Timing Considerations Using FFT-based Measuring Receivers for EMI Compliance Measurements

Transcription

1 Timing Considerations Using FFT-based Measuring Receivers for EMI Compliance Measurements Jens Medler Rohde & Schwarz GmbH & Co. KG Abstract The use of FFT-based measuring receivers for EMI compliance measurements is motivated by the desire to reduce the scan time by several orders of magnitude and to gain additional insights by applying longer measurement times.usage of an appropriate measurement time is the key to comprehensively record the disturbance characteristic of the equipment under test (EUT). Keywords EMI receiver; FFT-based measuring receiver; spectrum analyzer; EMI compliance measurement; CISPR ; CISPR Introduction With rd the publication of Amendment 1 to the 3 Edition of CISPR [1] in June 2010, FFTbased measuring receivers were introduced for EMI compliance measurements. Usage of such receivers is motivated by the desire to reduce the scan time by several orders of magnitude without entailing any degradation of the accuracy. To achieve this significant improvement, FFT based receiversmeasure spectral segments that are much wider than the resolution bandwidth during the measurement time by performing parallel calculations at multiple frequencies (Figure 1). In contrast, classic EMI receivers measure the signal within the resolution bandwidth in a given measurement time, thereby requiring a longer scan time for the entire frequency range. Figure 1 FFT-based measurement versus classic stepped scan Related specifications for the measurement methods to be used with FFT-based measuring receivers were published in the relevant parts of CISPR 16-2 (see section II) [2]. This article focuses on the selection of an appropriate measurement time using an FFT-based receiver for measuring broadband disturbance and intermittent signals. This paper is the author's continuation of the topic "Use of FFTbased EMI test receivers for EMI compliance measurements" which was published in SAFETY & EMC in 2013 [3]. Concept of CISPR 16-2 The CISPR 16-2 basic standard series describes methods for measuring disturbance and immunity. For disturbance measurements, the following parts are relevant: CISPR for conducted disturbance measurements CISPR for measurements of disturbance power CISPR for radiated disturbance measurements In all three parts, definition of the minimum measurement and scan times for continuous disturbances represents a major requirement. In general, it is important to choose a setting that allows measurement of the maximum emission while ensuring that disturbance signals are not overlooked. To achieve this objective, CISPR 16-2 requires usage of minimum measurement times which are equivalent for stepping and FFT-based EMI receivers (see Table 1). In addition, the frequency step size must be equal to half of the resolution bandwidth (band A: 200 Hz, band B: 9 khz, band C/D: 120 khz) or less. These requirements are also applicable to spectrum analyzers, resulting in the minimum scan times shown in Table2. Note: The terms sweep and scan time are used interchangeably within CISPR The sweep/scan time T s is defined as the time between the start and stop frequency of a sweep or scan. Tables1 and Tables 2 apply to CW signals. 22 SAFETY & EMC 2015

2 Table 1 Minimum measurement times for CW signals acc. to CISPR [2] Frequency band Minimum measurement time T m A 9~150 khz 1 0 ms B 0.15~30 MHz 0.5 ms C and D 30~1 000 MHz 0.06 ms E 1~18 GHz 0.01 ms approach for this purpose. It is essential to use a sufficient number of sweep points and a long sweep time (> 15 s). For CISPR band B (150 khz to 30 MHz), around sweep points are sufficient, which corresponds to a step size of about one half of the resolution bandwidth. The thick curve in Figure 3 is a good indication that intermittent signals exist. As long as the spectrum display continues changing, there may be more intermittent signals to discover. Table 2 Minimum scan times for CW signals in the three CISPR bands with peak and quasi-peak detectors acc. to CISPR [2] Frequency band Scan time T s for peak detection A 9~150 khz 10 ms s=47 min Scan time T s for quasi-peak detection B 0.15~30 MHz 0.5 ms s=99.5 min=1 h 39 min C and D 30~1 000 MHz 0.06 ms s=323.3 min=5 h 23 min Depending on the type of disturbance, the measurement time T m or the scan time T s may have to be increased - even for quasi-peak measurements. In extreme cases, the measurement time T m at a certain frequency may have to be increased to 15 s if the level of the observed emission is not steady. The measurement time used in stepping and FFT-based measuring receivers and the sweep time used in spectrum analyzers are the key to ensure that disturbance signals are not missed during automatic scans or sweeps over frequency spans. Therefore, a timing analysis of the EUT disturbance characteristic has to be performed before executing automated measurements (Figuer 2). Figure 3 Frequency sweep in spectrum analyzer mode; sweep time > 15 around sweep points; EUT: touch dimmer of halogen lamp Next, select a frequency of interest and switch to Zero Span mode to analyze the signal in the time domain. The example in Figure 4 shows an interval between pulses of 20 ms. Figure 2 Analysis steps for automated or semi-automated measurements To demonstrate such timing analysis, a dimmer of a halogen lamp was measured using the spectrum analyzer's Zero Span mode. As an alternative, an oscilloscope may be connected to the receiver's IF output. First, an overview measurement was performed with the peak detector to find critical frequencies for the timing analysis. Performing multiple sweeps with maximum hold is a good Figure 4 Zero span measurement in spectrum analyzer mode shows a pulse interval of 20 ms; EUT: touch dimmer of halogen lamp. For automated measurements, two cases must be considered: single and multiple scans or sweeps. Single scans or sweeps For a single scan or sweep, the measurement time at each frequency must be larger than the intervals between pulses for intermittent signals. If the measurement or sweep time is too short, erroneous measurement results will be obtained. SAFETY & EMC

