A short connection from an RF chip to a nearby matching network may have little distributed line loss, while a long feed to an antenna on the same product can consume a meaningful part of the loss budget. Decide whether to upgrade the laminate by estimating the recoverable loss over the actual route. The operating frequency alone does not settle the choice.
This article develops a screening worksheet for that decision. The numerical example below is deliberately hypothetical: it is not measured performance of FR-4, RO4350B, RO4003C or another named material. Use characterized attenuation for your frequency, copper and stackup before making a procurement decision. For the broader comparison, see Rogers materials versus FR-4.
Allocate the loss budget to identifiable parts of the path
Write down the reference planes first. A requirement from a coaxial connector to an antenna feed point includes different structures from a requirement measured between the RF IC package and its matching network. Record which connector launches, vias, components and transmission lines sit between those planes. Otherwise a material comparison can accidentally attribute a connector improvement to the laminate.
For a preliminary budget, treat a reasonably matched path as a sum of positive insertion-loss contributions in dB: fixed structures plus attenuation per unit length multiplied by line length. This is a screening approximation. Cascaded mismatched networks require an S-parameter analysis because reflected waves interact; adding independently measured loss numbers can then misrepresent the assembled result.
Analog Devices’ AN-1370 layout example combines extracted trace S-parameters with component S-parameters and discusses keeping the antenna connection short. Its design uses an FR-4 substrate. That is a useful example of considering the entire RF path, rather than treating an RF operating band as an automatic laminate prescription.
Compare a short connection with a remote antenna feed
Assume, only for this worksheet, that two proposed constructions have total matched-line attenuation of 0.008 dB/mm and 0.003 dB/mm at the same selected frequency. Assume unchanged launch and component contributions total 0.30 dB, and the path allocation is 0.60 dB. These values are teaching inputs, not supplier specifications or recommended acceptance limits.
| Quantity | 10 mm connection | 80 mm feed |
|---|---|---|
| Construction A line loss | 0.080 dB | 0.640 dB |
| Construction B line loss | 0.030 dB | 0.240 dB |
| A total including fixed 0.30 dB | 0.380 dB | 0.940 dB |
| B total including fixed 0.30 dB | 0.330 dB | 0.540 dB |
| Estimated improvement | 0.050 dB | 0.400 dB |
| B margin against 0.60 dB allocation | 0.270 dB | 0.060 dB |
The short route meets this assumed allocation with either construction. The longer route changes from failing the assumed allocation to nominally meeting it. The remaining 0.060 dB margin on the longer route is small: connector repeatability, production spread and temperature behavior still need to fit inside it. A nominal calculation is not evidence of a robust release.
The sensitivity is easy to reuse. Here the attenuation difference is 0.005 dB/mm. Recovering an assumed 0.10 dB requires 20 mm of changed line, provided the fixed structures and impedance remain equivalent. If only 5 mm of a long route changes material, apply the benefit to those 5 mm, not the entire route.
Identify how much of the loss a material change can recover
A laminate change can alter both dielectric behavior and the geometry needed for the target impedance. Copper type, surface profile, dielectric height and trace width also matter. Keep them in the comparison record. If a new construction simultaneously uses smoother copper or a wider trace, report the combined construction benefit rather than claiming that the dielectric alone produced it.
Use the dielectric-loss calculator to explore one contribution under its stated assumptions. Do not enter its output as the complete measured transmission-line attenuation: it does not establish connector, conductor, radiation and discontinuity losses. A useful worksheet keeps separate columns for calculated, simulated and measured quantities.
For a candidate such as RO4350B, return to the Rogers RO4000 series data sheet. It distinguishes process and design dielectric constants and gives conditions for reported properties. A dielectric value detached from its test method is an inadequate input to a construction comparison. This worksheet intentionally does not substitute generic FR-4 properties for the actual baseline laminate.
Check routing changes before committing the stackup
Place the baseline and candidate routes side by side. Can the connector move closer to the RF section? Can an unnecessary layer transition disappear? Is a long detour caused by a component placement that can still change? Recalculate the path after those adjustments. A shorter route can save line loss without introducing a second dielectric system into fabrication.
A narrower trace is not, by itself, a loss-reduction strategy. At a fixed stackup it also changes impedance, and conductor loss may worsen. If mechanical constraints force a different trace width, solve the required geometry and reassess the construction together. Record any trade against isolation, return-path continuity, thermal needs or antenna placement.
If the short route is already within budget, a material upgrade may still be justified by phase stability, thermal behavior or another requirement. Document that requirement separately. Conversely, a lower-loss line does not fix an incorrectly tuned antenna or a poor connector transition.
Set a measured decision gate for the prototype
Prepare line coupons on both candidate constructions using the intended copper and geometry. Include representative launches and a suitable calibration or de-embedding structure. Analog Devices’ EVAL-ADRF5045 documentation illustrates the use of a through calibration line to estimate board loss. The appropriate calibration for your coupon depends on its topology and measurement reference planes.
- Measure the operating band, not just the center frequency, and record the calibration planes and instrument settings.
- Repeat connector attachment sufficiently to estimate whether the expected improvement exceeds measurement variation.
- Compare return loss as well as insertion loss, so that a change in mismatch is visible.
- Test the relevant temperature and environmental conditions selected for the product.
- Accept the construction against a documented margin that includes manufacturing variation. Do not treat the illustrative 0.60 dB allocation above as a universal limit.
Keep the results with the construction revision and prototype identifier. The material selection workflow helps carry that evidence into the final laminate decision, while the test and measurement application guide connects the coupon to the wider measurement setup.