W/h between 0.1 and 10 and frequencies where the quasi-static assumption holds (roughly below the first transverse resonance of the substrate).

For stripline, the Wheeler equation gives:

Z₀ = (60 / √Dk) × ln(4b / (π × d_e))

where b is the ground-plane spacing and d_e is the effective round-conductor diameter mapped from the rectangular trace. Stripline formulas are exact in the TEM limit, making them more reliable than microstrip formulas at moderate frequencies.

Pre-Fabrication Design Checklist

Before releasing Gerbers for a controlled-impedance 50 Ω board, walk through every item below. Missing even one—especially the stackup sign-off—is the most common reason RF boards come back with impedance out of spec.

Loading Design Checklist Builder…

After generating your checklist, verify that it includes at minimum:

  • Stackup drawing with exact dielectric thicknesses, Dk values and copper weights per layer.
  • Impedance table specifying target Z₀, tolerance (±5 Ω or ±10 %, whichever your spec requires), trace width, and reference layers.
  • Impedance coupon location on the panel (IPC-2141 recommends coupons on every panel, not every lot).
  • Solder-mask keep-out over controlled-impedance traces—mask adds roughly 1–2 Ω shift depending on thickness.
  • Notes on surface finish (ENIG, immersion silver, OSP) since finish plating thickness affects the conductor cross-section.
  • Via backdrilling or microvia requirements for any 50 Ω lines transitioning layers above 10 GHz.

For a deeper dive into RF DFM rules, including minimum annular ring, acid trap avoidance, and panel utilization, see our dedicated guide.

Common Mistakes and How to Avoid Them

Using Bulk Dk Instead of Effective Dk

Plugging the datasheet Dk directly into a microstrip formula overestimates the dielectric loading. The effective Dk for microstrip is always lower than bulk Dk because part of the field travels through air. On RO4350B (bulk Dk 3.48), the effective Dk for a typical 50 Ω microstrip is closer to 2.85–2.95 depending on W/h.

Ignoring Solder Mask

Leaving solder mask over a 50 Ω trace adds a dielectric layer (Dk ≈ 3.3–4.0) that lowers impedance by 1–3 Ω. Either remove mask from signal traces or include it in the solver model.

Not Accounting for Etch Compensation

Copper etching undercuts the trace, making the finished width narrower than the artwork. Fab houses apply etch compensation, but you should confirm the compensated width matches your impedance target, not just the artwork width.

FAQ

What is the 50 ohm trace width on FR-4?

On standard FR-4 (Dk ≈ 4.3) with 1 oz copper, a 50 Ω outer-layer microstrip is approximately 1.85 mm (73 mil) wide over a 1.0 mm (39.4 mil) dielectric, or about 0.46 mm (18 mil) wide over a 0.254 mm (10 mil) dielectric.

Does trace width change with frequency?

The quasi-static trace width itself does not change with frequency, but Dk shifts with frequency (FR-4 drops from ~4.5 at 1 GHz to ~4.0 at 10 GHz), and conductor losses from surface roughness become significant above 10 GHz. Both effects mean the “correct” width at 77 GHz is not the same as at 1 GHz on the same laminate.

How tight should impedance tolerance be for RF?

Most RF designs specify ±5 Ω (i.e., 45–55 Ω) on 50 Ω lines. Tighter tolerances like ±3 Ω are achievable but increase cost because the fab must use tighter dielectric-thickness and etch controls. For applications below 1 GHz, ±10 % is usually sufficient.

Can I use FR-4 for a 50 ohm RF trace?

Yes, up to about 3–6 GHz depending on loss budget. Above that range, FR-4’s high and variable Df (0.020–0.025) introduces excessive insertion loss, and its Dk tolerance (±5–10 %) makes impedance control unreliable. Low-loss laminates like RO4350B (Df 0.0037) are preferred for anything above 6 GHz.

Microstrip or stripline for 50 ohm?

Microstrip is simpler and cheaper (outer layer, no additional lamination), but radiates more and is sensitive to solder mask and nearby components. Stripline offers better shielding and tighter impedance control, at the cost of narrower traces that are harder to etch and vias required to reach the buried layer. Use stripline when isolation between adjacent channels matters or when routing through a dense multilayer stackup.

Ready for a quotation?

Send your requirements for a material, stackup and manufacturing review. Price and schedule are confirmed for your project.

Upload Gerber