Accelerate System-Level Analysis With Upfront RF Planning

April 23, 2009
High-frequency technology didn’t earn its reputation as black magic without reason. Unlike lowfrequency circuits, microwave circuits don’t behave in a totally predictable way

1. The traditional Excel spreadsheet approach to system-level will include analysis component mismatches and phase noise.

2. Here, the Visual System Simulator (VSS) is co-simulated with the Microwave Office design environment.

3. A WiMAX test bench in VSS has all of the required tools in one place.

4. VSS’s digital pre-distortion uses TestWave software coupled with Rohde & Schwarz instruments.

5. This RF Budget Analysis screen is for an RF gain control circuit with automatic gain loop (AGC).

6. In this example, the RF Budget Analysis screen shows an RF link, cascaded noise figure measurements, and cascaded available gain at the image frequency.

7. The RF Inspector is applied to a converter circuit that has test points (TPs), combiners (COMBs) and Butterworth bandpass filters (BPFBs).

Joel Kirshman

High-frequency technology didn’t earn its reputation as black magic without reason. Unlike lowfrequency circuits, microwave circuits don’t behave in a totally predictable way. Consequently, “tweaking” has been an accepted mainstay of the microwave design approach/flow.

Fortunately, high-frequency design tools have dramatically improved so that tweaking of prototype circuits is much less common, and today’s engineer has powerful tools that can make sense of the black magic. One such tool is AWR’s Visual System Simulator (VSS) design environment, a software suite for design and optimisation of communications systems.

VSS is a natural evolution of the way most designers accomplish system-level analysis, which is accomplished with software such as Microsoft Excel (Fig. 1). It accounts for the most important factors— component mismatches and phase noise—that affect system performance. It also allows circuit designs or S-parameter files to be imported into them, offers a library of components, and can pinpoint the source of a spurious response.

While spreadsheets can be powerful and account for numerous system-level effects, they don’t fully exploit nor correctly calculate all the effects of a system (for example, noise) that determine a receiver’s RF link quality inclusive of cascaded noise figure and image noise. This source of noise is typically incurred in the downconversion process, which produces a desired (centre) signal from the difference between a mixer’s local oscillator and input frequencies.

The centre frequency will also incur a noise penalty from a higher image frequency that places a signal at the centre frequency after downconversion. The amount of added image noise depends on how much of the image frequency band is present at the input of the mixer. Failure to account for it can result in significantly degraded receiver performance.

Image noise can be reduced through filtering. However, if the system-level analysis is performed solely with a spreadsheet, the effect of image noise can only be assumed, since it can’t be accurately calculated. As a result, a designer can either ignore it and hope for the best or overcompensate for its effect based on an assumed noise value. Either of these solutions is imprecise at best.

That’s simply not acceptable in today’s design environment, where systems use complex protocols such as orthogonal frequency division multiplexing (OFDM), and the “timeto- market” window continues to shrink. In short, performance-killing system-level effects must be dealt with early in the design, as quickly and painlessly as possible.

VSS was created for the express purpose of advanced system-level planning, enabling designers to account for variables that aren’t feasible within a spreadsheet-based approach. It works seamlessly with the Microwave Office high-frequency design suite, which provides a complete solution from concept through verification (Fig. 2). The “heavy-lifting” is done by VSS, so that designers can devote their efforts to creating microwave products rather than writing software to evaluate them.

Both the baseband and microwave portions of a design can be evaluated together. As a result, all possible component interactions can be considered.

Test benches integrated into VSS are dedicated to specific wireless standards, such as GSM, EDGE, HSPA, DVB-H, DVB-T, W-CDMA, IEEE-802.11a/b/g, and WiMAX (Fig. 3). They include standard-specific measurement criteria that must be applied to ensure compliance with an applicable standard.

System-level variables are dealt with early in the design. This will often dramatically reduce and usually eliminate the need for any kind of circuit rework.

