Sunking PCB

High Frequency PCB Manufacturing Process Control: Lamination to Impedance Verification

Design Guide2026-09-16Sunking Circuits Technical Team
High Frequency PCB Manufacturing Process Control: Lamination to Impedance Verification

Choosing the right laminate does not guarantee a compliant board

The usual sequence looks like this: the RF engineer works out target impedance and a loss budget, picks a Rogers, Taconic or Isola grade, copies Dk and Df into the drawing, and sends out the RFQ. Nothing wrong with that, but it only locks down the as-supplied properties of the material.

What determines the finished board's electrical behaviour is the dozen-plus process steps that material passes through inside the factory. If dielectric thickness shifts after lamination, characteristic impedance shifts with it. If etched line width comes out narrower than drawn, impedance rises. If hole-wall plating is uneven, the RF return path picks up one more discontinuity. The Df 0.0009 on a datasheet is a laboratory-condition value, not a measurement of the board in your hand.

So qualifying a high-frequency fabricator is not about which laminates it can buy — that is a distributor's capability. The questions worth asking are: what is controlled during lamination, what is measured after etching, what is judged after drilling, and what reports come with the shipment. Sunking PCB's published high-frequency envelope is Rogers/Taconic/Isola base materials, Dk 2.2–10.2, Df 0.0009–0.003, a 1GHz–77GHz frequency range, ±5% impedance control and 2–16 layers. The sections below map each of those numbers to the step that holds it.

Step one: mixed lamination controls dielectric thickness consistency and interlayer bond

High-frequency boards rarely use RF laminate throughout. The reasons are practical: PTFE-family materials are expensive, less rigid, and harder to drill and plate, while only one or two layers on a typical board actually carry high-frequency signals. The normal approach is mixed dielectric lamination — RF layers and FR-4 layers pressed in one stackup, the former carrying the RF chain, the latter carrying power, control and structure.

Mixed lamination puts three problems on the fabricator's side at once.

First, dissimilar materials expand and flow differently. RF laminate and FR-4 use different resin systems, so flow behaviour and cure shrinkage differ under pressure. The control objective is reliable interlayer adhesion with no delamination, which depends on the press programme and prepreg pairing. Those are process parameters, not drawing values.

Second, dielectric thickness consistency drives impedance directly. Characteristic impedance is highly sensitive to dielectric thickness. If post-lamination thickness varies across the panel, impedance varies with it. This is the single most watched quantity in mixed lamination, and it decides whether a ±5% impedance window is achievable at all.

Third, lamination registration. High-frequency boards often carry shield cavities, openings in reference planes, and impedance layers paired to specific references. Layer-to-layer shift changes the effective reference spacing. The registration logic is the same as any multilayer build, but the tolerance for error is smaller.

How to arrange the stackup — RF layers on the outside or inside, how references are paired, what order to press in — has no universal answer. It depends on frequency band, layer count and structure, and it belongs in engineering review. What the design side should supply is the frequency band, target impedance and reference-layer intent, rather than a prescribed press sequence.

Step two: imaging and etching control line-width deviation

The second dominant impedance variable is line width. The site-wide tolerance for line width is ±20%, a general value for standard boards, while the high-frequency product page states ±5% impedance control. Seeing both numbers together is confusing, but they do not conflict: ±20% is the fabrication tolerance of one geometric dimension; ±5% is the electrical outcome of controlling several variables together.

To land impedance inside ±5%, the fabricator cannot let line width wander across the full ±20% band. It compensates deliberately: artwork dimensions are pre-adjusted for copper weight and etch behaviour so the etched result lands near target. The compensation amount depends on copper weight, etch line condition and pattern density. It is factory process data, not a formula that transfers between shops.

The practical implication for the design side is that a tight impedance tolerance has to be called out. Whether the drawing says ±10% or ±5% changes the etch control strategy, the test frequency and the release method. Per the claim-register tiering, ±10% is the standard tier and can be accepted unconditionally; ±5% sits in the advanced tier and requires engineering review to confirm feasibility. Without a callout, the standard tier applies.

Step three: drilling and hole metallisation control position, wall condition and aspect ratio

High-frequency boards are harder on drilling than standard boards, not because of the hole itself but because of the field distribution around it. Positional error changes the real spacing from via to reference plane and to adjacent traces. A rough or unevenly plated wall introduces a local impedance step in the return path.

The controlled quantities match the site-wide envelope: 0.2 mm minimum mechanical hole (0.15 mm in the advanced tier, subject to engineering review), 0.1 mm minimum laser hole (0.075 mm advanced, subject to review), PTH diameter tolerance ±0.075 mm, hole position accuracy ±0.05 mm, and PTH aspect ratio 10:1. Board thickness runs 0.2–6.0 mm.

The 10:1 aspect ratio deserves attention as a real constraint on RF designs. A thick board with small holes hits that line quickly: a 3 mm board with 0.25 mm holes is 12:1, outside the standard envelope. This combination is self-checkable at design time — state board thickness and minimum hole size together and feasibility can be answered at quotation instead of after a rejected job.

The laminate itself also affects drilling. PTFE-family materials are soft and prone to smearing, so hole-wall condition differs from standard FR-4 and both drilling and wall preparation are adjusted per material. The parameters are process content; what can be confirmed externally is that drill and plating feasibility for these materials is a review item, and the answer can differ between grades.

