Spectral signal analysis limits RF power and parasitic noise emissions

　　Frequency domain measurement of RF power is the most basic measurement performed using spectrum and vector signal analyzers. Such systems must comply with the limits of power transmission and parasitic noise radiation in the relevant standards, and must also be equipped with appropriate measurement techniques to avoid errors.

　　Key frequency-related controls such as frequency range, center frequency, resolution bandwidth (RBW), and measurement time all affect the measurement results.

　　Frequency span refers to the total spectral components that the analyzer can capture, while the center frequency is equivalent to the center of the frequency span. It should be noted that frequency controls such as frequency span determine the frequency span on the front panel of the instrument. On the other hand, FFT signal analyzers have two distinct acquisition modes depending on the size of the frequency span.

　　The frequency range up to the RBW is achieved in the instrument by down-converting a frequency segment and then digitizing the down-converted signal. For the frequency range beyond the RBW, the spectrum segments are converted and digitized in sequence. The RBW controls the frequency resolution on the frequency axis. In traditional analyzers, the spectrum display is achieved by scanning the frequency range with a narrowband filter. The filter bandwidth determines the resolution on the frequency axis and is therefore the control flag.

　　Meanwhile, FFT-based analyzers do not have analog filters, but use FFT and associated windowing parameters to determine frequency resolution or RBW. Unlike traditional spectrum analyzers, the latest FFT-based analyzers can select windows to limit spectral leakage and improve resolution of closely spaced frequency bands in the frequency domain. Those familiar with FFT analyzers and FFTs may ask, what is the relationship between RBW frequency resolution and the width of the FFT tap? Table 1 shows the relationship between the RBW frequency resolution parameter (specified at 3dB and 6dB RBW resolution) and the FFT tap width in new RF signal analyzers.

Table 1: RBW frequency analysis resolution is related to the tap width of the FFT analyzer.

　　Analyzers that use FFTs have window selections to limit spectral leakage and improve resolution of closely spaced spectra in the frequency domain. Traditional spectrum analyzers do not have this capability. The measurement time (or sweep time) of a traditional swept analyzer is inversely proportional to the square of the RBW, which is determined by the settling time of the analog filter. If you want to improve frequency resolution by reducing the RBW, the sweep time increases exponentially.

　　Conversely, as the RBW decreases, the acquisitions made by the FFT signal analyzer are longer and require more computation. As DSP devices become faster, measurements are made faster, allowing for higher resolution or narrower RBW measurements.

Figure 1: Frequency and amplitude relationship of spectrum analyzer measurement results.

　　Amplitude setting

　　Various amplitude controls also affect the measurement results, these include the reference level (ref level), attenuator settings and detection mode. The reference level sets the maximum input range of the spectrum analyzer. It controls the Y axis, similar to "volts/div" on an oscilloscope , and must be set to just above the maximum power measurement expected.

　　The optimum reference level is a balance between minimal instrument distortion (caused by very low reference levels saturating the input signal) and minimal noise floor (caused by too high a reference level, reducing the instrument's sensitivity and dynamic range). Sometimes it is beneficial to set a low reference level for broadband noise measurements, despite introducing some instrument distortion. Doing so will improve the sensitivity of the instrument when distortion can be appreciated, and ensure that it is excluded from the measurement.

　　The attenuator setting control also determines the input range of the instrument. This setting is usually set to automatic mode, where the software adjusts the attenuator value based on the reference level.

　　In firmware, spectrum analyzers couple the display's Y-axis to either the reference level or the attenuator. Virtual instruments have no such restrictions and the display's Y-axis can be decoupled from these controls if desired. This feature allows for visualization of the spectrum without affecting the instrument's amplitude settings. Note that both the reference level and attenuator settings affect the programmable attenuator, so only one needs to be set.

　　Detection mode is another amplitude control method that can be used in traditional swept spectrum analyzers, but not in FFT-based analyzers. It can be divided into normal, peak, sampling or negative peak modes. The specific detection mode determines how the spectrum analyzer reduces the spectrum information, or how to compress the spectrum information.

　　It also affects the overall power measurement. When there are more spectrum data points than the spectrum analyzer can display, the analyzer will benefit from data reduction strategies. This will cause the detection mode to change the power measurement.

Table 2: Spectrum analyzer measurement modes can affect power measurements.

　　Factors affecting accuracy

　　The spectrum analyzer uses a frequency sweep between a start and stop frequency. An analog ramp signal generates the frequency sweep signal, and the start frequency is synthesized from a high-precision time reference signal. Therefore, the measurement accuracy is determined by the center frequency of the analog ramp signal and the IF filter.

　　FFT-based analyzers do not have such analog ramp signals, so they are not limited by these factors, and have consistent accuracy over the entire measurement range. The accuracy within the range depends on the time base and measurement algorithm, so frequency accuracy and repeatability can be obtained more easily.

　　In a traditional swept analyzer, causes of frequency error include reference frequency error, frequency range accuracy (5% of range), and RBW (15% of RBW). Correspondingly, frequency error in an FFT-based analyzer includes reference frequency error and RBW, which can range from > 50% to < 10% of RBW, depending on the measurement algorithm.

　　To compare these errors, the reference frequency error must be ignored because it can be compensated for using a precision frequency source such as a rubidium clock . In a swept spectrum analyzer, measurement performance will be affected when the frequency span is greater than 50kHz and the RBW is set above 1kHz unless optimization techniques are used, such as placing a 100MHz frequency in the center of the frequency span.

　　If a smaller RBW is used, the test time will be longer because of the sweep time, as a typical spectrum analyzer requires a sweep time of 150-200ms. The measurement algorithm limits the measurement accuracy of the FFT-based analyzer. For example, the advanced spectrum measurement analysis toolkit uses interpolation technology to achieve higher resolution than the RBW can achieve. As in the example above, setting the RBW to 2kHz will ensure higher accuracy.

　　FFT-based analyzers use high RBW settings that enable accurate measurements, even without using accuracy-optimized measurement techniques. This means faster and more precise measurements can be made in the same test time. Signal analyzers are able to perform test samples less than 20ms in length, which is six times higher than spectrum analyzers.

　　Unless the proper measurement setup is used, the measurement results can vary widely even for the same test instrument. Therefore, a deep understanding of the operating principle is crucial to correctly set up the measurement instrument.