Frequency planning is a critical first step in the design of a radio communications system. Several frequency schemes are usually considered to ensure system reliability as well as compatibility with existing systems and compliance with applicable regulations. External performance is typically controlled by a regulatory agency, such as the Federal Communications Commission (FCC) or European Telecommunications Standards Institute (ETSI), and the designer must comply with requirements before products are shipped and revenue collected.

Spurious signal generation is a major obstacle in achieving performance and regulatory compliance in a system design, and the search for spurs can be tedious and time consuming. Fortunately, a new technique can reduce spurious searches from weeks and days to hours and minutes. The technique is embodied in WhatIF (pronounced What-I-F), a new frequency-planning tool incorporated in GENESYS™ 2005 from Eagleware-Elanix (Norcross, GA).

The new method shows the spurious performance of all intermediate frequencies (IFs) on a single graph. Spurious-free regions are identified along with multiple frequency-band conversions to a common IF. Operators can see performance trade-offs between IFs and can identify all spurious offenders for complete control over mixer requirements and specifications. The spurious performance is predicted and analyzed by a new, extremely fast synthesis technique which determines the responses and their amplitudes based on the characteristics of selected mixers, system frequencies and bandwidths, and desired IF bandwidths.

For example, Fig. 1 identifies all spurious-free regions above −100 dBc for a dual-band mobile telephone operating in the cdmaOne bands. The performance of each RF band is shown in a different color. Each horizontal Gantt bar shows the IFs, amplitude, and identification information for each spurious combination. In the graph, IF is shown on the horizontal axis; amplitude in dBc relative to the IF output is shown on the vertical axis. The shaded regions highlight the range of IFs that are spurious free above −100 dBc for a given spurious order. On an average personal computer (PC), this entire simulation requires less than a fraction of second to run, saving an enormous amount of time during the system-development process.

How are spurious products predicted so quickly? Spurious products can be determined from the frequency-conversion equation:

FIF = | m × FRF ± n × FLO | or FRF = | m × FIF ± n × FLO |

where:

m = 0, 1, 2, ...,

n = 0, 1, 2, ...

FIF = the intermediate frequency,

FRF = any frequency in the RF band, and

FLO = any frequency in the local-oscillator (LO) band.

From this equation all spurious products can be determined. It's easy to analyze frequencies that fall in a given IF band. Determining the amplitude of these spurious products is more challenging. But, selecting the best IF can be difficult at best because of the time needed to create and analyze the data sets representing all of the possible spurious combinations for a single IF. The process is further complicated for multiple-band operation where it is desirable to use a common IF for all bands. Also, locating an input IF while trying to maintain spectral purity across a wide output band is troublesome.

An Internet search of current analysis tools for performing spurious searches does not yield a unified process to address frequency-planning issues. In fact, commercial frequency-planning software is virtually nonexistent. Traditional tools include spreadsheets, spur charts, and many other types of custom tools. Such tools have their applications, but they also have many weaknesses:

  1. Interpretation of the results can be complicated and confusing.
  2. Most of these tools assume that the LO band is independent of the RF band.
  3. These tools don't account for IF bandwidth.
  4. The tools don't account for spurious amplitude.

Spurious searches typically involve interpretation of large amounts of data. Charting techniques typically require normalized data and become very complicated when bandwidths are taken into account. Custom tools are difficult to use for engineering groups since many RF engineers have their own sets of custom tools they like to use. This makes it very difficult to pass the design from one engineer to another.

The dependence of the LO and RF bandwidths represents one of the most glaring problems with spurious search tools. To truly determine the performance of a given IF, the LO and RF bands are always dependent and cannot be separated. The relationship between the RF, LO, and IF bands is:

FRFBW = FLOBW + FIFBW

where:

FRFBW = the RF bandwidth,

FLOBW = the LO bandwidth, and

FIFBW = the IF bandwidth.

From this simple equation, it is apparent that the LO bandwidth must always be smaller than the RF bandwidth by the IF bandwidth. Violation of this rule falsely predicts spurious products appearing across wider frequency ranges. True characterization of every IF can only be obtained by preserving this bandwidth relationship for every case.

