High‑Frequency PCB Supplier Checklist: 5 Must‑Verify Data Points

Engineering Summary

Five measured data points separate high‑frequency PCB suppliers who truly control their process from those who only claim capability.

You must verify:

• 1) Dk/Df lot‑acceptance data
• 2) Copper roughness (Ra) on RF layers
• 3) Δ‑Loss coupon results
• 4) Impedance distribution (Cp/Cpk)
• 5) Thermal cycling / CAF test reports

Rule of thumb: A supplier that cannot provide measured data for all five cannot guarantee repeatable RF performance across production lots.

Why a checklist, not a marketing tour

Every high‑frequency PCB supplier claims “tight impedance,” “low loss,” and “high reliability.” But when production boards drift, the cost of failure lands on you – not them.

This checklist replaces trust with verification. Use it before you release prototypes, qualify a new vendor, or scale to volume.

1. Dk/Df lot‑acceptance data (material stability)

Why it matters: Laminate Df variation of just ±0.001 adds ≈0.15 dB/cm loss at 28GHz. Datasheet Dk/Df are typical values at 10 GHz – your production lot may differ.

What to ask for:

• Incoming test report per lot using T‑resonator or split‑post resonator (IPC‑TM‑650‑2.5.5.13)
• Dk and Df at your operating frequency (e.g., 28GHz or 77GHz), not only 10GHz
• Lot traceability to your production panels

Common supplier gap: “We use Rogers material – it’s already qualified.” 
Reality: Rogers qualifies the material, not your supplier’s lot handling. Always demand lot‑specific data.

2. Copper roughness (Ra) on RF layers

Why it matters: At 28‑77GHz, skin depth is 0.23‑0.39 µm. Standard ED copper (Ra ≈2.0 µm) can increase loss by 30‑50% vs HVLP.

What to ask for:

• Profilometry measurement (ISO 25178) for each production lot
• Ra (arithmetical mean height) and Rz (average max height) on RF layers
• Statement that HVLP or VLP copper is used, not left to “fabricator’s choice”

Common supplier gap: “We use low‑loss copper.” 
Reality: “Low‑loss” is not a specification. Demand Ra numbers and verify with profilometry.

3. Δ‑Loss coupon results (measured insertion loss)

Why it matters: Impedance coupons confirm geometry – they do not measure loss. Δ‑Loss coupons (IPC‑TM‑650‑2.5.5.13) remove connector and launch effects, giving true PCB‑only loss.

What to ask for:

• Δ‑Loss coupon data for each critical RF layer and line type
• Loss per cm (or per inch) at your operating frequency
• Panel‑to‑panel and lot‑to‑lot variation statistics

Common supplier gap: “We test impedance and it passes.” 
Reality: Impedance passing does not guarantee loss. Demand Δ‑Loss data or prepare for surprises.

4. Impedance distribution (Cp/Cpk) – not just min/max

Why it matters: A single impedance reading or “within ±10%” claim hides distribution tails. Cp/Cpk tells you how consistently the supplier holds target impedance across panels and lots.

What to ask for:

• TDR measurement data from at least 30 panels (or statistically significant sample)
• Cp and Cpk values for each controlled impedance net group
• Upper and lower specification limits (USL/LSL) and process mean

Common supplier gap: “Our impedance is within ±10%.” 
Reality: A process can be within ±10% but have Cp=0.8 – meaning 5‑10% of boards may fail after shipment. Demand Cp/Cpk, not just range.

5. Thermal cycling / CAF test reports (long‑term reliability)

Why it matters: High‑frequency PCBs often operate in harsh environments (‑40°C to +125°C, high humidity). Without thermal cycling and CAF testing, you have no proof of long‑term reliability.

What to ask for:

• Thermal cycling report (IPC‑6018 Class 3 or equivalent) – cycles, temperature range, resistance monitoring
• CAF test report (IPC‑TM‑650‑2.6.25) – spacing, voltage, temperature/humidity, time to failure
• Cross‑section images of microvias and plated through‑holes after stress

Common supplier gap: “We follow IPC‑6012.” 
Reality: IPC‑6012 Class 2 does not require thermal cycling or CAF testing for many structures. Demand reports, not certifications.

Supplier response evaluation table

How to use this checklist in procurement

1. Before quoting: Ask for existing data from similar designs (same material, layer count, frequency band).
2. During prototype phase: Require that all five data points be measured on the first article.
3. Before volume production: Establish acceptance limits and agree on sampling frequency (e.g., Cp/Cpk per lot, Δ‑Loss per panel).
4. Ongoing: Track trends – a drifting Cp or increasing Δ‑Loss warns of process degradation before failure.

Frequently Asked Questions

Related Engineering Resources

References: IPC‑TM‑650‑2.5.5.13 (Δ‑Loss), IPC‑TM‑650‑2.5.5.7 (TDR), ISO 25178 (profilometry), IPC‑6018 Class 3 (thermal cycling), IPC‑TM‑650‑2.6.25 (CAF).