Copper Roughness in 5G, Radar, and SatCom RF Hardware: The Hidden Loss Mechanism

At 77 GHz, copper skin depth is only about 0.24 µm. A 2.0 µm RMS copper surface is therefore roughly eight skin depths rough.

That microscopic geometry can become a system-level RF-loss mechanism. It affects mm-Wave feeds, Massive MIMO, Radar, and Satellite Communications.

At these frequencies, copper is no longer an ideal conductive sheet. Its surface morphology becomes part of the transmission line.

The Core Physics: Skin Depth Versus Copper Roughness

The classical skin depth of a good conductor is:

δ = √[2/(ωμσ)]

where:

• ω = 2πf

• μ is conductor permeability

• σ is electrical conductivity

Ansys uses the same skin-depth relationship within its conductor-roughness modelling framework. [Ref. 1]

Using copper conductivity:

σ ≈ 5.8 × 10⁷ S/m

the approximate skin depths become:

A 2026 peer-reviewed mm-Wave substrate review likewise uses 5.8 × 10⁷ S/m for copper conductivity. The same review calculates copper skin depths of approximately 0.382 µm at 30 GHz, 0.330 µm at 40 GHz, and 0.270 µm at 60 GHz. [Ref. 2]

Now compare those dimensions against real PCB copper.

Rogers reports approximately 2.0 µm RMS surface roughness for standard ED copper. The same Rogers comparison reports approximately 0.35 µm RMS for rolled copper. [Ref. 3]

At 28 GHz: 2.0 µm / 0.395 µm ≈ 5.1 skin depths

At 77 GHz: 2.0 µm / 0.238 µm ≈ 8.4 skin depths

At 110 GHz: 2.0 µm / 0.199 µm ≈ 10 skin depths

That is the critical transition. The RF current no longer experiences an approximately planar copper interface. It propagates along an electrically significant microscopic surface.

Why the Current Path Becomes More Expensive

Skin effect confines high-frequency current near the conductor surface. Once roughness approaches the skin-depth scale, surface topology matters strongly.

A 2026 review identifies copper-surface roughness as increasingly important above roughly 20–30 GHz. The same review explains that roughness increases current crowding and effective path length. [Ref. 2]

Rogers likewise states that increasing copper roughness increases conductor loss. Rogers also reports that rougher copper causes greater propagation phase delay. [Ref. 4]

Ansys states that surface roughness can increase conductor power loss beyond twice smooth-conductor loss. [Ref. 5]

That means copper roughness influences more than S₂₁ magnitude. It can influence:

• conductor attenuation

• phase velocity

• effective dielectric constant

• group delay

• resonator Q

• antenna-feed efficiency

• channel-to-channel phase tracking

Ansys specifically identifies higher insertion loss and reduced resonator Q as consequences of rough conductor surfaces. [Ref. 6] At mm-Wave, copper loss and phase become coupled manufacturing problems.

Pain Point 1: Copper Roughness Can Dominate Thin mm-Wave Transmission Lines

PCB insertion loss can be considered as several combined mechanisms:

αTOTAL = αC + αD + αR + αL

where:

• αC = conductor loss

• αD = dielectric loss

• αR = radiation loss

• αL = other leakage-related loss

Rogers identifies conductor, dielectric, and radiation losses as key mm-Wave transmission-line loss mechanisms. It also states that thinner circuits are more strongly affected by copper roughness. [Ref. 4]

That matters because many practical mm-Wave circuits use thin substrates. Thin substrates can reduce microstrip radiation losses compared with thicker substrates. But thin dielectric sections also require narrower transmission-line geometries.

Those narrower structures can increase conductor-loss sensitivity. Copper roughness therefore becomes much more than a fabrication detail.

A Real Rogers Benchmark: Same Laminate, Different Copper

Rogers performed a particularly useful controlled experiment. The dielectric material remained the same: 5-mil RO3003 laminate. Only the copper type changed.