Step four: surface finish, direction of effect only

Surface finish sits in the high-frequency current path, which is why it matters more here than on standard boards. Two directional statements are safe:

  • nickel is magnetic and adds loss at high frequency, so nickel-bearing finishes are generally not the first direction to consider for RF paths;
  • surface roughness affects conductor loss, and the effect becomes more pronounced as frequency rises.

Beyond that, any statement needs measured data for a specific material at a specific frequency. Claims of the form "finish X costs Y dB less in band Z" should not be used for selection unless the test conditions and method come with them. Finish selection belongs in engineering review alongside solderability and storage requirements for the actual band and product life. Comparison dimensions for each finish are on the surface finish page.

Step five: the verification chain controls what evidence can be delivered

The first four steps are process control. All of it has to resolve into deliverable evidence. The high-frequency release chain has four elements:

  1. Impedance testing. TDR measurement on a coupon or designated points, the direct check on whether the preceding steps held the impedance window. The drawing needs to identify impedance layers, target values, tolerance and coupon requirements.
  2. 100% electrical test. Site-wide, every order gets 100% electrical test by flying probe or fixture, verifying opens and shorts.
  3. FQC and visual inspection. 100% FQC is one element of the 10+ quality gates, which include 100% electrical test, 100% FQC and per-batch reliability testing.
  4. Per-batch reliability testing. Run by batch to verify batch-to-batch consistency.

The output is a record set that ships with the boards. Published quality figures are a 99.8% outgoing pass rate (defined per production batch as shipped good units ÷ total units produced × 100%) and a scrap rate below 0.3% (scrapped units ÷ total units produced × 100%). System-level certifications are ISO 9001:2015, ISO 14001:2015 and IATF 16949:2016, all issued by DNV, plus UL (US and Canada), RoHS and REACH. Acceptance and visual criteria follow the applicable class of IPC-6012 and IPC-A-600, with the class agreed per order.

Confirmable at quotation vs requiring engineering review

Two lists, which is the most useful form for an RFQ.

Confirmable at quotation (standard tier): 2–16 layers; 0.2–6.0 mm board thickness; mechanical holes ≥0.2 mm and laser holes ≥0.1 mm; aspect ratio ≤10:1; ±10% impedance tolerance; PTH diameter ±0.075 mm and hole position ±0.05 mm; ±20% line width; common grades inside Dk 2.2–10.2 and Df 0.0009–0.003; 1GHz–77GHz.

Requires engineering review (advanced tier or structural judgement): ±5% impedance; 0.15 mm mechanical or 0.075 mm laser holes; the specific mixed-lamination arrangement and press sequence; RF-to-FR-4 pairing and prepreg selection; drill and metallisation feasibility for a particular PTFE grade; finish selection for the target band; thick-board-plus-small-hole combinations that approach the aspect ratio limit.

The line is not a hedge. It keeps quoted numbers meaningful: standard-tier values can go straight into the order, advanced-tier values need the actual stackup and artwork before they can be committed. The full capability envelope is on the manufacturing capabilities page and tolerance definitions are on the tolerance page.

What to include in an RFQ

Working backwards from the control logic, a high-frequency RFQ that can go straight into review should carry: target frequency band with target impedance and tolerance for the critical nets; stackup intent (which layers carry RF, where the references sit); material grade or an acceptable substitution range; board thickness and minimum hole size; finish requirement or service environment; and test and acceptance requirements (whether an impedance report is needed, which IPC class applies).

With those in hand, feasibility and a process route can be answered in one pass. Without them, a fabricator either quotes on the most conservative assumptions or comes back for repeated clarification. Submission and quotation entry is on the quote page.

FAQ

Q1: Does a high-frequency board have to be all Rogers?

No. Mixed lamination is the normal approach: RF layers carry the high-frequency chain while FR-4 layers carry power, control and structure. An all-RF build is usually only necessary when structure or frequency band demands it, decided in stackup review.

Q2: If the drawing says ±5% impedance, is it guaranteed?

±5% is an advanced-tier tolerance and requires engineering review. Feasibility depends on stackup, line width, copper weight and material combination, and not every design can be held to that window. The standard tier is ±10%.

Q3: Do ±20% line width and ±5% impedance contradict each other?

No. ±20% is the general fabrication tolerance for one geometric dimension; ±5% is the electrical result of controlling several variables together. Reaching ±5% requires deliberate etch compensation rather than letting line width use the full band.

Q4: What is the minimum hole size on a high-frequency board?

Standard tier is 0.2 mm mechanical and 0.1 mm laser; advanced tier is 0.15 mm and 0.075 mm respectively, subject to engineering review. The 10:1 PTH aspect ratio applies as well, so thick boards with small holes may be limited by it.

Q5: How do I verify the impedance of boards I receive?

Ask for the impedance test report (TDR on a coupon or designated points) and identify impedance layers, target values, tolerance and coupon requirements on the drawing. Every order also gets 100% electrical test by flying probe or fixture, 100% FQC, and per-batch reliability testing.

Q6: Is 77GHz the ceiling?

The published frequency range on the high-frequency page is 1GHz–77GHz. Applications beyond that range need separate review and fall outside the published envelope.

Related links

Need Professional PCB Manufacturing?

Founded in 2009, we provide one-stop PCB solutions from prototype to mass production.