Benefits of an All-Digital IF in Spectrum/Signal Analyzers
by Dipti Chheda, Agilent Technologies, Inc.

Spectrum and signal analyzers (the latter term increasingly used for those who do vector signal analysis along with spectrum analysis) are a vital tool for the RF and microwave engineer in high technology design and manufacturing. The ADC (analog to digital converter) and DSP (digital signal processing) technologies that enable functionality and performance improvements in communications gear are doing the same for signal analyzers. The result is a dramatic improvement in both the performance and functionality, for example measurement accuracy, frequency resolution and speed, of many different types of measurements ranging from spectrum measurements to modulation analysis.

So, if you are a design engineer whose goal is the accuracy of your design; or a manufacturing engineer whose goal is the integrity of the tests run; or a manager whose goal is the quality of your products, you will find in this article how the all-digital final IF* technology in today’s spectrum/signal analyzers help you achieve your goals.

The Role of the Intermediate Frequency (IF) in a Spectrum/Signal Analyzer
In an RF or microwave spectrum analyzer, the input signal is first downconverted to a fixed IF by multiple stages of frequency conversion and filtering. [1]

Because of the frequencies and bandwidths involved, the earlier downconversion and filtering is generally performed using analog techniques. The essential signal analysis is then all performed at this fixed IF, and swept analysis is performed by sweeping the input signal through the IF filter rather than sweeping the filter itself. The fixed frequency approach simplifies the filter design and enables stable, high performance filtering.

In a traditional swept spectrum analyzer, the signal at this IF then passes through an analog logarithmic amplifier (so that the analyzer’s results are presented on a log/dB scale), an envelope detector, and then a video filter. It is after this stage that the signal is digitized and displayed, as shown in Figure 1. Thus, the ADC is in the last stages of the signal processing path, a common architecture since the 1970s.

Taking Advantage of Improving Digital Technology
To gain the advantages of digital technology, today’s state-of-the-art spectrum/signal analyzers digitize incoming signals immediately after the signal is downconverted to the fixed IF, much earlier in the signal path than before. This is the concept of the all-digital final IF, shown in the lower portion of Figure 1. Thus, the log amplifier, resolution bandwidth filters, video filtering (display averaging) etc., are all implemented using digital signal processing. This technology was first used in lower frequency spectrum analyzers in the late 1980s (HP 3588A) and RF analyzers in the early 1990s (HP 89440A). Taking advantage of the improvements in digital technologies, today Agilent’s RF/microwave PSA, MXA and EXA analyzers utilize this technology to deliver significant improvements in performance.

Measurement Benefits of All-Digital IF Technology
With sufficient ADC and DSP performance, the all-digital IF technology can yield significant benefits in all of the most important performance categories for the RF engineer: frequency and amplitude accuracy, frequency resolution and measurement speed.

Improved measurement accuracy: Predictable time and frequency response of the digital filters and circuits enable highly accurate corrections, improving the amplitude and frequency accuracy of the analyzer. When all the resolution bandwidth or IF filters are implemented digitally, this accuracy will be same irrespective of resolution bandwidth setting. When a combination of analog and digital filters is used, as in the older spectrum analyzers, there will be higher uncertainty when using analog filters or switching from digital to analog filters and vice versa. [2]

(1.) Log scaling for the display (previously known as log amplification) is implemented by digital computation, so there are no reference level dependent gain stages and there is no IF gain error/reference level switching uncertainty. Digital implementation also improves log linearity error (same as scale fidelity) to an almost negligible level.

(2.) The uncertainty due to switching between resolution bandwidths with digital filters improves to about one tenth of that with analog filters.

(3.) The actual power bandwidth (independent of signal type) can now be known and specified much more accurately than traditional RBW figures—accuracy of ±2% or better, vs. the typical ±10-20% specification for analog IF filters. This power bandwidth accuracy improves many measurements, including the noise marker, band power markers, channel power, and all adjacent channel power measurements.

(4.) Switching between log and linear display scale is implemented in the processor, so there is no scale switching error when using both scale types.

For Agilent spectrum and signal analyzers, the accuracy improvements have been dramatic; approximately an order of magnitude better. The Agilent PSA Series, for example, offers amplitude accuracy as good as ±0.19 dB over a usefully broad amplitude and frequency range. [4] Spectrum analyzer amplitude accuracy can now approach that of a power meter.

Increased usable dynamic range: In an analog spectrum/signal analyzer architecture, setting the reference level automatically sets the saturation level for the analog log amplifier. Hence, any signal that goes above the reference level drives the log amplifier into saturation, causing overload of the IF and significant measurement error.