The measured roughness values were:

ED copper ≈ 2.0 µm RMS

Rolled copper ≈ 0.35 µm RMS

Rogers measured 50 Ω microstrip lines from DC through 110 GHz for this comparison. The rougher ED copper produced greater conductor and insertion loss. The same experiment showed that thinner laminates were more strongly affected by rough copper. [Ref. 3]

This comparison is powerful. The dielectric remained essentially controlled. The dominant changed variable was the conductor surface.

How Electrically Different Are Those Copper Types?

At 28 GHz:

Rolled copper: 0.35 / 0.395 ≈ 0.89δ

ED copper: 2.0 / 0.395 ≈ 5.1δ

At 77 GHz:

Rolled copper: 0.35 / 0.238 ≈ 1.47δ

ED copper: 2.0 / 0.238 ≈ 8.4δ

The same physical foil therefore becomes electrically rougher as frequency increases. This explains why a stack-up acceptable at 6 GHz may disappoint at 77 GHz. The laminate name alone does not define mm-Wave insertion loss.

The copper profile matters. Rogers explicitly identifies copper-surface roughness as an important mm-Wave conductor-loss variable. [Ref. 4]

A 2026 Review Quantifies the Same Engineering Trend

A peer-reviewed 2026 review examined conductor roughness across 30–60 GHz. [Ref. 2]

Its representative roughness values were:

• RA copper ≈ 1.8 µm RMS

• LP copper ≈ 0.9 µm RMS

• VLP copper ≈ 0.4 µm RMS Electronics 2026

Using a Hammerstad-type roughness correction, the review estimated an 11–15% conductor-loss reduction when moving from RA to VLP copper across 30–60 GHz.

Its calculated roughness-factor reductions were approximately:

• 15.0% at 30 GHz

• 13.4% at 40 GHz

• 11.3% at 60 GHz Electronics 2026

The same study also examined a 50 Ω microstrip on RO4350B from 30–60 GHz. At 60 GHz, its calculated conductor loss changed from approximately 0.094 dB/cm to 0.062 dB/cm when moving from RA to VLP copper. That corresponds to roughly a 34% reduction in the calculated conductor-loss value for that specific model.

That number should not be treated as universal. Its magnitude depends on substrate, geometry, roughness, and modelling assumptions. But the engineering direction is unambiguous. Copper morphology materially changes mm-Wave transmission loss.

Pain Point 2: Rough Copper Changes Phase, Not Just Loss

This is often missed during PCB material selection. Rough copper does not only raise attenuation. It also alters propagation delay.

Rogers reports that rougher copper increases propagation phase delay. It also reports that circuits using rougher copper can exhibit higher effective Dk. The company describes this as a wave-slowing effect that becomes increasingly important at high frequencies. [Ref. 3]

That creates a serious phased array problem.

Why This Matters for Phased Arrays

Beam steering depends on controlled relative phase. For a transmission line:

β = 2πf√εeff / c and φ = βl.

Therefore: φ ∝ √εeff

For small perturbations: Δφ/φ ≈ ½ · Δεeff/εeff

Consider an illustrative 30 mm feed line at 28 GHz.

Assume: εeff ≈ 3

The guided wavelength becomes approximately: λg ≈ 6.18 mm.

A 30 mm line therefore spans roughly: 4.85 guided wavelengths or approximately: φ ≈ 1,747°

Now assume modelling error changes effective permittivity by only 2%. The accumulated phase difference becomes approximately: Δφ ≈ 17.5°.

That calculation is illustrative. But its physical premise is not.

Rogers confirms that copper roughness can increase effective Dk and propagation phase delay. [Ref. 3]

A 17° Phase Error Is Not Small in Beamforming Hardware

Such an accumulated feed-phase error can influence:

• beam pointing

• sidelobe level

• null depth

• polarization purity

• channel calibration

• coherent combining

A uniform roughness offset may potentially be calibrated. Channel-dependent and lot-dependent variation is more difficult.

