Two decades ago, a spectrum analyzer was a discrete piece of test equipment that sat on your table in the lab. It had knobs, buttons, and a cathode-ray tube (CRT) display reporting the power versus frequency of whatever signal was connected to the input. Today, the spectrum analyzer is no longer just a collection of RF front-end hardware designed to condition and measure RF signals. It’s a PC.
As test equipment has evolved from discreet, single-function boxes to modular and highly flexible PC-based technology, so has the ability to utilize these tools to improve microwave design. As a result, modular, PC-based test equipment offers a wide range of functionality beyond just taking measurements—and these features are often highly valuable to microwave circuit and system designers.
Here, we’ll walk through two specific applications of how engineers can use PC-based test equipment in conjunction with microwave design software to improve their design process. By explaining how engineers are using this new technology, I hope to trigger new ideas in your mind of how you can design microwave systems more efficiently.
Application #1: Model Extraction
Imagine you’re building a wireless receiver designed to communicate with a satellite over a microwave link. In this design, you might choose to build some of the “easier” components yourself, but you’ll also choose to buy some components off the shelf as well.
As a sophisticated system designer, you’ll use a system simulation software tool (like AWR’s Visual System Simulator) to simulate the circuit’s performance. In your simulation, you might perform a cascade noise figure analysis or spur analysis, or you might even attempt to estimate the bit error rate (BER) of the modulated signal you plan to use.
In this application, the accuracy of your simulation is only as accurate your ability to model the performance of each component in the system. There’s little concern for the components you’ve designed natively in the simulation environment.
But for components you’ve elected to buy off-the-shelf, metrics such as the component’s 1-dB compression point and third-order intercept point (IP3) only tell a partial story of how the component will behave in your application. Wouldn’t it be nice if you could somehow measure your component’s performance and use that information in your simulation?
Enter model extraction, which measures a component or subsystem’s behavior with test equipment (in this case RF test equipment) and then incorporates that measurement information in your simulation of the entire system. With microwave components, one of the most common forms of model extraction that engineers employ is the measurement of AM-AM and AM-PM responses of an active device.
For active RF components such as mixers and amplifiers, AM-AM represents the relationship between the input and output power of a signal. Conversely, AM-PM represents the output phase of a signal as a function of input power. Using these characteristics, engineers can model the performance of active microwave components within system simulation software tools.
To measure AM-AM and AM-PM, an RF vector signal generator and RF vector signal analyzer can be automated to produce and acquire the appropriate signal. On the generation side, a ramp signal is generally loaded onto the arbitrary waveform generator and generated along with a digital output trigger and the 10-MHz reference clock. On the analyzer side, the RF signal analyzer begins acquisition upon receiving the start trigger and acquires the signal as IQ samples (Fig. 1).
Once the “ramp” signal has been acquired, some simple math can be employed to calculate the relationship between the input and output phase and power of the signal through the microwave components. Generally, information such as AM-AM and AM-PM is shared with system simulation software through a basic text file, which then can be imported into the simulation environment.
In Figure 2, we observe that the AM-AM and AM-PM results are referenced in the system design—and the measurement-based model is used to more accurately predict system performance. As Figure 2 suggests, given the use of real-world data to model the power amplifier, the performance of an example receiver now can be more accurately modeled in the simulation environment.
2. System simulation uses extracted AM-AM and AM-PM data.
AM-AM and AM-PM extraction is just one example of how test equipment is being used in conjunction with EDA software to improve the accuracy of system simulation. As you might expect, engineers are continually experimenting with new ways of capturing data to predict system performance. Going forward, we’ll continue to see more sophisticated measurement and modeling techniques for microwave system simulation.
Application #2: HIL Simulations
A second but closely related application to model extraction is hardware-in-the-loop (HIL) simulation. In Application #1, suppose that we decide that the power amplifier (PA) in question is too difficult to model. After all, techniques such as AM-AM and AM-PM extraction are most appropriate for narrowband signals. Moreover, other component characteristics such as memory effects as a function of temperature are dynamic.
If our signal in question was a wideband signal or if we wanted to introduce other variables, we might consider alternate approaches to introducing measured data into the simulation environment. With HIL simulation, a combination of mathematical models and actual hardware is used to simulate how a given system will perform. In this case, the system simulation software interfaces to the component in question through RF test equipment.
In an RF system such as the receiver mentioned earlier, simulated components produce an output signal as a function of input through mathematical models. But for the portion of the circuit simulated through hardware-in-the-loop, a vector signal generator produces the input signal and a vector signal analyzer produces the output signal.
Implementation of an HIL setup requires our system simulation software environment to interface to RF instrumentation in real time. Figure 3 illustrates an example implementation of an HIL setup using AWR’s Visual System Simulator (VSS) software. The connectivity between the system diagram and hardware occurs through the LabVIEW element. LabVIEW is a classic tool for instrument control hardware (among other things). It allows IQ samples to be passed between the system diagram and hardware controlled through an instrument driver.
3. LabVIEW serves as the interface between EDA software and instrumentation.
It’s interesting to note in Figure 3 that PCI-based instrumentation architectures such as PXI are ideal for applications such as HIL. Because PXI enables extremely fast data transfer to and from the instrument through a high-speed PCI or PCI Express data bus, it allows the entire simulation to run more quickly.
As one might expect, HIL simulation enables engineers to use a combination of real and simulated (or modeled) components to simulate a system like an RF receiver. This is especially useful in scenarios where a component is particularly difficult to model or when a subsystem’s performance is variable according to environmental conditions. In this application, it’s easy to imagine an engineer placing the real amplifier in a temperature chamber to determine the effect of thermal characteristics on overall system performance.
Parting Thoughts
Twenty years ago, the process of getting measured data into a PC was so slow that model extraction and HIL simulation might have seemed too difficult to implement. Today, PC-based instruments are designed with faster data buses, digital sampling technologies, and high-performance processing power. As a result, this evolution of software-defined instrumentation is enabling engineers to use measured data to improve RF system simulations.
In fact, model extraction and HIL simulation are just the start. Going forward, growing connectivity between the EDA environment and programming languages will continue to enable new applications that use instrumentation in the design flow. In the long term, expect to see increasing connectivity between design and test—a technology that will ultimately increase the productivity of microwave design engineers.