A laminate comparison is only meaningful when the values describe comparable tests and constructions. Before sorting HDI PCB materials by Dk, Df or Tg, record the measurement method, frequency, conditioning and exact material form. A missing condition is an open question, not a value you can safely assume.
This guide builds a comparison worksheet rather than naming a universal best laminate. Start with the HDI laminate-selection guide for construction requirements, then use the process below to separate documented evidence from convenient headline numbers.
Capture the identity of each data point
Save the original manufacturer document and its revision date. Record whether a number applies to laminate, cured prepreg or a named construction within a family. Include units and any stated qualifier: typical, minimum, maximum or guaranteed specification. A typical property value should not silently become a purchase acceptance limit.
Use one worksheet row per property and condition. If a supplier gives several frequencies, preserve the separate rows. If a required condition is absent, label it unknown and request clarification. This makes gaps visible before a simulation model or purchasing drawing depends on them.
| Property group | Conditions to retain | Decision supported |
|---|---|---|
| Dk and Df | Frequency, test method, relevant direction and construction | Suitable inputs for the chosen electrical model |
| Tg | Measurement method and reported definition | Comparison of the stated thermal transition evidence |
| Expansion | Axis, temperature interval and units | Review of expansion through the relevant thermal history |
| Delamination or thermal stress | Test temperature, duration, specimen and criterion | Understanding the test actually performed |
| Moisture and adhesion | Conditioning, copper type and test procedure | Process and environment questions to verify |
Do not treat every Dk value as the same model input
Rogers explains in its discussion of material dielectric constant that different measurement configurations can yield different values for the same material. Its Design Dk approach uses circuit measurements intended to support design modeling. That does not mean every manufacturer’s typical Dk can be relabeled Design Dk.
Ask which value is appropriate for the field solver, frequency range and construction you will actually use. Keep the source of that recommendation with the model. If two data sheets use different methods, do not invent a universal correction factor between them. A supplier-supported model or a suitable measurement comparison is stronger evidence than an unexplained average.
Also separate material loss from complete channel loss. Your finished traces, copper and geometry remain part of the electrical problem. A lower listed Df is a reason to investigate a candidate, not proof that a particular assembled link will meet its loss budget.
Use a deliberately incomplete example to expose the trap
Imagine candidate A lists Df at one frequency and candidate B lists it at a substantially different frequency, with no common test context. Ranking them numerically answers the wrong question: it compares two reported observations, not necessarily performance under the same conditions.
The useful action is to add columns for method, frequency, construction and intended model use. Mark the comparison unresolved until you obtain compatible evidence. Do not fill a blank cell with a value from another grade merely because it belongs to the same product family. In a review meeting, that unresolved row is more valuable than a falsely precise winner.
For a quick dimensional check, keep percentage and fractional loss representations distinct and retain the original units for expansion and adhesion. Unit conversion can be verified arithmetically; differences in test method cannot be fixed by unit conversion alone.
Read thermal properties as separate tests
The Isola 370HR revision C data sheet is a useful worked document: it identifies Tg by DSC, decomposition by TGA with a stated weight-loss criterion, and delamination testing by TMA. These are different measurements. The document also distinguishes expansion before and after Tg and total expansion over a temperature interval.
This article does not transfer that material’s numbers to another grade or prescribe it for every HDI board. The lesson is to preserve what was measured. A Tg comparison is not a complete assessment of assembly survivability, and a reported decomposition temperature is not a recommended continuous operating temperature.
Describe the actual manufacturing and assembly exposure to the fabricator: sequential laminations, reflow and any expected rework. Ask what evidence applies to the proposed stack and via structure. The high-Tg FR-4 discussion helps frame the questions, but the released material system still needs construction-specific review.
Bring the comparison back to the available construction
A promising family name may not identify the glass style, resin content, copper foil or pressed dielectric your supplier can provide. Ask for the exact proposed bill of materials and construction. Keep the electrical model tied to those selections rather than to a generic trade name on a summary chart.
When a substitute is offered, record what changed and which calculations or process evidence must be revisited. An unchanged nominal board thickness does not establish unchanged electrical or thermal behavior. Require an explicit approval path for substitutions in the fabrication documentation, including who resolves a disagreement between the model and the quoted construction.
Release a comparison with an evidence status
- Attach manufacturer documents and revision dates.
- Retain methods, units, frequencies and specimen conditions.
- Distinguish typical properties from acceptance requirements.
- Identify the exact stocked construction proposed by the fabricator.
- Mark unsupported comparisons and request the missing evidence.
- Approve model inputs and manufacturing requirements as separate decisions.
Use the PCB material selector to organize the initial requirements and the materials library for related topics. For a material and stackup review, send the comparison worksheet with its unresolved rows intact. A traceable decision is more useful than a ranking whose test conditions have disappeared.
Sources reviewed 9 October 2026. The cited documents support interpretation of test conditions; this is not a material qualification or a supplier availability guarantee.