Rogers notes that copper-surface variations can affect both insertion loss and phase response at mm-Wave frequencies. That is no longer only a laminate-selection problem. It becomes a production-repeatability problem.

Massive MIMO Makes the Problem Statistical

Imagine a 256-channel mm-Wave radio. Each channel may contain several centimetres of PCB feed network.

Now introduce small channel-dependent copper variations. The consequences can include:

• unequal S₂₁

• unequal phase delay

• broader calibration spread

• frequency-dependent residual error

• channel-dependent realized EIRP

The system may calibrate successfully at one frequency. Broadband behaviour can still remain unequal. This becomes particularly relevant to massive MIMO and open RAN radios.

Pain Point 3: Smooth-Conductor Simulation Can Be Dangerously Optimistic

A perfect-electric-conductor model establishes an ideal upper boundary. Finite conductivity improves realism. But finite conductivity alone does not describe manufactured copper morphology.

Ansys states that surface roughness is important for accurate RF-component and PCB modelling. HFSS supports both Groiss and Huray surface-roughness models. It's enhanced Huray implementation is causal and broadband. The enhanced model also predicts increased phase delay produced by surface roughness. [Ref. 6]

That final point is crucial. A magnitude-only correction is insufficient for phased array analysis.

Groiss, Hammerstad, or Huray?

Not every roughness model serves the same purpose.

Hammerstad-Type Correction

Hammerstad-based corrections are useful for approximate roughness-loss estimation. Ansys implements a modified Hammerstad model in SIwave and 2D Extractor workflows. The model uses RMS roughness and a scale factor. [Ref. 7]

Groiss

The Groiss model relates conductor roughness directly to copper skin depth. Its roughness factor is limited to a maximum value of 2.

Ansys therefore describes it as appropriate for relatively polished conductors. It states that it is not appropriate for significantly rougher PCB copper. [Ref. 1]

Huray

The Huray approach models copper morphology using spherical surface nodules. Its inputs include nodule radius and Hall-Huray surface ratio.

Ansys' enhanced Huray model supports rough manufactured PCB copper. It can model both additional conductor loss and increased propagation delay. [Ref. 5]

For critical mm-Wave modelling, the chosen roughness model therefore matters.

What Should Go Into Your HFSS Model?

Do not enter only:

“Copper conductivity = 5.8 × 10⁷ S/m”

and assume the conductor is solved.

For higher-fidelity modelling, consider:

• conductivity

• finished copper thickness

• etch profile

• sidewall geometry

• top-surface roughness

• bottom-surface roughness

• sidewall roughness

• laminate Dk

• laminate Df

HFSS 3D Layout allows separate roughness definitions for top, bottom, and side conductor surfaces. [Ref. 8]. That capability matters for etched PCB conductors.

Rogers identifies the copper-to-substrate interface as particularly important for roughness-induced conductor loss.

The electromagnetic model should therefore represent the actual fabricated conductor. Not merely its nominal Gerber outline.

Do Not Confuse RMS Roughness With Other Surface Metrics

Copper-surface characterization can use several roughness descriptors.

Examples include:

• Ra

• Rq / RMS

• Rz

• areal roughness metrics

Those values describe different statistical characteristics. They should not be treated as interchangeable.

The Rogers 2.0 µm and 0.35 µm benchmark values are explicitly stated as RMS roughness. [Ref. 3]

Therefore, a procurement requirement saying: “Copper roughness ≤ 1 µm” is incomplete. The measurement definition must also be specified.

Why This Matters for 5G Infrastructure

At mm-Wave, feed-network loss directly reduces useful RF delivery.

Consider an illustrative additional feed loss of 0.5 dB. The linear transmitted-power ratio becomes:

10^(−0.5/10) ≈ 0.891

Therefore approximately 10.9% of RF power is lost before reaching the next RF stage.

For a simplified 256-channel system carrying 1 W per channel:

Ideal aggregate RF power: 256 W

After an additional 0.5 dB loss: ≈ 228 W

Difference: ≈ 28 W

That calculation does not claim copper roughness always contributes 0.5 dB. It illustrates why fractions of a decibel matter architecturally.

