A low-loss laminate can still produce more channel loss than expected when the copper profile, trace geometry or launch differs from the assumptions in the simulation. Copper roughness deserves attention when a critical RF or high-speed path has limited loss margin, but a foil's roughness number alone cannot predict the loss of a finished board.
Start with the channel's frequency range and loss budget, then compare the actual copper and dielectric construction proposed for the critical layer. The low-loss PCB selection guide covers the wider choice of materials, transmission lines and verification. Here the focus is the copper surface: what a specification means, what to ask a fabricator and how to compare candidate constructions.
Why does copper roughness affect a high-speed trace?
As frequency rises, the skin effect concentrates more of the alternating current near the conductor surface. The copper is not perfectly smooth, particularly at the treated surface bonded to the dielectric. Surface texture can increase effective conductor resistance and contribute to attenuation. The amount depends on the surface geometry and the electromagnetic field distribution, not simply on whether the foil is described as smooth.
Both the signal conductor and the conductors carrying its return current matter. A stripline has different surrounding surfaces from a surface microstrip, and the relative influence of each interface depends on the actual stackup and geometry. Surface finish can also affect exposed conductors, while dielectric loss, transitions, connectors and mismatch contribute to the measured channel result.
That distinction changes the purchase decision. If the line already has ample measured margin, changing foil may not be necessary. If it does not, a smoother candidate is worth evaluating alongside the dielectric and the complete channel. It is not proof that the revised board will pass.
Roughness is not copper thickness
Copper thickness describes the amount of copper through the conductor; roughness describes variations in a surface. A thickness callout does not identify the profile or treatment. Likewise, two foils with the same nominal thickness can have different surface profiles and bond treatments. For thickness units and conversions, see PCB copper thickness.
Electrodeposited (ED) and rolled foils describe different manufacturing routes. Low-profile, very-low-profile and similar terms describe product positioning or foil profiles, but the label alone is not a measured value for every supplier and treatment. Ask for the exact product, thickness, surface and data rather than assuming that one manufacturing route is always smoother or lower loss.
The side of the foil matters
An ED foil can have different surfaces on its two sides. A bonding treatment may be applied to a particular side, and reverse-treated foil uses a different treatment orientation from a conventional construction. In a completed multilayer, additional preparation of copper surfaces can also be relevant to the final interface.
For a critical layer, ask the fabricator to identify the dielectric-facing surface and relevant treatment in the proposed construction. For a surface trace, also identify the exposed surface and finish. A data-sheet number taken from an unspecified side may not describe the interface used in your loss model.
How do you compare roughness data sheets?
A surface can be summarized in several ways. Ra is an arithmetic-average profile parameter; Rq is a root-mean-square profile parameter. Rz is a height-based profile parameter, but its precise calculation can depend on the referenced method. Surface-area parameters such as Sq describe a measured area rather than a single profile. Their values are not interchangeable, and there is no reliable universal conversion from one to another for an unknown texture.
Before comparing two figures, put the surrounding information beside them:
| Check | Why it matters | What to request if missing |
|---|---|---|
| Foil product and nominal thickness | A supplier can publish different values for different products or thicknesses. | Exact designation, thickness and current technical data sheet. |
| Surface and treatment | The treated dielectric-facing side may differ from the other side. | Which side was measured, its treatment, and which side faces the dielectric in the proposed stackup. |
| Parameter and unit | Ra, Rq, Rz and Sq summarize texture differently. | Parameter definition, units and the applicable measurement document or method. |
| Measurement setup and condition | Instrument, sampling, filtering and treatment state affect comparability. | Measurement method and enough setup details to establish whether the data can be compared. |
| Value type | A typical value is not necessarily a guaranteed production limit. | Whether the number is typical, maximum or another statistic, and its applicable specification. |
| Intended use | A foil property is not a channel measurement. | The proposed finished construction and any available representative electrical and adhesion evidence. |
For example, a supplier's product table may state Rz for a particular foil thickness while a laminate data sheet gives Sq for a treated surface. Those numbers should not be ranked directly. Return to the suppliers for comparable measurements or use each dataset with the model and assumptions for which it was supplied.
The Rogers Corporation document Copper Foils for High Frequency Circuit Materials discusses both mechanical and optical roughness measurement and illustrates why the measurement method and the treated surface must accompany a value. A foil manufacturer's product table is useful for identifying a candidate; it does not establish performance of a different finished stackup.
Should you select the smoothest available foil?
