Promo Figure 1

Building Blocks for 28-GHz Small Cells

May 13, 2020
One of the more significant 5G innovations is the ability to operate in the millimeter-wave spectrum. Support for high-frequency bands opens underused spectrum to bolster 5G network capacity and provides opportunity to leverage small-cell technology.

Built to support 5G networks, 28-GHz small cells are compact, lightweight devices mounted on “street furniture” like lampposts or telephone poles. Immediately, this application presents design concerns. These devices need to be small and lightweight to accommodate the installation process. Also, they must withstand temperature extremes, yet they can’t always afford the large heat fins often used to manage temperature due to size and weight restrictions. 28-GHz small cells also need to be extremely efficient to be an economically viable approach to building an access network.

A 256-element, 28-GHz array, for example, might be used in a g Node B (gNb), higher-power 28-GHz radio (Fig. 1). Many components are involved in creating a base station; they include antenna assembly, feed network, beamforming ICs, RF filters, RF switches, RF amplifier, mixer, and transceiver. A large macrocell might have the space for all of these components with separate cards for some of the switches, amplifiers, and filters, but as the radio gets smaller, the task is to compress the design without compromising on performance.

In a 28-GHz small cell, reducing the total number of antenna elements helps to reduce cost and complexity to transition from the 256-element array in the gNb above into something more like a 64-element tile. Ideally, all RF components fit into such a tile or card. RF component dimensions are constrained by the size of the array, and array size is dictated by wavelength. To reduce loss in RF components, shrinking the footprint to fit within a single tile is particularly important. The challenge is to do this while optimizing RF performance.

Figure 2 shows a breakdown of how components might fit on a tile:

  • Step 1 moves the element array to the front of the board.
  • Step 2 moves the beamforming ICs onto the rear of the board, shown with a single beamforming chip that connects to the four different element arrays.
  • Step 3 displays feed networks for a Wilkinson power divider. It’s important to shrink these feed networks as much as possible without compromising on performance.
  • Steps 4, 5, and 6 show filter placement examples. To identify potential filtering options, consider what level of performance is needed and how much space there is to implement components.

Filtering in the 28-GHz Small Cell

When implementing frequency-control components such as filters, size and performance remain critical factors. Among the key considerations for implementation are:

  • Size: In dense tiles, consider the amount of board area available for passive components; otherwise, move components off-board to a separate card, increasing loss.
  • Performance: Guard banding is influenced by such characteristics of RF components as tolerance and repeatability, as well as temperature stability at higher temperatures

In general, performance affects size—the better the performance, the larger the overall component size. Figure 3 shows four different Knowles Precision Devices (KPD) filter offerings at the 28-GHz frequency. The dark-green device is the largest component with the highest selectivity. The other three designs employ KPD second-generation components with higher-permittivity dielectric materials. Adjusting filter performance can reduce the overall size and performance of the system at a reduced cost in manufacture.

Given the small-cell thermal concerns, maintaining temperature stability over a wide range is imperative. Temperature stability impacts guard bands, potentially affecting spectral efficiency.1 Tolerance is also a key driver. If tolerance isn’t considered in manufacture, it can impact the yields of the overall system and further increase the need for guard banding, taking up useful spectrum.2

Building Feed Networks

As the RF moves from the transceiver and fans out into the beamforming network ICs, there must be some power dividers on the board to split the signal. Feed networks should be implemented between the beamforming ICs, further complicating the component (see Table 1 for key considerations).

Considering the 1:4 divider/combiner components of the network in our small cell, there are several different approaches for implementation (Table 2 and Figure 4). We can build these Wilkinson power dividers in the PCB; utilize three 1:2 Wilkinsons; or take an all-in-one approach with a single surface-mount 1:4 Wilkinson divider.

Using a fully integrated design featuring thin-film technology and a high-permittivity dielectric gives the designer the opportunity to shrink the overall size and integrate the resistor, reducing variation from resistor tolerances and improving on the overall RF performance.

By managing the thermal properties of ceramic materials (PG above), elements of the feed layer also act as heat-management components. They’re temperature-stable by design and capable of operating across a range of temperatures inside the small cell. In addition, the high thermal conductivity of the ceramic material aids in heat distribution in the antenna tile. This second feature has benefits for mechanical complexity (i.e., cost of assembly) and reliability.

Performance

Effective isotropic radiated power (EIRP) is an important characteristic when discussing 5G antenna system performance. In the equation shown in Figure 5 for EIRP in a beamforming antenna, the engineer can use three factors to optimize EIRP, all of which have an impact on the overall system:

  • Element TX Power (P_out) is a function of the RF path loss and component gain; we can balance the cost of adding gain by minimizing loss. P_out affects system dc power dissipation and cost.
  • Number of elements (N_elem) is the contribution from the array. N_elem can affect PCB area and cost.
  • Individual Element Gain (Elem_gain) is the gain of an individual element; this value is driven by how effectively the antenna radiates. Elem_gain affects element complexity and cost.

In small-cell applications, N_elem and Elem_gain might suffer in the name of cost because of their potential impact on cost and complexity. To influence EIRP, P_out can be optimized. To do this cost-effectively, we look at loss in the passives we have been discussing.

Innovation in active components drives design improvements in mmWave small-cell design. However, the impact of careful passive-element selection shouldn’t be underestimated—thoughtful implementation can reduce both the size and overall cost of the system. The right passive components lay the groundwork for a successful small cell through better efficiency at every level of design.

Peter Matthews is Senior Technical Marketing Manager at Knowles Precision Devices.

References

1. “Spectral Efficiency and mmWave Bandpass Filter Temperature Stability” 

2. “Millimeter Wave Filter Manufacturing: Tolerance and Size