FR-4 Datasheets Make Sense Only When You Read the Test Conditions

An FR-4 datasheet is useful only when each number stays attached to its material grade, measurement method, and test condition. A dielectric constant without a frequency, or a glass transition temperature without a method, is an incomplete input to a PCB design. The practical task is not to collect impressive values: it is to decide which values belong in a model, which need a purchasing limit, and which require verification on the finished board.

This guide uses an actual laminate product table as a reading exercise. It does not recommend that material for every design or treat a supplier's typical data as a production guarantee.

Start with a Material Identity, Not Just “FR-4”

FR-4 identifies a class of glass-reinforced epoxy laminate, not one universal recipe. Different resin systems and constructions can share the classification while offering different thermal and electrical behavior. Likewise, solder-mask color does not identify the underlying laminate.

Before copying any parameter, record the product name, source document or page, and revision if stated. For a live webpage with no document revision, record the access date instead. The example below uses the manufacturer's FR408HR product table as accessed on September 9, 2026; that date is not a datasheet revision.

Read One Real Table Across, Not Down

The Isola FR408HR product table labels the following as typical values. Its Tg entry is 190 °C by DSC, using IPC-TM-650 2.4.25C. Its decomposition entry is 360 °C by TGA at 5% weight loss. At 1 GHz, the table lists Dk 3.69 and Df 0.0091, identifying IPC-TM-650 2.5.5.9 for that frequency. Those are separate measurements, not four interchangeable measures of “better FR-4.”

The useful extraction is a complete record such as: FR408HR; Tg; 190 °C typical; DSC; IPC-TM-650 2.4.25C; manufacturer product table. If a spreadsheet leaves out “typical” or “DSC,” the shortened entry changes what the number means.

Figure 1: Preserve the complete data record before deciding whether it is suitable for simulation or acceptance.

Separate Thermal Events Before Comparing Temperatures

Glass transition, decomposition, brief soldering exposure, and long-term operation answer different questions. Tg describes a transition in polymer behavior; it is not a melting point or permission to operate continuously at that temperature. Td describes chemical degradation under a specified measurement procedure. Neither number alone qualifies a plated hole, a solder joint, or a complete assembly.

A useful review begins with the actual exposure. Is the concern a short assembly excursion, repeated temperature cycling, or years of operation near a local hot spot? Write that question beside the data request. It prevents a team from answering a long-term reliability question with a short-duration material test.

For example, a proposed drawing note saying “Tg ≥ 170 °C” is ambiguous if two suppliers report different methods. Improve the note by specifying the accepted method and the required supporting reliability evidence. IPC-TM-650 2.4.25 Revision D defines a DSC method and includes specimen preparation and conditioning. Preserve the revision actually cited by the material supplier—Revision C in the table above—rather than silently replacing it with a newer document. Do not average incompatible measurements to make candidates look comparable.

Dk and Df Belong to a Construction and a Frequency

Dk helps describe electric-field storage and affects impedance and propagation. Df represents dielectric dissipation; it does not include every loss mechanism in a PCB channel. Copper profile, geometry, discontinuities, and the chosen interconnect construction also matter.

The same product page includes construction-level prepreg data by glass style and resin content. That is a reason to request the proposed stackup's model inputs rather than lifting a single catalog headline. Ask the fabricator which dielectric thickness, construction, and design values it will use for the impedance calculation.

Consider a deliberately simplified check: for a fully dielectric-filled, uniform TEM line, propagation delay scales with the square root of relative permittivity. Changing the assumed value from 4.0 to 3.6 changes that factor from 2.00 to about 1.90, roughly 5%. This is an illustrative sensitivity calculation, not a prediction for a real microstrip or a claim about either material's tolerance. Its purpose is to show why an undocumented model input can consume timing margin.

Keep Typical Data Separate from Acceptance Limits

A typical value supports understanding and preliminary modeling. An acceptance criterion needs a defined limit, method, sampling basis, and agreement with the supplier. Writing a typical value into a purchase order with an invented tolerance does not create a validated requirement.

For each important property, classify the next action:

Model input: identify the construction and the simulation assumption.
Purchasing requirement: agree on a limit and how conformance is established.
Qualification item: identify the representative board, exposure, and pass/fail criterion.

The distinction is particularly important when reviewing alternatives. A proposed substitute with the same nominal Tg may still require new electrical inputs or a different process. Approve the relevant properties and construction, not a vaguely worded “equivalent FR-4.”

FAQ

Can I use a generic FR-4 dielectric constant for initial layout?

It can be an explicitly provisional estimate. Before releasing controlled-impedance or timing-sensitive routing, replace it with inputs appropriate to the proposed material construction and modeling method. Record the change so an early assumption does not silently become a production specification.

Does a higher Tg automatically make a laminate the better choice?

No. First establish which failure or performance limit needs improvement. A higher Tg does not, by itself, demonstrate lower channel loss, a better heat path, or an acceptable service life for the finished assembly.

Turn the Datasheet into a Traceable Design Input

Keep five things together: the material identity, property, value and unit, condition and method, and intended use. That small discipline makes a material table useful to design, sourcing, and quality teams without asking one number to prove something it cannot.