3 Multiple scans or sweeps For performing multiple sweeps with maximum hold, the observation time at each frequency needs to be sufficient to intercept intermittent signals, i.e. the observation time has to be selected according to the pulse repetition interval of the disturbance signals (see timing analysis above). If the sweep timeper sweep is too fast, the probabilityof intercept per sweep is very low and erroneous measurement results will be obtained (see Figure 5). Increasing the sweep time per sweep gives a higher probability of intercept (see Figure 6). Figure 5 Visualization of intercepts low probability of intercept per sweep using very high sweep rate steady. The following equation can be used to calculate the minimum sweep time for multiple sweeps: T s min=2 ( f/b res 2 ) (1) where T s min is the min. sweep time for multiple sweeps f is the frequency span B res is the resolution bandwidth Timing Considerations using FFTbased Measuring Instruments Today's FFT-based measuring receivers are limited by the currently available analog-to-digital converters(adc), e.g. for a 1 GHz measuring receiver an ADC is necessary with 2 GS/s sampling rate to meet the Nyquist criterion. Furthermore, preselection filters and high resolution ADCs (e.g. 16 bits for 30 MHz FFT bandwidth) are necessary to meet the CISPR overload requirements for quasi-peak detection. Therefore, today's FFT-based receivers cannot sample and evaluate the complete CISPR bands C/D (30 MHz to MHz) and E (1 GHz to 18 GHz) in one shot. Instead, they combine parallel calculations at N frequencies with a stepped scan. For this purpose, the frequency range of interest is subdivided into several segments that are measured sequentially (see Figure 7). Figure 6 Visualization of intercepts high probability of intercept per sweep using lower sweep rate. In all cases, the fastest possible sweep time per sweep is the multiplicative inverse of the resolution bandwidth based on a step size of 0.5 RBW, e.g. 194 ms for RBW=100 khz over the frequency range 30 MHz to MHz.This fast sweep must be applied continuously with maximum hold for a period of time equal to or greater than the time that would have been spent for sweeping in line with Table 2, e.g s for peak detection of CW signals corresponding to 5 sweeps of 194 ms each for the frequency range 30 MHz to MHz. However, depending on the type of disturbance, the sweep time T s may have to be increased, e.g. to 15 scorresponding to 77 sweeps of 194 ms each for the frequency range 30 MHz to MHz if the level of the observed emission is not Figure 7 FFT scan in sequence;source: CISPR [2] The scan time T scan is calculated as: T scan = T m N seg (2) where T scan Tm is the measurement time for each segment, and N seg is the number of segments. 24 SAFETY & EMC 2015

