By Mark Elo

Choosing the Correct Analyzer for Your Transmitter or Component Test Needs

There are many analyzers available in the market today. Some are very specialized, while others offer more generic radio measurement capability; some are called

click to enlarge

Top, Figure 1. Third-generation CDMA base stations transmit a composite downlink signal formed by summing up different physical channels.
spectrum analyzers, while others are called signal analyzers. All measure and display frequency vs. amplitude. In our cost-conscious industry, you need to get the best value for the company’s investment. This paper will help you to ask the relevant questions when looking at the broad selection of tools available in the market and help you make a well-informed decision when purchasing an instrument.

Understanding Price vs. Performance

The price of an analyzer is in part derived from its cost. If the analyzer is designed using low cost components, its price will look very attractive; however, usually its performance will be limited. What is meant by performance? The choice of mixers, LOs amplifiers, A/D converters, IF FPGA/ASIC, and microprocessor all contribute to cost and performance. For example, a single loop local oscillator design may be very cost-effective; however, it may introduce enough phase noise distortion to render the measurements useless. Or a low cost microprocessor may seem attractive from a cost perspective; however, if the analyzer uses this microprocessor for any type of DSP to perform demodulation, then the spectrum analyzer will execute at a very slow pace.

Here are innovative ways to provide performance while keeping cost and, ultimately, price under control:

•To sweep or not to sweep: Many traditional analyzer manufactures still use a sweeping architecture. While this is good for microwave and millimeter-wave spectrum analysis, many innovative RF analyzer suppliers forego this traditional swept system and create a similar — in most cases, better — measurement using signal processing techniques. The Keithley Model 2810 Vector Signal Analyzer with spectrum analyzer capabilities is a great example of this.

•Measurement speed. When thinking about buying an analyzer, a good question to ask is "How is the measurement data processed?" Some instruments use multiple processors to get a fast result, while others leave the main processor for general instrument housekeeping and use an FPGA or ASIC to execute the measurement. Some just have a single microprocessor to do all the work. Obviously, while the latter is the most cost-effective for the supplier, it could be unacceptably slow with more complex modulation schemes. The Keithley Model 2810 provides the industry’s highest performance measurement infrastructure, utilizing a unique high speed architecture based on a DSP-based IQ measurement engine.

Frequency Range

Don’t buy more frequency range than you need: one of the largest contributors to instrument price is the frequency range. Many analyzers have frequency breaks around 2.5 GHz, 6 GHz, 13 GHz, and 26 GHz. High performance instrumentation can often go up to 50GHz. If you are working on cellular products or products that operate in the ISM band, such as 802.11b/g wireless LAN

click to enlarge

Figure 2. Two tones (non-linear amplifier distortion)
devices, often the most cost-effective analyzer would be one with an upper frequency range of less than 3 GHz. CW Measurements Let’s take a look at the simple spectral situation illustrated in Figure 1. Here, we have two signals: a CW, or carrier wave, and a smaller interfering signal. The carrier wave has a number of characteristics. It has amplitude, frequency, phase noise, and broadband noise. The amplitude is the spectral energy emitted by your device at a specific frequency. The phase noise, represented by the skirt of the signal, tells you how stable or spectrally pure your signal is. Usually the local oscillator in your product contributes to the phase noise of the signal. To the left, we see an unwanted signal, or a spurious signal. This signal may be due to a large transmitter close by or be generated by some other part of your system; for example, it may be derived from the microprocessor clock.

Amplitude Measurements

The better the amplitude measurement, the more reliable or certain your result will be. When looking for analyzers, don’t settle for worse than 0.6dB or less than 3GHz measurements.

Noise Measurements and Low Level Signals

When measuring noise and/or low level signals, make sure that your analyzer has a pre-amp. Also, again take a look at the analyzer’s measurement architecture. Analyzing low level signals often means you’ll want to set a very narrow span. When you compare speed in narrow spans across a number of analyzers, you’ll notice swept-based analyzers slow down considerably, while analyzers that employ digital signal processing techniques don’t suffer this type of degradation. Finally, you’ll want to express your noise measurement result in terms of noise density within a certain bandwidth. Unlike other analyzers that define the resolution bandwidth as the 3dB point on a Gaussian shaped filter, the Keithley Model 2810 can specify its filters as noise bandwidths, making it very easy to perform this type of measurement.

Intermodulation Measurements

This type of measurement determines the distortion a device or system can create when stressed under specific signal conditions. Looking at Figure 2, we can see we are stimulating our device with two CW carriers or tones. The tones are causing the device to generate distortion that can easily be seen in the frequency domain as the two distortion products, left and right of the input tones. As the analyzer is also a receiver with active components in

click to enlarge

Figure 3. Modulated signal
its signal path, there is a risk that the analyzer can also generate this type of distortion, making the measurement invalid. One easy check to verify the signal’s integrity is to increase the attenuation setting of the analyzer. If the signal reduces in amplitude as you increase the attenuation, then the products are generated by the analyzer. If changing the attenuation has no effect on the products, then the measurement is valid. When you increase the attenuator value, you will notice that the noise floor increases by the same amount of dBs. This is so the amplitude of the carrier remains constant with different levels of attenuation. The increase in noise floor, however, could mean that the noise could mask the intermodulation product.

