Rogers RO4835 vs RO4350B: Which Laminate for Your mmWave PCB?
Within the Rogers RO4000 series, the RO4350B and RO4835 represent two formulations that designers often conflate or select based on incomplete information. Having processed thousands of panels in both materials for automotive radar (77 GHz), 5G infrastructure (28 GHz), and satellite communications (24 GHz Ka-band), here's what actually matters in the selection.
The Quick Decision
| Application | Frequency | Recommendation | Why |
|---|---|---|---|
| WiFi 6E/7, sub-6 GHz 5G | <24 GHz | RO4350B | Lower cost, adequate loss, better availability |
| 5G mmWave (short paths <3cm) | 24-40 GHz | RO4350B | Loss difference negligible for short lines |
| 5G mmWave (long feed networks) | 24-40 GHz | RO4835 | 0.5+ dB advantage on 5cm+ paths |
| Automotive radar | 77 GHz | RO4835 | 25-30% lower loss tangent directly impacts range |
| Ka-band satellite | 24+ GHz | RO4835 | Every 0.1 dB matters in space-constrained budgets |
What's Actually Different Between Them
Both materials belong to Rogers' hydrocarbon ceramic family — woven glass reinforcement with thermoset resin filled with ceramic particles. The critical distinction is in resin formulation and resulting high-frequency loss:
- RO4350B: Dk = 3.48 at 10 GHz, loss tangent = 0.0037
- RO4835: Dk = 3.48 at 10 GHz, loss tangent = 0.0032
That 0.0005 difference looks trivial on a datasheet. It isn't trivial at mmWave frequencies.
Why the Difference Grows with Frequency
Published datasheet values are measured at 10 GHz. Real mmWave applications operate 3-8x higher. Based on our in-house split-post resonator testing and customer VNA characterization data:
At 28 GHz:
- RO4350B: effective Dk ≈ 3.52-3.55, loss tangent ≈ 0.0042-0.0048
- RO4835: effective Dk ≈ 3.50-3.53, loss tangent ≈ 0.0035-0.0040
At 77 GHz:
- RO4350B: loss tangent rises to ≈ 0.0055-0.0065
- RO4835: loss tangent ≈ 0.0040-0.0050
That's a 25-30% reduction in dielectric loss at 77 GHz. For an automotive radar antenna feed network with 5cm of propagation path, this translates to 0.5-0.75 dB less insertion loss — directly measurable in radar detection range and receive path SNR.
Fabrication: What's the Same and What's Not
From a PCB manufacturer's perspective, both process similarly. Unlike PTFE materials requiring sodium etching or plasma treatment, both RO4000 materials accept:
- Standard copper bonding treatments
- Conventional desmear and electroless copper
- Standard carbide drill tooling
- Normal multilayer lamination processes
The subtle difference: RO4835 exhibits approximately 5-8% less resin flow during lamination at equivalent pressures when bonding with Rogers 4450F bondply. This means lamination recipes validated for RO4350B hybrid stackups cannot be directly transferred to RO4835 without verification. Not a major challenge — but a process engineer needs to know which material they're running.
Glass weave effect: RO4835 typically uses 1080 glass fabric (finer weave), reducing periodic Dk variations for signals at certain angles relative to the weave. RO4350B is available in both 1080 and coarser 1674 fabric, offering more thickness options but potentially greater Dk variation in weave-sensitive applications.
Cost and Availability Reality
RO4350B benefits from higher production volume and longer market history:
- Standard thicknesses (0.254mm, 0.508mm, 0.762mm) in 1oz copper: 2-4 week lead time from stock
- Well-established supply chain, stable through recent disruptions
RO4835 costs 10-20% more with slightly longer procurement:
- Standard configurations: 3-6 week lead time
- Occasional allocation constraints during high-demand periods for automotive radar substrates
- 77 GHz autonomous driving market consuming increasing share of RO4835 production capacity
Hybrid Stackup Strategy
Most practical mmWave designs don't use Rogers for all layers. Cost optimization dictates hybrid stackups where Rogers appears only on layers demanding low-loss performance.
A common 8-layer hybrid for 77 GHz automotive radar:
- Layers 1-2: RO4835 (antenna and feed network)
- Layers 3-6: Megtron-6 (digital control)
- Layers 7-8: Standard FR-4 core (mechanical rigidity)
Bonding uses Rogers 4450F bondply specifically formulated for the interface. Our standard process validates each new hybrid combination through thermal shock testing and cross-sectional analysis before production release.
Common design pitfall: Specifying impedance calculations based solely on laminate Dk while ignoring bondply Dk at the Rogers-to-ground-plane interface. The bondply acts as a dielectric spacer with different Dk than the Rogers laminate, causing systematic impedance errors of 3-5 ohms if not accounted for in simulation.
Thermal and Power Considerations
| Parameter | RO4350B | RO4835 |
|---|---|---|
| Glass transition (Tg) | >280°C | >280°C |
| Thermal conductivity | 0.62 W/mK | 0.66 W/mK |
| CTE (X-Y) | ~14 ppm/°C | ~14 ppm/°C |
| CTE (Z-axis) | ~46 ppm/°C | ~46 ppm/°C |
| Moisture absorption | 0.06% | 0.06% |
The matched CTE values mean both materials can be combined in hybrid stackups without introducing mismatch stresses — useful for designs using different Rogers materials on different signal layers.
For power amplifier circuits: at 28 GHz with 2W RF power on 50-ohm microstrip (0.254mm thickness), dielectric heating in RO4350B generates ~18 mW/cm vs ~15 mW/cm for RO4835. Small individually, but meaningful in dense circuits with many parallel feed lines.
The Bottom Line
- Below 24 GHz: RO4350B. Lower cost, better availability, adequate performance.
- 24-40 GHz with short paths (<3cm): RO4350B still works. Save the budget.
- 24-40 GHz with long feed networks (>5cm): RO4835 delivers measurable improvement.
- 60 GHz and above (77 GHz radar, WiGig): RO4835 should be default selection. 10-20% material premium is insignificant relative to the performance gain.
Regardless of material choice, successful mmWave fabrication requires close collaboration between design team and fabricator during stackup definition. Material selection, layer registration, surface finish (immersion silver preferred over ENIG for RF surfaces), and impedance tolerances all interact and must be optimized as a system.
Based on fabrication data from thousands of panels processed in both RO4835 and RO4350B for automotive radar, 5G infrastructure, and satellite communications applications.
Further Reading:
- Rogers 4350B PCB Stackup Design Guide — Detailed stackup recommendations
- 5G Antenna PCB Fabrication: Via Fencing for 28 GHz Arrays — Manufacturing guide for 5G phased arrays
- FR-4 vs Rogers PCB: When to Switch Materials — The fundamental material decision framework
Designing a mmWave PCB? AtlasPCB maintains Rogers RO4835 and RO4350B inventory for rapid-turn RF prototypes. Hybrid stackup design support included with every RF quotation.
Top comments (0)