As modern systems push for increased bandwidths and IFs become wider, chances increase that a spurious product will fall in-band. As the IF bandwidth increases, the required LO bandwidth decreases yielding spurious combinations that cover smaller frequency ranges. A designer must account for these differences to minimize design time and components over or under specification.

Most spurious searches are tedious and time consuming. Accounting for amplitude is yet another layer of complexity. Depending on the spurious combination and mixer input drive level, legitimate IFs can be selected even though spurs may appear in-band if their amplitude is low enough. Knowing spurious amplitudes is very helpful when making trade-offs during the frequency-planning process.

A new technique was developed that significantly reduces the time required to understand the performance of all IFs. An implementation of this technique is found in the WhatIF software. User input is simple. Several parallel conversion schemes can be used to determine a single IF. The conversion scheme can be selected such that the IF is an input, as in the case of a transmitter, or output, as in a receiver. Two modes of operation are available: the performance synthesis for all IFs or the traditional analysis technique for a selected IF. Furthermore, mixer performance is specified as a double-balanced mixer or an intermodulation table. After user entry, the performance of every valid IF for the given configuration is displayed on an easy-to-read graph (Fig. 1). This technique is unique and avoids the traditional iterative techniques that are time consuming and difficult to use, hard to understand, with results that are complicated to interpret.

For example, assume a dual-band receiver operating in the 869-to-894-MHz and 1930-to-1990-MHz bands. To minimize the cost of downstream components, it is desirable to find a single IF of this dual-band receiver. Is there an IF to minimize the possibility of self-generated interference?

Figure 2 shows the settings of WhatIF. The number of parallel mixers is 2, the IF should appear at the output, and the maximum spurious order is 10 with an amplitude range of −100 dBc. Since the IF is unknown, the worst-case behavior of all IFs is examined.

Figure 3 shows configuration information for the 1930-to-1990-MHz band only. Each parallel conversion is configured independently. Sum and difference schemes are selected including high- and low-side LO injection. Known design parameters, such as RF center frequency and bandwidth, IF bandwidth, and drive levels, are specified. On the 'Type' page of the "Frequency Planner Properties" dialog box (not shown), a generic double-balanced mixer or intermodulation table mixer can be specified. These mixer models coupled with the drive levels determine the spurious performance used during the analysis.

After clicking the "Apply" button, the results in Fig. 1 appear in a fraction of a second, in an easy-to-interpret format. The performance of every valid IF that meets the requirements for both RF bands appears on a single graph. Mouse fly-over text is used to identify each spurious and spurious-free region. In Fig. 1, it is an easy matter to identify the half-IF spurious product (the 2 × 2 spur from 0 to 119.375 MHz) on the left-hand side of the graph as well as all spurious-free regions shaded in green.

To demonstrate the analysis capability of the WhatIF software, a spurious-free IF of 772.85 MHz (right side of Fig. 1) was selected and analyzed. On the "Settings" tab of the "Frequency Planner Properties" dialog box, when this frequency is specified as the IF center frequency, an appropriate LO is created internally for both RF bands. This LO, along with the specified RF bands, will be analyzed using the appropriate equation (i.e., FIF = | mFRF ± nFLO |). The mixer output is then displayed as shown in Fig. 4. This display is equivalent to looking at the single IF spectrum output of the dual-band conversion scheme on a spectrum analyzer. If the RF and LO frequencies were varied across their ranges, the 0 × 8 and 2 × 8 spurious ranges could also be tracked. These ranges have also been plotted in Fig. 4. As predicted by the performance shown in Fig. 1, the 772.85-MHz IF is indeed spurious free to −100 dBc. The valid IF region is also shown.

This new spurious prediction and analysis technique can slice weeks and days on spurious searches to hours and minutes. It applies to all upconversion and downconversion schemes, including conducted emissions from spurious mixer products, allowing engineers to finally discard their spreadsheets. Eagleware-Elanix Corp., 3585 Engineering Drive, Norcross, GA 30092; (678) 291-0995, FAX: (678) 291-0971, e-mail: sales@eagleware.com, Internet: www.eagleware.com.