Every additional feed loss affects the link budget or thermal balance.

For Receive Arrays, the Penalty Can Be Even More Expensive

Loss before the LNA directly damages receive-system sensitivity.

For satellite communications, such front-end loss reduces effective G/T. A feed-loss increase of 0.5 dB can correspond approximately to a 0.5 dB G/T penalty when other quantities remain unchanged.

That is particularly expensive in LEO and Ground-station architectures. These systems already contend with:

• free-space path loss

• atmospheric attenuation

• polarization mismatch

• scan loss

• aperture limitations

Copper roughness therefore belongs inside the receive link budget, not merely inside the fabrication notes.

For Radar, Phase Can Matter as Much as Raw Loss

77/79 GHz Radar depends heavily on coherent channel behaviour. Angle estimation and beamforming rely on predictable phase relationships.

Rogers identifies phase-angle stability as especially important in 77 GHz automotive Radar circuits. This requirement also explains modern laminate development. [Ref. 4]

Rogers positions RO3003G2 specifically for next-generation 77/79 GHz automotive radar. RO3003G2 uses very low profile ED copper to reduce insertion loss.

Its published Design Dk is approximately 3.07 at 77 GHz.

For a 5-mil laminate, Rogers reports approximately 1.3 dB/in insertion loss using its microstrip differential-phase-length method.

That is direct commercial evidence. At mm-Wave, copper profile becomes a product-level design variable.

A Recent 2026 Benchmark Reinforces the Trend

A 2026 peer-reviewed review evaluated representative roughness effects from 30 to 60 GHz. Using representative RA and VLP copper roughness values, its Hammerstad-based calculation produced approximately 11–15% lower conductor-loss factor across that range. [Ref. 2]

The review also reports that rougher copper consistently produces greater conductor loss at mm-Wave frequencies.

That result is model-dependent. It is not a universal laminate guarantee. But it reinforces the strategic conclusion:

Copper selection can materially change mm-Wave transmission performance.

Aerospace & Defense Hardware Adds Another Constraint

Radar and EW systems often require tight phase repeatability. Phased arrays require predictable channel-to-channel behaviour.

Environmental qualification can add:

• temperature cycling

• vibration

• supplier traceability

• long service life

Copper-profile variability therefore enters the RF repeatability problem. MIL-STD-461 addresses electromagnetic compatibility. It does not specify copper-surface roughness.

But additional RF loss can force higher internal transmit power that can reduce EMC margin elsewhere. PCB material engineering and EMC engineering therefore interact indirectly.

Satellite Communications Needs Consistency, Not Only Low Loss

Ground stations may tolerate larger physical hardware. LEO payloads generally face tighter mass and packaging constraints.

Payload electronics can simultaneously require:

• low RF loss

• thermal efficiency

• stable phase

• radiation resilience

Rad-Hard layout addresses radiation-related semiconductor reliability. It does not reduce copper-surface resistance.

Multipactor mitigation addresses high-field vacuum discharge. It does not correct PCB conductor roughness.

These are separate failure mechanisms. A robust payload must address each one independently.

The Strategic Engineering Solution

Top-tier RF organisations manage copper roughness before layout release. They do not discover it during VNA correlation.

1. Put Copper Profile Into the Stack-Up Specification

Do not specify only: “1 oz copper.”

Also define:

• copper foil type

• roughness metric

• maximum roughness

• treated side

• finished copper thickness

• plating requirements

At mm-Wave, foil profile belongs beside Dk and Df.

2. Request the Correct Supplier Data

Ask for:

• RMS or Rq roughness

• Rz where relevant

• treated-side morphology

• profilometry or equivalent data

• lot variation

• foil manufacturer and grade

Avoid relying only on the laminate product name. Rogers demonstrates that the same 5-mil RO3003 dielectric produces different mm-Wave insertion loss when copper roughness changes. [Ref. 3]

3. Use Low-Profile Copper Where the Link Budget Justifies It

Potential options include:

• rolled copper

• VLP copper

• HVLP copper

• reverse-treated copper

Do not select these from marketing terminology alone. Use measured foil morphology and validated RF loss.

