The decision between FR-4 and Rogers for your PCB is fundamentally about loss budget and impedance predictability. Engineers who default to Rogers for everything waste money, while those who force FR-4 into RF applications waste prototype cycles.
30-Second Decision
| Application | Frequency | Recommendation |
|---|---|---|
| IoT / BLE / WiFi 2.4 GHz | < 3 GHz | FR-4 (or hybrid) |
| 5G sub-6 GHz | 3-6 GHz | Hybrid stackup |
| WiFi 6E / UWB | 6-7 GHz | Rogers (RF layers minimum) |
| 5G mmWave / Radar | 24-77 GHz | Full Rogers or PTFE |
| Digital only (PCIe/DDR) | Baseband | FR-4 |
The Real Engineering Tradeoff
FR-4 is a composite of woven fiberglass and epoxy resin, optimized for mechanical reliability and cost. Its dielectric constant varies between 4.2 and 4.8 depending on resin content, glass weave style, frequency, and humidity. At 1 GHz, this variation is manageable because wavelengths are long relative to trace dimensions. At 10 GHz, where a quarter wavelength in FR-4 is approximately 4.5mm, that same +/-10% Dk uncertainty shifts your matching network center frequency by hundreds of MHz.
Rogers RO4350B uses a ceramic-filled hydrocarbon thermoset with tightly controlled filler particle distribution. The result is Dk = 3.48 +/-0.05 from DC to 40 GHz, with a dissipation factor of 0.0037 at 10 GHz.
Dielectric Loss: The Numbers
| Property | FR-4 (Standard) | FR-4 (Low-Loss) | Rogers RO4350B | Rogers RO3003 |
|---|---|---|---|---|
| Dk @ 10 GHz | 4.2 | 3.9 | 3.48 | 3.00 |
| Df @ 10 GHz | 0.020 | 0.010 | 0.0037 | 0.0013 |
| Loss (dB/cm @ 10 GHz) | 0.10 | 0.06 | 0.025 | 0.012 |
| Dk tolerance | +/-10% | +/-5% | +/-1.5% | +/-1.5% |
A 5cm microstrip trace at 10 GHz loses 0.50 dB in standard FR-4 versus 0.125 dB in RO4350B. For a 77 GHz automotive radar with multiple cascaded filter stages, that difference accumulates to the point where FR-4 simply cannot close the link budget.
The Hybrid Stackup Approach
The most cost-effective approach for mixed-signal boards is a hybrid stackup that places Rogers material only where needed:
- L1: Rogers RO4350B — RF signal, antenna, matching network
- L2: Copper — ground reference for L1 microstrip
- L3-L6: Standard FR-4 — digital signals, power
This costs roughly 80-100% more than all-FR-4, compared to 200-300% more for all-Rogers. The RF performance on L1 is identical to a pure Rogers board.
The critical manufacturing challenge is CTE mismatch at the Rogers/FR-4 boundary during thermal cycling. This requires controlled lamination pressure profiles and symmetric stackup construction.
Where FR-4 Actually Fails: Real Production Cases
Phase inconsistency across production lots. A 5.8 GHz bandpass filter designed on FR-4 with Dk=4.3 shipped 200 boards. The first lot measured center frequency at 5.78 GHz. The third lot, using a different FR-4 supplier's material with Dk=4.5, shifted to 5.62 GHz — a 160 MHz shift the receiver could not tolerate.
Insertion loss at temperature extremes. FR-4's dissipation factor increases significantly above 100°C. For automotive radar modules seeing 125°C junction temperatures, the additional loss from FR-4's thermal Df increase often violates the link budget.
Impedance predictability. When you simulate a 50-ohm microstrip, Rogers measures 50.2 ohms versus 47-53 ohms on FR-4. If your matching network has 0.5 dB return loss margin, FR-4 variation eats your entire budget.
Process Compatibility: Why RO4350B Wins
RO4350B was engineered for compatibility with standard FR-4 fabrication processes. Same drill bits, etch chemistry, solder mask, and press cycle (with adjusted temperature). This means hybrid boards run through production without special tooling — keeping lead times at 5-7 days versus 10-14 days for PTFE designs.
For applications above 40 GHz where RO4350B's Df becomes limiting, RO3003 or RO3006 (ceramic-filled PTFE) delivers Df=0.0013 at 10 GHz, but requires modified processing.
Decision Framework
- Below 1 GHz: FR-4, always sufficient
- 1-3 GHz: Depends on tolerance stack and loss margin
- Above 3 GHz: Rogers on RF layers is the professional choice
The hybrid stackup represents the best value proposition for modern wireless products. It delivers Rogers-grade RF performance where needed while keeping digital layer costs at FR-4 levels.
Originally published at AtlasPCB Engineering Blog. We specialize in hybrid FR-4/Rogers PCB fabrication for RF and 5G applications, with in-house impedance verification on every panel.
Further reading: Rogers 4350B Stackup Design Guide
Top comments (0)