Répartiteur de puissance à microbandes de Wilkinson ou à cavité : lequel choisir pour un système DAS 5G ?

Both are power splitters. One is the size of your palm and costs a few bucks. The other is a chunk of metal that costs ten times more. Where does the money go? And is it worth it?

If you design DAS systems, you have probably wrestled with this question at some point.

With 5G rolling out, frequencies are climbing, power levels are going up, and PIM requirements are tightening. The old microstrip splitters that worked fine for indoor coverage start falling apart when you put them in macro sites or high‑power setups. On the other hand, cavity splitters perform great, but they are bulky and expensive – using them on a small project feels like overkill.

This article skips the textbook fluff and walks you through the real engineering differences between these two splitter types.

The Wilkinson Topology: A Quick Primer

Let us get one thing straight.

A Diviseur de puissance Wilkinson is a specific circuit topology – it uses two quarter‑wave transmission lines and one isolation resistor to split power equally. In a 50Ω system, the quarter‑wave lines have a characteristic impedance of 70.7Ω, and a 100Ω resistor bridges the two output ports. That resistor is key – when the loads on the two outputs are unbalanced, reflected power gets dissipated in the resistor instead of bouncing back and forth between ports.

A microstrip Wilkinson splitter implements this topology on a PCB – the microstrip lines are etched on a dielectric substrate, and the isolation resistor is soldered on. A cavity splitter, by contrast, uses a coaxial structure inside a machined metal housing – no PCB, no dielectric loss, and no isolation resistor.

Different principles, very different performance.

Insertion Loss: Cavity Splitters Win by a Wide Margin

Microstrip Wilkinson splitters have three sources of loss: the theoretical splitting loss (3dB for a 2‑way, 4.8dB for a 3‑way, 6dB for a 4‑way), dielectric loss from the PCB substrate (tan δ), and conductor loss from the microstrip traces. The excess loss above the theoretical value is typically 0.3‑0.5dB. That does not sound like much by itself, but in a DAS, splitters are cascaded – a 4‑way splitter feeding three 2‑way splitters adds up fast. By the time the signal reaches the antenna, a big chunk of it has already disappeared.

Séparateurs de cavité use air as the dielectric – no dielectric loss. The conductors are silver‑plated for excellent conductivity. Excess loss is typically <0.1dB. MS TELERF’s 790‑2700MHz cavity splitter series, for example, has a total loss (including the theoretical 3dB) of just 3.1‑3.3dB for 2‑way models – that means excess loss is only 0.1‑0.3dB.

In plain numbers: for the same 2‑way splitter, the cavity version delivers 0.2‑0.4dB more output power than the microstrip version. In a 5G DAS with a tight link budget, that 0.3dB can be the difference between “good signal” and “dropped calls at the edge.”

Power Handling: 100W vs. 200W+

Microstrip Wilkinson splitters are limited by the isolation resistor and the microstrip traces themselves. Crank up the power, and the resistor heats up, the traces heat up, dielectric loss increases, and performance drifts. They are usually fine up to 100W; beyond that, things get dicey. MS TELERF’s 2000‑8000MHz microstrip 4‑way splitter is rated at 30W – it is meant for lab use and low‑power indoor scenarios.

Séparateurs de cavité have an all‑metal structure with no resistor to heat up. Typical power handling starts at 200W and can go to 500W or more. MS TELERF’s 790‑2700MHz cavity splitters handle Puissance moyenne de 200 W, and the 600‑6000MHz cavity series handles 200‑300W CW.

Macro RRUs often put out tens to hundreds of watts, and when you combine multiple carriers, the total power is even higher – cavity splitters can take it; microstrip splitters would burn out.

Port Isolation: Microstrip Takes This One

This is one of the few areas where microstrip splitters come out ahead.

Microstrip Wilkinson splitters have that isolation resistor, which provides ≥20dB of isolation between output ports. If one output gets shorted or the antenna VSWR goes bad, most of the reflected power gets soaked up by the resistor and does not leak into the other output. That is a real advantage in multi‑antenna systems – one bad antenna does not drag down the whole system.

Séparateurs de cavité do not have an isolation resistor. If one output is open or shorted, reflected power goes straight back to the input. That means cavity splitters require a better VSWR on the output side – if your antennas and feed cables are not up to snuff, a cavity splitter can actually cause more trouble than a microstrip one.

But here is the catch: even without a resistor, cavity splitters can still achieve ≥25‑30dB isolation. That isolation comes from the electromagnetic field distribution inside the cavity, not from resistive dissipation. So the isolation of a cavity splitter is not actually worse – the mechanism is just different.

VSWR: Cavity Splitters Are Better

VSWR tells you how well the input impedance is matched.

Microstrip Wilkinson splitters typically have a VSWR of 1.3‑1.4. At higher frequencies, radiation loss and parasitic effects in the microstrip lines get worse, and VSWR degrades along with them.

Séparateurs de cavité can achieve ≤1.25 VSWR. MS TELERF’s 300‑4000MHz cavity splitter maintains ≤1.30 across the entire 380‑4000MHz range. For power amplifiers and transmitters, lower VSWR means less reflected power, less heat, and longer equipment life.

PIM Performance: Cavity Splitters Have a Natural Edge

PIM matters more and more in 5G systems. As noted earlier, using low‑PIM components is industry best practice for cellular applications.

Microstrip Wilkinson splitters have multiple PIM sources: non‑linearities in the PCB substrate, oxidation at solder joints, and parasitic effects from the isolation resistor. These non‑linearities get worse at higher frequencies.

Séparateurs de cavité are made with non‑magnetic materials and precision plating. No PCB, no resistor, no solder joints – fewer PIM sources to begin with. MS TELERF’s 600‑2700MHz cavity splitters achieve ≤‑153dBc (@2×43dBm), and some models go down to ‑160dBc or better.

For 5G networks that use 256‑QAM or even 1024‑QAM modulation, low PIM is a prerequisite for maintaining signal‑to‑noise ratio and hitting those high data rates.

At a Glance: Which One Fits Your Project?

Choose microstrip Wilkinson when:

• You are doing indoor coverage or low‑power applications under 100W
• Space is tight and you cannot fit a bulky cavity splitter
• Budget is a concern
• You need port‑to‑port isolation protection (so one bad output does not affect the others)

Choose cavity when:

• You are deploying macro sites or outdoor high‑power systems over 100W
• Every 0.1dB of loss matters and your link budget is tight
• Outdoor installation requires IP65 weatherproofing and long‑term durability
• You have strict PIM requirements to support high‑order modulation
• You want something that will run for years without maintenance

A simple rule of thumb: look at what is sitting next to the splitter. If it is a BBU in an equipment room, microstrip is probably fine. If it is an RRU on a tower, go with cavity.

MS TELERF offers both types. The microstrip Wilkinson line covers 500‑8000MHz – ideal for indoor DAS, Wi‑Fi 6E/7, and lab environments. The cavity splitter line covers 300‑6000MHz – built for macro sites, outdoor deployments, and high‑power DAS. Which one to pick? Let your power levels and installation environment decide.

If you have questions about splitter selection, or need detailed specs and test reports, feel free to reach out to the MS TELERF technical team.