To get the best measurement performance, fine attenuator steps are important. Coarse attenuation steps can move the noise floor by 10dBs, quickly masking the signals that you want to measure.

The ability to measure small signals in the presence of large signals is a key use of any type of spectrum analysis instrument — this performance attribute is defined as the dynamic range of an instrument. Dynamic range is often expressed as a combination of the analyzer’s third-order intermodulation performance (e.g., the two-tone measurement we discussed previously), the instrument’s noise floor performance, and its phase noise. It is often quite difficult to directly compare the dynamic range of an instrument, as different manufacturers can optimize the instrument for noise floor performance or distortion performance. An easy way to compare the dynamic range of multiple analyzers relatively is to examine the W-CMDA adjacent channel power. This measurement takes into account all of the above parameters.

Modulated Signals

So far, we’ve only looked at carrier wave (CW) signals. When measuring modulated signals, you need to ensure that your spectrum/signal analyzer has the ability not only to measure the spectrum of the signal but also the ability to measure the quality of the modulation.

Figure 3 shows a typical digitally modulated signal in the frequency domain. This signal uses a modulation scheme that does not have a constant power envelope, so its amplitude varies over time. A key measurement an analyzer must be able to perform is the average power of this type of signal. This is usually specified over a defined bandwidth. The intermodulation and phase noise distortion manifests itself in the skirts of the signal. The adjacent channel power feature of the analyzer helps quantify the intermodulation and phase noise performance of the device under test.

The ability to demodulate the signal and express the quality of the signal in terms of a metric such as EVM (error vector magnitude) is a key requirement for modern analyzers. Key analyzer performance characteristics that enable this type of measurement are the instrument’s digitizing bandwidth and its corresponding frequency and phase response. For example, the Keithley Model 2810 has the ability to capture and digitize signals up to a maximum bandwidth of 35MHz. For popular modulations schemes such as GSM or W-CDMA, demodulation and quality metrics are often built into the analyzer. However, your choice of analyzer needs to adapt to evolving communications technologies. Figure 4 shows how the Keithley Model 2810 can be used as a calibrated IQ acquisition engine. In this measurement setup, the Model 2810 is capturing a signal from the device under test and storing it as calibrated IQ pairs in its 50Megasample memory. You can then export this record from the instrument and import it into any commercial analysis package, such as MATLAB®. This will give you the flexibility to continue to make measurements as communication technologies evolve.

Analyzer Connectivity

Most modern analyzers are LXI-C compliant. LXI (LAN eXtension for Instrumentation) is a standard that defines instrument connectivity over LAN. There are three versions of the LXI standard: A, B, and C. C means you can control the instrument over LAN, and it contains a web server for remote operation. For example, if you are sharing measurement information across global teams, you simply type the IP address for the analyzer into your web browser, and the analyzer’s display appears in your web

click to enlarge

Figure 4. Analyzer Connectivity
browser. B and A are still evolving. They are supersets of C that provide more advanced measurement triggering functionality.

Of course, today most of your instruments are still controlled through the GPIB interface. When choosing an analyzer, ensure that it has the connectivity for your legacy test needs using GPIB and is set for the future with at least LXI Class C compliance.

With the advent of LAN-enabled instruments, Internet security and safety are becoming key issues, especially across large enterprise systems. For example, if an instrument is based on Windows® XP, it has all the characteristics of a PC. You’ll need to talk with your IT Department to put it on the network; it is susceptible to viruses and attacks, just like any other PC. Some instrument manufacturers have chosen Linux, although this in turn reduces the instrument’s connectivity to Microsoft-based tools. The Keithley Model 2810 uses Windows CE, which is a good compromise between operating systems, offering both connectivity and safety. Summary A spectrum or signal analyzer can be a large investment, and there are many analyzers in the marketplace from which to choose. To help you compare analyzers, ask yourself the following questions:

•How much money do I have to spend?
•What frequency range do I need to measure?
•What amplitude accuracy do I require?
•What are my dynamic range requirements? (Use ACP as a quick cross-check.)
•What are my phase noise requirements?
•What type of signal do I need to measure: CW or Modulated?
•Do I need demodulation capability? What is my signal’s bandwidth?
•How do I plan to control the instrument, remote or manually?

If you have an unlimited budget, then all the questions are answered easily. If, however, you want to ensure you are getting the best performance for your investment, then work with your instrument sales representative to determine the type of analyzer that’s the best fit for your needs and budget.