With the all-digital final IF architecture, the logarithmic display scaling is implemented entirely in digital circuits. This means that the reference level does not set saturation level of any analog log amplifier, because there is no analog component to the log amplifier. In other words, the display range does not set the measurement range. This increases the usable dynamic range of the analyzer, makes the display more flexible for the user, and removes a potential source of error (an incorrectly set reference level).

This is a key advantage in an automated test environment, where the operator is remote from the analyzer and cannot see the display screen. This ensures that the measurements are reliable and accurate, and improves confidence in the measurement. It also lets the front panel user examine a low level signal with high amplitude resolution, such as might be needed in manual circuit adjustment.

Figure 2 shows an example of zooming in on a low level signal while an off-scale signal is still measured accurately. The measurement on the left uses a scale of 10 dB/division, while the measurement on the right is zoomed in to 2 dB/division to get more resolution on the smaller signal. In this example, the Agilent MXA makes accurate measurements even when the signal is outside the display range of the analyzer.

Improved frequency resolution and sweep speed:
Digital filters and other DSP technologies improve frequency resolution and sweep speed in a number of ways. Digital technology provides better filter shape, more filter choices, and more predictable filter response which can be used to further increase sweep speed.

(1.) Selectivity or shape factor is defined as the ratio of the 60-dB bandwidth to the 3-dB bandwidth of the analyzer’s IF (resolution bandwidth) filters. Smaller numbers indicate a filter with sharper transition bands and a more desirable shape. The digital resolution bandwidth filters have more than three times better selectivity (4.1:1) than analog filters, enabling the use of wider resolution bandwidth (faster sweep speed) to resolve closely spaced signals which are significantly different in amplitude.

(2.) The digital architecture further improves sweep speed by providing the flexibility to better optimize resolution bandwidth setting for resolving the desired signals. For example, Agilent’s PSA Series and X Series analyzers provide a range of 160 RBW bandwidth settings, compared to typically a dozen RBW settings for a traditional analog spectrum analyzer.

(3.) The predictable time and frequency response of digital filters enables highly accurate frequency, amplitude and time corrections, allowing for a more than a 2x increase in sweep rates, without compromising accuracy.

(4.) Because the IF signal is precisely digitized, other types of DSP can be used in addition to RBW filtering. This includes FFT analysis, which can be a better technique (in terms of speed and resolution) for narrow resolution bandwidths. Based on the measurement settings, analyzers can automatically select the best approach between swept and FFT operation or the user can set the mode manually. The combination of FFT analysis and swept analysis allows for the fastest possible measurements in widely varying measurement scenarios. [3]

(5.) The log/linear display selection and the dB/division scaling are digital post-measurement functions. This will enable the data to be viewed on any flexible scale without the need to re-measure.

All of these enable today’s spectrum/signal analyzer to make measurements faster without compromising accuracy.

In Figure 3, you can see that 2 tones, 200 kHz and 57 dB apart, can be resolved completely with 120 kHz digital resolution bandwidth compared to an analog resolution bandwidth of 30 kHz (available in only 1-3-10 sequence).

Summary
Rapid improvements in ADC and DSP technology have eliminated the bandwidth and cost drawbacks of digital filters in spectrum and signal analyzers. The result is a dramatic improvement in both performance and functionality for today’s most demanding signal measurements.

References:
1. Spectrum Analyzer Basics, Agilent application note 150, literature number 5952-0292, http://cp.literature.agilent.com/litweb/pdf/5952-0292.pdf
2. PSA High-Performance Spectrum Analyzer Series: Amplitude Accuracy, Agilent product note, literature number 5980-3080EN, http://cp.literature.agilent.com/litweb/pdf/5980-3080EN.pdf
3. PSA Performance Spectrum Analyzer Series: Swept and FFT Analysis, Agilent application note, literature number 5980-3081EN, http://cp.literature.agilent.com/litweb/pdf/5980-3081EN.pdf
4. Gorin, Joe. “Achieving Amplitude Accuracy in Modern Spectrum Analyzers.” Microwaves & RF, September 2008.

Acknowledgments
The author wishes to thank Ben Zarlingo and Joe Gorin of Agilent Technologies for their time and contributions to this article.

Author Bio
Dipti Chheda is a sales and technical support engineer at Agilent’s Signal Analysis Division in Santa Rosa, California, focusing primarily on communications and general signal analysis topics.

Agilent Technologies Inc.
www.home.agilent.com
TXTLINX.COM85
Email this article to a friend!