Select the construction that can meet both the electrical requirement and the project's mechanical and manufacturing requirements. Bond treatment helps the copper adhere to the dielectric; foil chemistry, treatment and laminate compatibility all matter. Lower measured roughness does not by itself establish either good or poor adhesion.
| Project situation | Practical next step | Evidence to request |
|---|---|---|
| Short route with comfortable measured margin | Keep the proposed construction unless another requirement justifies a change. | Representative electrical result and the approved stackup. |
| Long or loss-sensitive path with limited margin | Compare a lower-profile candidate with the existing construction over the relevant frequency range. | Candidate foil and treatment details, modeled assumptions and representative insertion-loss data. |
| A foil change improves a model but its bonding information is missing | Hold the substitution for a manufacturing and reliability review. | Laminate/foil compatibility, applicable adhesion or reliability evidence and the proposed process. |
| Alternate material is needed for supply | Treat it as a construction change, not merely a different trade name. | Updated stackup, data-sheet basis, impact assessment and any agreed requalification. |
The table is a decision aid, not a guarantee that a given foil type will pass. A manufacturer should confirm the construction it can actually build; the product team remains responsible for confirming the final system performance.
How should you compare two candidate constructions?
Suppose a designer has a high-speed path that misses its allocated insertion-loss budget in simulation. The fabricator proposes two possible copper profiles on the critical layer. The following is a hypothetical test plan, not a measured result or an APTPCB capability claim.
- Define the decision first. State the frequency range, reference planes, measurement uncertainty and the channel loss allocation. Decide in advance what electrical result and material documentation would justify changing the construction.
- Lock the common construction. Specify dielectric material and thickness, trace and reference-plane geometry, nominal copper thickness, layer location, and relevant finish. If a change to one of these is unavoidable, record it as a second variable instead of attributing the entire difference to roughness.
- Document the different interface. Identify foil product, treatment, dielectric-facing side, roughness parameter and data source for each candidate. Request the fabricator's actual stackup and any approved changes before manufacture.
- Agree on test structures and method. Use representative transmission-line structures and, where useful, more than one length so that fixture and line contributions can be investigated. Define calibration and de-embedding, connector launches, reference planes, frequency range and reporting format with the test provider before fabrication.
- Compare the electrical and physical evidence. Review the measured insertion-loss response and impedance for the relevant structures, then check construction records and the agreed adhesion or reliability evidence. Investigate unexpected differences in geometry, finish, dielectric or fixtures before assigning the result to foil profile.
- Make the release decision. Select a candidate only after the agreed electrical, manufacturability and reliability requirements are satisfied. Record the foil, treatment and stackup as controlled attributes and define how substitutions are reviewed.
Matching the design files alone does not ensure that two fabricated lines are identical. Include manufacturing variation in the interpretation, and test enough representative structures to support the intended decision. An impedance coupon can show whether its tested geometry meets an impedance requirement; it does not by itself measure insertion loss or qualify the entire channel.
What about roughness models in a field solver?
Models can help compare candidates before hardware is available, but they require inputs that match the chosen model. A single Rz value should not be entered as an Ra or Rq value, or converted into detailed surface geometry without a documented method. Some models need more information about texture than a basic supplier data sheet supplies.
When data is incomplete, label the assumption and compare a reasonable range of model inputs rather than presenting one predicted curve as a measurement. Ask the material supplier or fabricator for the applicable foil information and the model provider for the meaning and calibration of its inputs. Correlate the prediction with an agreed test structure when the decision depends on a narrow margin.
What should be included in the fabrication handoff?
Send the fabricator the relevant stackup and layer spans, critical transmission-line structures, target impedance, frequency range and loss budget. Identify any restrictions on laminate, foil product, treatment and substitution. Explain whether the critical path is internal or external and where the test reference planes should be.
Request the proposed pressed stackup, the foil designation and available data for the relevant interfaces, and an explanation of any alternate construction. Agree which impedance, insertion-loss, adhesion or reliability evidence is required, which test structures represent the design, and how results and material changes will be documented. Availability and test scope should be confirmed for the specific quotation.
To discuss a particular stackup, request a quote or contact APTPCB.
Frequently asked questions
Is a lower roughness number always better?
Not by itself. The measurement must refer to a comparable parameter, method, surface and treatment. The selected foil must also work with the laminate and process and meet the project's electrical and reliability requirements.
Does a controlled-impedance result prove low insertion loss?
No. Impedance and insertion loss answer different questions. An impedance coupon can meet its target while a long channel still exceeds its loss budget because of conductor, dielectric or transition effects.
Can I calculate board loss from a foil data sheet alone?
Not reliably. A useful estimate also needs the dielectric construction, geometry, relevant surfaces, frequency and a suitable model. A measured channel result includes additional features and test setup effects.
Does rolled copper always outperform ED copper?
No. Those terms describe manufacturing routes, not a universal ranking of surface profiles or finished-board loss. Compare the specific products, treatments and completed constructions for the application.