4 FFT-based measuring receivers have to meet the general requirements for the minimum measurement time and step size as defined in CISPR 16-2 (see section II). Therefore, the measurement time T m must be selected longer than the pulse repetition interval for correct measurement of a broadband spectrum. If the measurement time is too short, enormous measurement errors can result. In a worst case scenario, the FFT-based measuring receiver may not capture the disturbance signal at all. This is problematic if the segment size has a large width, e.g. 30 MHz or more. Fast Sweep versus Fast Time Domain Scan For demonstration purposes, the same EUT was used: a touch dimmer of a halogen lamp. The frequency span was also reduced to 5 MHz for better visualization. First, a single fast sweep without sweep time adjustment(sweep time SWT = 20 ms) was performed. It is clear that the result does not match the spectrum envelope at all (Figure 8). Figure 9 Spectrum analyzer A single sweep with sweep time adjustment based on timing analysis of the EUT (a touch dimmer of a halogen lamp). 1 s 20 ms pulse interval sweep points). However, the result in Figure 10 shows that critical frequencies are still missed! For this EUT, a SWT=2 s for 20 sweeps gives a good match with the envelope. Figure 8 Spectrum analyzer A single sweep without sweep time adjustment does not matchthe spectrum envelope at all Next, a single sweep with sweep time adjustment based on the above timing analysis was performed; this results in a sweep time of 20 s (20 ms pulse interval sweep points). The result in Figure 9 exhibits a good match with the envelope. If the pulse repetition interval is unknown, many users perform multiple sweeps with fast sweep times using the maximum hold function to determine the spectrum envelope. For low repetition impulsive signals, many sweeps are necessary to fill up the spectrum envelope of the broadband component. The correct sweep time can also be determined by increasing it until the difference between maximum hold and clear/write displays is below 2 db, for example. For demonstration purposes, a multiple sweep with sweep time adjustment based on the above timing analysis was performed; this results in a total observation time of 20 s for 20 sweeps (20 SWT = Figure 10 Spectrum analyzer multiple sweep (20x) with sweep time adjustment based on timing analysis of the EUT. Finally, a timedomain scan with measurement time adjustment based on the above timing analysis was performed. An FFT based measuring receiver with 30 MHz segment size can measure CISPR band B (150 khz to 30 MHz) in one shot. According to Equation (1), this results in a pure scan time of 20 ms; in reality, however, a highperformance receiver can generate the result in about 100 ms. The achieved curve is a full match with the envelope (see Figure 11). An overall comparison of the different approaches is shown in Table 3. More Speed with the Time Domain Scan FFT-based measuring receivers can attain measurement speeds a few thousand times faster than can be achieved in conventional, stepped frequency scan mode. As a consequence, frequency scans SAFETY & EMC

5 Figure 11 FFT-based receiver time domain measurement with measurement time adjustment based on timing analysis of the EUT Pulse repetion frequency Measurement time to get interception Table 3 Match with envelope versus scan time Single sweep SWT=20 ms Single sweep SWT=20 s in the CISPR bands using the peak detector can be performed in only a few milliseconds, and even with the quasi-peak and average detector it only takes seconds, thereby rendering preview measurements with the peak detector obsolete (see Figure 12). 20x Multiple single sweep SWT=1 s Figure 13 Measurement speed of timedomain scan versus stepped frequency scan with the R&S ESR EMI test receiver 20x Multiple single sweep SWT=2 s Conclusions FFT-based measuring receivers can be used for EMI compliance measurements in accordance with Amendment 1 to the 3 rd Edition of CISPR The use of FFT-based measuring receiversis motivated by the desire to reduce the scan time by several orders of magnitude without loss of accuracy while gaining deeper insight by applying longer measurement times.another benefit is that the disturbance spectrum can be directly measured with the final detector, making measurement of fluctuating disturbances more reliable.for precise and reproducible measurements, the use of an appropriate measurement or sweep time based on timing analysis of the EUT's disturbance characteristic is essential. Time domain scan, 0.15~5 MHz Step=2.25 khz MT=20 ms 50 Hz 20 ms s 20 s 20 s 40 s 0.1 s Match with envelope no good not all good good Figure 12 Analysis steps for automated or semi-automated measurements with an FFT-based measuring receiver Rohde & Schwarz has introduced a new generation of FFTbased EMI test receivers for disturbance measurements compliant with CISPR 16: the R&S ESR [4]. Its FFT-based time domain scan can deliver measurement speeds up to times faster than can be achieved with the traditional single-channel filtering approach (see Figure 13). The ultra-fast measurement speed is particularly useful if the equipment under test can be operated only during a short period of time, e.g. a starter motor in cars. Part of the time savings can also be exploited to apply longer measurement times in order to reliably detect narrowband intermittent signals or isolated pulses. References [1] Amendment 1: to CISPR : (Edition 3) Specification for radio disturbance andimmunity measuring apparatus and methods - Part 1-1: Radio disturbance and immunity measuring apparatus - Measuring apparatus. [2] Amendment 1: to CISPR : (Edition 3) Specification for radio disturbance andimmunity measuring apparatus and methods - Part 2-3: Methods of measurement of disturbances and immunity - Radiated disturbance measurements. [3] MEDLER, J., More speed, more insight - Use of FFT-based EMI test receivers for EMI compliance measurements, SAFETY & EMC 2013, [4] R&S ESR EMI Test Receiver, Product Brochure Version 02.00, www. rohde-schwarz. com 26 SAFETY & EMC 2015