What to know about electromechanical interconnects for 5G systems

The interconnect challenges of delivering 5G to the device

It may feel like 5G is everywhere, but the consensus is that it will still take several years for the latest cellular network to reach 4G/LTE levels of availability. This fits with the historic trend for a new network every 10 years, and we can expect 6G systems to start appearing around 2030. It also makes anyone now targeting 5G a “trailing-edge early-adopter,” if that’s not too much of a contradiction.

That’s a good thing. The early adopters are already out there but haven’t hit anything like mainstream yet. Now is a good time to develop the first generation of your 5G solution. Manufacturers can expect to develop several generations to span that decade before 6G starts to take over.

Any first-generation product is the ideal place to establish design excellence. It can be done while encountering and overcoming the challenges associated with new technologies. For 5G, the main challenge is also the biggest feature: performance. Everything goes faster in 5G. Designers and systems engineers will immediately understand that for them, fast isn’t always fun.

Signal integrity is a big issue for any high-frequency signal, whether it is traveling through a conductor or through the air. All the headline benefits of 5G — such as highly dense traffic, more efficient spectrum utilization, more bandwidth and lower latency — need to be delivered at every point in the network. That includes at the board level, between boards and between components on those boards. This is where the Pulse Amplitude Modulation 4-level (PAM-4) scheme is so important. It will deliver speeds of 56 Gbps and 112 Gbps, much higher than the non-return to zero (NRZ) modulation scheme it is displacing. While NRZ will still be used for some time, PAM-4 is going to be enabling for 5G.

The key capabilities of 5G networks as defined in ITU-R M.2083 (Source: ETSI)

The European Telecommunications Standards Institute (ETSI) defines eight aspects of speed, latency and efficiency that must be met by 5G systems, as shown below. ETSI puts increased emphasis on the three main use cases identified by the International Telecommunication Union, Radiocommunication Sector (ITU-R): enhanced mobile broadband (eMBB), massive machine-type communications (eMTc), and ultrareliable and low latency communications (URLLC). To meet the demands of these use cases, the industry is utilizing new techniques and methods, such as millimetre wave transmission, smaller and more cells, beamforming and MIMO antenna technology. The nature of MIMO increases the antenna density in a transmitter, pushing the antenna array density up to as much as 256 elements.

From a device manufacturer’s point of view, this points toward devices that are physically smaller and more power efficient, yet they are handling many more signal paths. Those physical connections need signal integrity levels that can support PAM-4 transfer rates. All vertical markets are expected to want 5G, predominantly because of its low latency. They will rely on systems that can support that level of throughput at the component level.

The new network

Part of the reason the move from 4G/LTE to 5G is so significant is because it redefines the network topology. Previous generations have been built on legacy systems with a continuation of technology and methodology. For 5G, those legacy systems have been put aside. This can be characterized by the New Radio standard adopted by 5G. In reality, every aspect of the network has been redefined. Of course, it must achieve this while still relying to some extent on existing technologies, such as radio frequency connectors able to carry the mmWave signals.

The increased openness of 5G has seen fundamental changes in network structure. The radio access network (RAN) of 4G included a baseband unit and remote radio heads. In 5G, this has become the fronthaul network comprising centralized units, distributed units, radio units and MIMO antennas.

This is where the new remote radio units, active antenna units and baseband units will be deployed. The fronthaul will connect to the core network, while the actual devices found in the use cases outlined earlier will connect through the fronthaul network using MIMO antennas.

While optical interconnect will be used in many places throughout the network, copper interconnect still has an important role to play in the new 5G topology.

Molex Global Product Manager Mike Hansen explained that the main demand for copper interconnect solutions is for routing between boards and, interestingly, across boards. Using a cable assembly with twinax allows high-speed signals to be routed while avoiding the losses associated with PCB traces.

Hansen explained that 5G has created a shift, with the development of active antenna units (AAUs). These AAUs feature massive MIMO architectures and a lot of processing, all in a very small area. High-density copper interconnects are essential in the system architecture.

Routing high-speed 5G signals

At the board level, the PCB is becoming a major hurdle for high-speed signals. While moving to optical interconnect takes some of that pain away, at some point electrical signals will still need to interface to integrated circuits. This is where advanced interconnect solutions can provide higher signal integrity with the low insertion losses needed for sensitive differential signals.

Improved edge connectors provide the density needed to squeeze differential pairs and single-ended signals alongside power on a single, highly compact connector. This is achieved while still avoiding signal integrity issues

Instead of being routed across congested and lossy PCBs, signals are now routed from one side of a PCB to the other, or directly from the I/O to the integrated circuit, using assemblies made with twin coaxial cable. These so-called bypass cable assemblies avoid the losses associated with conventional PCBs without the need to move into and out of the optical domain, keeping the cost and latency to a minimum.

To support the bandwidths involved with 5G networks, operators are using 56 Gbps PAM-4 signalling and, where possible, 112 Gbps. Routing signals of this speed relies on careful impedance matching at the connector level. Solutions for routing PAM-4 signals include the NearStack 100 Ohm and 85 Ohm families from Molex. The NearStack PCIe connectors support 32 Gbps with NRZ encoding.

More radical solutions, like the NearStack On-the-Substrate system, use direct-to-chip substrate twinax connections, putting the point of contact right on the surface of the ASIC. This system supports 56 Gbps and 112 Gbps PAM-4 connections today.

Twinax cable assemblies allow sensitive high-speed differential signals to be routed directly from the I/O to an ASIC without incurring the losses associated with FR4 tracks in PCBs. (Source: Molex)

For more information, check out the AcceleRate HD Ultra-Dense Multi-Row Mezzanine Strips from Samtec, the high-frequency RF connectors from Bel and the twinax cable assemblies mentioned in the article from Molex. 

Dive deeper into 5G  

Follow Avnet Silica on LinkedIn