When used in conjunction with AWR’s TestWave software, circuits, subsystems, and even complete systems can be evaluated using actual standards-based waveforms generated by signal generators, and evaluated by vector signal analysers, spectrum analysers, vector network analysers, and other external instruments. Complex test routines can be created, configured, and controlled by the TestWave software, and the results can be compared with a simulation performed in VSS (Fig. 4).

With VSS, designers can begin at the behavioural level, and subsequently progress to the component level using Microwave Office or Analog Office software. Then system performance can be verified using actual measurements:

RF Budget Analysis: A main component of VSS, it enables standard cascaded RF measurements, such as gain, noise figure, and third-order intercept (including image noise), to be performed throughout the project. • RF Inspector: Helps users identify the cause of impairments like intermodulation products or spurious signals anywhere in the signal chain. • TestWave: This tool makes it possible to verify system performance by stimulating and evaluating the performance of devices using external test equipment to reveal weaknesses.

The RF link with automatic gain loop (AGC) illustrates the benefits of the RF Budget Analysis tool (Fig. 5). The link consists of two amplifiers, a filter, and a variable gain amplifier (VGA). The amplifier model in VSS depends on a data file that accounts for the frequencydependency of common amplifier parameters, including input and output reflection coefficients. The VGA accounts for frequencydependency, hysteresis, and S-parameters. Frequency-dependent settings are interpolated for frequencies between the smallest and largest frequencies.

The models also include saturation shaping of the input signal, and, if desired, designers can work at the highest level of abstraction using RF models with parameter windows for P1dB, IP3, noise figure, gain conversion, and other parameters.

The VSS simulation is set to sweep through several frequency points and power levels. For each frequency, the cascaded noise figure, gain, and output power versus input power are monitored.

In the case of noise figure versus frequency, each input power level is represented by a different colour trace and the X-axis is defined as frequency. The power input versus power output graph in the figure shows the region of operation in which relatively stable gain can be achieved. In short, VSS lets the subsystem designer evaluate the AGC loop under realistic conditions, and can do so with eight mouse clicks in less than five minutes. This is obviously a dramatic improvement over what even the best spreadsheet can do.

The upper left area of Figure 6 shows a VSS project that consists of two RF links, each composed of a bandpass filter, low-noise amplifier, image-noise rejection filter, quadrature mixer, low-pass filter, and 50Ω terminating load. The mismatch between components is simulated in both links. The difference between the two links is the order of the image noise rejection filter—one employs a third-order filter and the other a seventh-order filter.

The seventh-order filter has a sharper cut-off in the stop band than the third-order filter. The cascaded noise figure graph of the RF link (lower left) shows that a lower noise figure can be achieved using a seventh-order filter rather than a third-order filter. The available cascaded gain at the image frequency is shown at the lower right. The reduction in cascaded gain after the mixer is caused by the frequency conversion.

To maintain the overall cascaded noise figure, the designer can either keep the third-order filter (which is less expensive to implement) and optimise the noise figure of the low-noise amplifier to provide the same effect as using a seventh-order filter. The available cascaded gain at the image frequency can be monitored as well. The graphical interface of VSS simultaneously shows both optimisation and yield analysis.

RF Inspector is a system-level, frequency- domain circuit simulator. It solves for voltages and currents at each RF node for each generated spectral component and includes the effects of conversions, harmonics, intermodulation, and mixer leakage (LO-to-IF, IF-to-RF, and RF-to-IF). The RF link can be simulated with continuous-wave (CW) or modulated signals.

Designers can monitor the full spectrum, only the spectrum of the signal, the spectrum of the distortion products, and the spectral content of any node after simulation. To determine the individual contributions to a particular tone, designers can simply double-click on a tone to open up the RF Inspector dialog box. The box contains detailed information about the content of the selected spectrum component. Flags of different colours are used to identify the desired signal, intermodulation products, and distortion products.

Figure 7 shows RF Inspector applied to a converter circuit that has test points (TPs), combiners (COMBs), and Butterworth bandpass filters (BPFBs). In this circuit, the signal is downconverted at 2.14 GHz and has undesired signals at 1.4 and 2.4 GHz.


To join the conversation, and become an exclusive member of Electronic Design, create an account today!