Rogers' current RO3003G2 platform deliberately uses VLP ED copper to achieve lower mmWave insertion loss. [Ref. 3]

Rogers also launched RO4830 Plus using VLP ED copper for 76–81 GHz Radar applications. [Ref. 9]

4. Build Roughness Into HFSS Before Optimizing Geometry

Run at least two electromagnetic cases.

Case A — Smooth Conductor

Establish the ideal conductor baseline.

Case B — Realistic Surface Roughness

Use supplier-supported roughness parameters. The difference reveals the predicted roughness penalty.

HFSS directly supports Groiss and Huray surface-roughness modelling. For rough PCB copper, Ansys recommends the Huray approach over Groiss when loss-enhancement factors exceed two.

5. Correlate Simulation With Differential-Length Coupons

One transmission line can leave connector effects embedded. Differential-length structures can better isolate propagation behaviour.

Rogers' mm-Wave copper comparison uses differential-length transmission-line measurements. Its published microstrip comparisons extend into the mm-Wave region. [Ref. 10]

Use those coupons to extract:

• attenuation per unit length

• phase delay per unit length

• effective Dk

• roughness-model correlation

This creates a far stronger material validation process.

6. Validate Both Magnitude and Phase

Do not correlate only: |S₂₁|

Also compare:

• S₂₁ phase

• group delay

• extracted effective Dk

• TDR impedance

• temperature response

Rogers states that copper roughness affects both insertion loss and propagation phase. [Ref. 4]

Ansys' causal Huray model likewise includes the additional phase delay caused by roughness. [Ref. 5]

A magnitude-only correlation can therefore hide modelling error.

7. Perform Lot-Level Validation for Critical Programmes

For high-volume Massive MIMO hardware: Measure multiple production panels.

For Radar: Track phase consistency across lots.

For Satellite Communications: Track feed loss after environmental qualification.

The goal is not one golden prototype. The goal is predictable production statistics.

The B2B Procurement Implication

A mm-Wave PCB procurement drawing should not simply state: “Use Rogers material.”

That is incomplete. A stronger requirement should define:

• laminate designation

• Design Dk

• Df

• copper foil type

• RMS/Rq roughness

• maximum acceptable roughness

• finished conductor thickness

• plating finish

• insertion-loss coupon

• phase-delay measurement

• lot traceability

For Phased Array programmes, also specify:

• channel-to-channel phase tolerance

• feed-loss tolerance

• calibration residual

• realized array gain

For Radar, include:

• 77/79 GHz phase stability

• thermal correlation

• production repeatability

For Satellite Communications, include:

• feed-network attenuation

• EIRP impact

• G/T impact

• environmental verification

At mmWave, material procurement becomes part of RF architecture.

The Executive Takeaway

At 28 GHz: δCu ≈ 0.395 µm

At 77 GHz: δCu ≈ 0.238 µm

Meanwhile, Rogers reports:

Rolled copper ≈ 0.35 µm RMS

Standard ED copper ≈ 2.0 µm RMS

That makes the 2.0 µm ED surface approximately:

5.1 skin depths at 28 GHz and 8.4 skin depths at 77 GHz

Rogers experimentally demonstrates that the rougher copper produces greater insertion loss in otherwise similar RO3003 microstrip structures. It also reports that rough copper increases effective Dk and propagation delay. [Ref. 3]

Copper roughness therefore influences:

• conductor loss

• insertion loss

• phase velocity

• effective Dk

• resonator Q

• phased-array calibration

• realized antenna efficiency

So the correct mm-Wave design question is not:

“Which PCB laminate am I using?”

It is:

“What electromagnetic surface does the RF current actually see?”