High-Frequency PCB

The core differences lie in three aspects: substrate, performance, and process:

• Substrate: High-frequency PCBs use specialized high-frequency substrates such as Rogers and Taconic, with stable dielectric constant (Dk 2.2-30) and extremely low dissipation factor (Df ≤ 0.003); ordinary FR-4 substrates have large Dk fluctuations (3.8-4.5) and high Df (0.02-0.03), only suitable for scenarios below 5GHz.
• Performance: High-frequency PCBs support signal transmission from 10GHz to 110GHz, with insertion loss ＜0.002 dB/inch@10GHz; ordinary FR-4 has signal attenuation exceeding 50% above 10GHz, which cannot meet high-frequency requirements.
• Process: High-frequency PCBs require special processes such as impedance simulation, back drilling, and precision etching; ordinary PCBs focus on basic circuit processing with lower precision requirements.

The industry generally defines PCBs with an operating frequency ≥10GHz as high-frequency PCBs, but there is no absolutely unified standard:

• Mid-to-high frequency: 10GHz-30GHz (e.g., 5G base stations, wireless communication);
• Ultra-high frequency: 30GHz-100GHz (e.g., millimeter-wave radars, satellite communication);
• Extreme high frequency: ＞100GHz (e.g., terahertz equipment, cutting-edge scientific research instruments).

The core judgment basis is "the signal wavelength is equivalent to or smaller than the PCB line width" (when the frequency ≥10GHz, the signal wavelength is approximately 30mm, which is close to the order of conventional PCB line width, so electromagnetic radiation and transmission loss need to be considered).

Dk selection needs to match three key factors: signal frequency, impedance design, and cost budget:

• High-frequency scenarios (＞30GHz, e.g., millimeter-wave radars): Choose low Dk substrates (2.2-3.0, e.g., Rogers RO3003) to reduce signal attenuation;
• High-density wiring scenarios (e.g., 5G optical modules): Choose medium Dk substrates (3.5-4.5, e.g., Rogers RO4350B) to balance impedance control and wiring space;
• Cost-sensitive scenarios (e.g., industrial sensors): Choose the hybrid lamination scheme of FR-4 and high-frequency materials, using specialized substrates in local high-frequency areas to reduce overall costs.

Note: Dk needs to be comprehensively judged in combination with the temperature coefficient (TCDk). In extreme environments, substrates with TCDk＜50 ppm/℃ should be preferred.

Minimum Order Quantity (MOQ): Most manufacturers support orders starting from 1 piece, but the unit price for small batches of 1-5 pieces is higher (3-5 times higher than mass production);

Small-Batch Quality Assurance: Regular manufacturers adopt the "same-line production" model, using the same equipment and process standards for prototyping and mass production. Only need to note:

• Small batches require separate first article testing (to confirm key parameters such as impedance and loss);
• If subsequent mass production is required, it is recommended to provide small-batch samples as "reference parts" to ensure parameter consistency.

Three types of test reports are core requirements to avoid "false parameter marking":

• High-Frequency Performance Report: Insertion loss (S21) and return loss (S11) curves tested with Vector Network Analyzer (VNA), which need to mark the test frequency range (e.g., 10GHz-60GHz) and actual measured values (which must meet design requirements);
• Impedance Test Report: Full-board impedance scanning data, focusing on checking whether the impedance fluctuation of key networks (e.g., RF transmission lines) is within the tolerance range;
• Reliability Test Report: Special tests for application scenarios (e.g., automotive scenarios require a test report of 500 thermal cycles at -40℃~125℃, and military scenarios require a GJB 362B vibration test report).

Troubleshooting should follow the order of "material → design → process → use":

• Material Issues: Substrate Df exceeds the standard (e.g., actual Df=0.005＞design value 0.003), uneven copper thickness (affecting impedance);
• Design Issues: Transmission line is too long (matching circuit required if ＞100mm), line width and impedance mismatch (e.g., designed 50Ω but actual 60Ω);
• Process Issues: Excessively long back-drilled stubs (＞0.3mm), excessively high contact resistance of surface treatment (e.g., oxidation of ENIG nickel layer);
• Usage Issues: Operating temperature exceeds the substrate's tolerance range (e.g., using a substrate only supporting 125℃ in a 150℃ environment), external electromagnetic interference (no shielding treatment).

The core role of back drilling is to remove excess copper stubs in holes and reduce signal reflection:

• Applicable Scenarios: Multi-layer boards with signal frequency ＞15GHz and blind/buried hole design (e.g., 5G AAU module PCBs with more than 10 layers), which can reduce signal delay from 12ps/in to 5ps/in;
• Inapplicable Scenarios: Single/double-layer boards with frequency ＜15GHz (e.g., simple RF modules), the benefit of back drilling is lower than the cost (it will increase the process cost by 20%-30%);

Key Indicators: The back drilling depth tolerance must be controlled within ±0.1mm, and the stub length must be ＜0.2mm; otherwise, the optimization effect cannot be achieved.