Off-the-shelf 5G and IoT testing

Designing 5G and IoT RF systems through accurate simulation with OTS components and fast multistage impedance matching synthesis on a PCB.

5G, the upcoming 5th-generation wireless mobile network, will operate from 24 GHz and up to 95 GHz. It promises extremely high data rate wireless connection for applications such as 4K/8K ultra high definition TV streaming.

The Internet of Things (IoT) is another fast-growing application of wireless technology that is all around us — from personal gadgets to industrial sensors and freight tracking around the world. By 2020, over 50 billion IoT objects are forecasted to be in operation.

This means tremendous work and pressure for RF and microwave engineers to design and build 5G and IoT products quickly to compete for a share of the market.

5G 28 GHz RF system simulation

Designing and building RF systems to operate at 24 GHz and above is challenging due to the parasitics of interconnects, peripheral biasing and passive components, and the absence of simulation models of available system components.

Calculating with spreadsheets and then breadboarding with actual hardware is very costly in terms of time, instrumentation and effort for every ensuing iteration.

A more efficient approach in designing, prototyping, and realising RF systems in one pass is now possible and is validated in Figure 1, which shows the block diagram of a 5G system with 28 GHz RF input and two down conversion LOs at 22 GHz and 7 GHz to a 1 GHz IF. The block diagram is simulated in the Keysight Genesys Spectrasys system simulator with system blocks modelled as:

• X-Parameter for nonlinear circuit data;
• Sys-Parameter for system datasheet behavioural data with frequency, bias and temperature dependence;
• S-Parameter for linear circuit data;
• behavioural equation-based models.

Figure 1. 28 GHz 5th-generation RF receiver system with dual down conversion to 1 GHz IF. Simulated with Keysight Genesys RF system simulator.

RF system simulation technology has come a long way since the use of spreadsheets. Improvements in accuracy and diagnostic capabilities are significant. An example of these improvements includes identifying the component origin and frequency equation of nonlinear intermods; and which system blocks and their specs in the system line-up are contributing to performance degradation such as error vector magnitude (EVM), BER and ACPR under digital-modulated RF stimulus.

Figure 2 shows the budget analysis of EVM error versus system component line-up, which instantly identifies the major contributors of EVM degradation as LO phase noise and linearity of mixer and amplifiers. Clicking on the culprit components enables their offending specs to be adjusted in order to improve the system performance. This enables proper selection of system component line-up without over or under spec’ing them, to achieve the best performance with minimum cost.

Figure 2. Breakthrough budget analysis of error vector magnitude diagnoses which components in the RF system line-up are causing degradation to digital modulated RF signals in the system path, to prevent wasteful hardware prototype iterations.

Specifying the behavioural specs during design and then trying to find real components that have such specs during realisation is a common but inefficient approach, which inevitably results in multiple iterations.

Sys-parameters representing simulatable datasheets of real off-the-shelf components or X-parameters of measured nonlinear components can be used directly in RF system simulation, so that when the design is complete, parts have also been already specified and verified to work in the system chain.

The system is now ready for hardware implementation.

5G system realisation with OTS parts

Off-the-shelf parts from vendors such as Mini-circuits, Analog Devices, Qorvo, Marki and Avago implemented into modular tiles by X-Microwave were used in the realisation of the 28 GHz 5G RF receiver system as shown in Figure 3.

Figure 3. Hardware prototype of a 28 GHz receiver system using X-Blocks from X-microwave. What-You-Simulate-Is-What-You-Get without the discrepancies caused by interconnect parasitics or inaccurate system models.

Each modular tile, called an X-Block, includes all the biasing and peripheral passive components for the active device such as LOs, mixers and amplifiers. They are characterised by measured X-parameters or Sys-parameters at their co-planar interconnect reference planes for simulation to accurately model how they are being used in the actual system hardware. They are connected together by a flipped, co-planar laminate that spans the small gap between the X-blocks and is held down by compression, without soldering, to work reliably up to 67 GHz.

The 1.9 mm test launchers are also held down by compression so that the X-blocks can be re-used with no damage. When the prototype is finalised, the same composite layout can be used directly for production, since they are all built on the same laminate material.

When the system was measured, the agreement with the simulated result was unexpectedly close, as shown in Figure 4, and within the uncertainty error of the vector signal analyser.

Figure 4. Measured vs simulated error vector magnitude with different input RF power. Excellent correlation to within measurement uncertainty of test instruments.

Designing for the IoT

Multiple standards are emerging for IoT radios based on range of coverage, data bandwidth and operating frequencies. The IoT frequencies can be broadly divided into two categories: sub 1 GHz, and those above, namely around the 2.4 GHz and 5.8 GHz ISM (industrial, scientific and medical) bands.

From the perspective of designing IoT physical radio links that work at these frequency bands, the main focus should be on impedance matching the IoT chipset to the antenna. For a longer range, amplifiers may be inserted in between the chipset and the antenna.

Ideally, the impedance matching network has to be compact and economical to build. Multistage impedance matching over a broad bandwidth (30% or more) to complex frequency-dependent impedances such as an antenna, measured S-parameters of an IoT chipset or an unstable non-unilateral discrete transistor amplifier is extremely difficult and tedious using traditional Smith chart or benchtop cut-and-try techniques.

A more efficient and optimal approach is the use of automatic impedance matching synthesis, which employs multiple algorithms from simple L-sections to the real frequency technique, for addressing the increasingly difficult above-mentioned impedance matching problems.

Because synthesis can accomplish difficult simultaneous multistage matching in seconds with distributed and/or lumped networks, the IoT radio designer can quickly experiment with multiple matching topologies to select one that is most economical to build.

Figure 5 shows the result of three-stage, simultaneous matching of an antenna to a low-noise stabilised transistor amplifier circuit, followed by the measured S-parameter of a chipset power amplifier to achieve -20 dB return loss match from 2 to 3 GHz and a gain of 35 dB.

Figure 5. Impedance matching synthesis and microstrip layout of a three-stage matching network from 2 to 3 GHz to achieve -20 dB return loss and 35 dB gain is done in less than 1 hour.

The microstrip layout dimensions were also synthesised with the automatic insertion of discontinuities such as tees and open stubs. The entire process was completed within one hour.

RF systems for 5G and IoT applications can now be efficiently simulated, prototyped and produced with off-the-shelf system components thanks to breakthrough diagnostic capabilities which pinpoint wrongly spec’ed components in the system line-up.

Accurate X- and Sys-parameter simulation models of off-the-shelf RF system components enable what-you-simulate-is-what-you-get efficiency in going from design to prototype and production with no iterations.

Impedance matching synthesis replaces tedious manual design and optimisation with an instant selection of various suitable matching topologies for the most economical realisation.

Opening image courtesy Michael Kumm under CC BY 2.0